AI safety is a marathon, not a sprint
The views of the AI safety community are quickly converging on a fast takeoff within the next year. But this relies on a few low-quality data points from AI companies and contested models of the world.
I worry that this convergence will degrade the epistemic environment and make good decisions about AI policy and deployment harder. Forget about predicting the future; it will be (and sometimes already is) hard to adjudicate claims about the present.
It also pushes the community’s energy toward blunt instruments such as generalized bans on AI development, at the cost of decentralized, society-wide approaches that improve resilience. And it normalizes extraordinary government interventions and authoritarian actions.
I hope that even people who are worried about imminent superintelligence and takeover risks can see that the epistemic homogeneity in the AI safety community is unhealthy. The community’s entire focus seems to be on one threat model, backed by contested data points, when it should be prepared for a range of scenarios.
Five ways in which this shows up:
1) Updating heavily on limited evidence.
Prominent commentators have shifted their views based on benchmark accuracy numbers. Of course, benchmarks have some signal. But anything that can be benchmarked is by definition verifiable, so it can be optimized for with reinforcement learning. Still, the community has coalesced around the view that these numbers point to superintelligence emerging in 2027.
2) Focusing analysis and interventions exclusively on AI companies.
Little of the community’s analysis covers how AI is diffusing, or how institutions outside the AI industry are adapting. So almost every incident is seen as a story about model capabilities. The recent outages at Microsoft, United, and the NSA, and the FAA ground stop, all saw vagueposting about the latest cyberattack by an advanced AI model; none of them was the result of an AI cyberattack. The NSA’s outage came from patching vulnerabilities that were found with AI tools, which is a consequence of AI diffusion. But this is not nearly as compelling as the AI cyberattack story, so it didn't get a lot of attention.
3) Ignoring institution- and capacity-building in favor of blunt (and potentially authoritarian) interventions.
In response to concerns around imminent superintelligence, many of the policy recommendations rely on extraordinary government action that veers into authoritarianism. Most are generalized bans on AI training; few build the institutional capacity to absorb AI’s impacts across society.
But the cost of building frontier models is decreasing rapidly, so even if we were somehow able to ban AI development today, these bans wouldn't be effective at preventing societal risks. They will also leave us worse prepared for when a capable model is openly released, or an adversarial nation state develops one, since our institutions will not be ready for the fallout.
4) Adopting a “soldier mindset” instead of forming independent views.
One dynamic I have observed firsthand is that many people in the safety community take the word of prominent commentators as gospel and do not question their claims or methods, which compounds these problems. Even the debates happen within the community’s frame, such as its reliance on subjective probability estimates for AI’s existential risks. So the community is homogeneous in its methods as well as its views.
5) Treating AI safety as a sprint.
To be clear, I think AI capabilities will continue to increase over the next few years, and we need to dramatically improve our responses to AI risks. But that means preparing society to absorb a transformative technology, which goes well beyond heading off the current worry about an intelligence explosion. Remaking our institutions is a decades-long process; we can’t solve it by pushing AI back in the box.
My worry is that today’s views lead us into a perpetual state of unproductive alarm. Even if “superintelligence in 2027” doesn’t materialize, AI progress won’t stop, so I expect the community will keep its narrow focus on AI development. A far more productive outcome would be to update its causal models of the world and treat resilience, institutions, and diffusion more seriously.
Brendan Nyhan
Brendan Nyhan ( ; born 1978) is an American political scientist and professor at Dartmouth College. He is also a liberal to moderate political blogger, author, and political columnist. He was born in Mountain View, California and now lives in Hanover, New Hampshire. More on Wikipedia
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sayash Kapoor
@sayashk
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tyler Kingkade
@tylerkingkade
For this story, I spoke w/ ppl who led edu & AI policy teams at Google
One said of Google's AI products: "We can’t in good conscience hand them off to our kids"
Another said if Google won't let schools get AI features out of Search, then they’re "showing who they are - an organization that’s prioritizing themselves over our children"
NBC News
@NBCNews
Amid growing concern about AI and screen addiction, America’s largest school districts are blocking Google products such as YouTube and Gemini, and are now threatening to abandon Google Search, upending longtime partnerships with the tech giant. www.nbcnews.com/news/education/google-faces-reckoning-ai-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Josh McCrain
@joshmccrain
In Chicago for ASC? Come see me present this paper, with @ian_t_adams and @jnixy
We show hot spots "quasi-experiments" (and similar place-based programs) are biased towards finding a crime-reducing effect. The size of the bias is significant & changes the evidence-base -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
It's not useful... except it is (and again, this is the worst it'll ever be). We need to be thoughtful about how to protect learning in higher ed, but this kind of denial of AI's practical utility is counterproductive
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jonathan Berk
@berkie1
🚨Boston City Council has voted 8-4-1 to eliminate minimum parking requirements for new housing (with some carveouts).
The change still needs approval from The Zoning Commission.
Congratulations to lead sponsor @RonDurk her team, advocates, @AbundantHomesMA and many more. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Mackenzie Arnold
@MackenZ_arnold
RSI has gone form a fringe issue to dead center of the policy window.
Here’s one of the four experts on Governor Newsom’s expert panel saying “If there's one thing the legislature can get done before the end of the year, I hope it is this.” Let’s go.
Gillian Hadfield
@ghadfield
Okay--great honor to have been asked to serve on the expert panel with @alondra, @goldman and @robreich to advise @CAgovernor for a 2 month sprint on what to do about AI regulation. But the most incredible thing is to have been given the opportunity to help produce such a substantive and concrete report, chock-a-block with actionable steps to take NOW. A clear framework, state or federal. Accelerating the IVO ecosystem is central to everything, to rapidly build the capacity needed to MANDATE independent and ongoing risk assessments, reviews of safety frameworks for adequacy, and more. Revising critical incident reporting to ensure Hugging Face incidents must be disclosed. Requiring risk containment protocols (distributed "kill switch"). But the biggest news is around strong recommendations for an immediate priority on reporting on where companies are on the path to RSI (compute reporting, for example, with IVO verification). If there's one thing the legislature can get done before the end of the year, I hope it is this. www.govops.ca.gov/wp-content/uploads/sites/11/2026/11/FIN…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dean W. Ball
@deanwball
There is a common belief in the AI field—that advanced AI means the inevitable forfeiture of human primacy in the world. Even many who disfavor that future believe it is likely. I call this the silent consensus, and it must not be allowed to remain silent any longer.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joshua Basseches (joshuabasseches.bsky.social)
@JoshuaBasseches
This and return of the measles will be two of the hardest things to “course correct” on. So much irreversible damage these last two years.
Jesse D. Jenkins
@JesseJenkins
One of America's greatest strengths —a source of economic dynamism, innovation & discovery—has been the ability to attract the best and brightest around the world to come here and study.
The Trump Administration's aggressive anti-immigration policies threaten to destroy that advantage: new international student enrollment at US colleges and universities plummeted 15% this fall, AP reports.
What an unnecessary self-inflict wound. How many US startups are founded by immigrants who first came here to study? How many new discoveries and breakthroughs come from immigrant researchers? We should be welcoming international students with open arms (and a path to long-term residency). Instead this Administration is telling them to go somewhere else. Sad. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jack
@tracewoodgrains
Basically every one of this guy’s posts is “Every popular free-to-play mobile game is made from demons from hell who lie to you because they want to eat your soul” and it’s hard to watch any of them and not conclude that these companies should be regulated into the ground
Ayo O
@ayoships
I reverse engineered monopoly go
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"According to DukeHub, Ariely is not set to teach any classes in the spring. He has not taught a course since spring 2022."
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sol Messing
@SolomonMg
The evals are not ok
New NeurIPS paper: Hidden Measurement Error in LLM Pipelines Distorts Annotation, Evaluation, and Benchmarking.
Evals drive deployment, safety standards, and research. Yet we ignore noise from prompt, judge, agent harness etc, & so underestimate error in benchmarking & annotation, & it gets worse with more data.
I'll get to what this means for CIs & how to fix, but first let's talk about how this makes benchmark hacking easier 🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Emu Brigadier
@emubrigadier
In case you are interested, these are the genres I tried for the Pangram bounty. Every single pre-2020 sample came out 100% human:
academic books on African politics and security
academic paper acknowledgements with grant lists
academic paper reference lists
advance-fee scam emails
African business news
African general news
African government news-agency items
African opinion essays
African poetry
African sermons and devotionals
African short fiction
African student and education writing
African technology news
amateur rhymed poetry
Arabic how-to articles
Arabic local news and press releases
astronomy telegrams
BBS-era amateur stories
BBS-era bulletin-board lore
BBS-era conspiracy texts
BBS-era humour files
BBS-era occult texts
BBS-era political tracts
BBS-era UFO reports
clinical trial descriptions
company annual reports to shareholders
Congressional Record floor speeches
Congressional Research Service reports
content-farm articles
corporate press releases (SEC 8-K exhibits)
cover-letter and interview-answer samples
data-analysis textbooks
developer mailing-list emails
dream reports
drug labels
drug-label adverse-reaction lists
early modern English printed books
earnings-call transcripts
education conference papers
education evaluation reports
education guides and books
education journal articles
education research reports
education statistics reports
Emu's private drafts *
English-language government news-agency releases
Federal Register notices and rules
Federal Reserve meeting minutes
flash fiction
Hacker News comments
horoscopes
horror fiction and creepypasta
internet standards documents (RFCs)
IRC chat logs
IRS private letter rulings
literary-magazine essays and criticism
LitRPG web serials from Royal Road
mathematics proofs in LaTeX
mathematics textbooks
medical dictation reports
medical-device adverse-event narratives
mid-century literary short story
NASA technical report abstracts
Nigerian forum posts
Nigerian lifestyle and entertainment posts
NIH grant abstracts
nonprofit and agency annual reports
NSF grant abstracts
organisational guidelines
patent descriptions
printed cookbook recipes
product recall notices
programming book indexes
public comments on a Medicaid docket
public comments on US regulations
PubMed abstracts
real-estate listings
recipe blog posts
safe-harbor legal boilerplate
Saudi Press Agency releases
science-fiction and fantasy magazine fiction
self-published free-verse poetry
slang dictionary definitions
small-business press releases
software bug reports
software manual pages
Stack Exchange answers
statistics textbooks
strategic plans
Supreme Court oral-argument transcripts
sustainability reports
synonym dictionaries (scanned books)
UK Acts of Parliament
UK government corporate reports
UK government policy papers
UK government research reports
UK parliamentary petitions
vehicle-owner safety complaints
vision statements
weather forecast discussions
white papers
wiki cookbook recipes
World Bank environmental assessments
World Bank project information documents
World Bank project status and completion reports
World Bank working papers
Yandex search engine poetry -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kareem Rifai 🌐
@KareemRifai
Really, really bizarre stuff:
Trump's appointed chairman for the U.S. Commission of Fine Arts, Rodney Mims Cook, ordered that Atlanta display a giant Russian flag to "honor" Putin's Ambassador to the US who died last week, Alexander Darchiev.
Darchiev was a hardcore defender of Putin's invasion of Ukraine and recently refused to testify before a Senate committee on Russia's mass-kidnapping of children.
Cook's role is expansive: he had a major hand in Trump's ballroom project and the plans for a triumphal arch on the Potomac.
Earlier this year, he traveled to Russia to meet with Putin and pro-Putin actor Steven Seagal, according to a profile in The New Yorker, which described Cook as an “unapologetic Russophile.”
www.the-independent.com/news/world/americas/us-politics/t… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
jenny
@jennygzhang
Why is there more outrage that this guy “fed text without consent into AI” (AI detector ≠ AI, pangram doesn’t use inputs for training data, and you don’t need consent to test a text) than outrage that AI frauds are cheating real human writers—of the same race!—out of opportunity
Tuhin Chakrabarty
@TuhinChakr
It’s unfortunate that mostly black authors have been outed for using AI to write fiction, but does that mean AI detectors are racist ? Sadly, the answer is a resounding NO. With the help of @pangram and @GPTZeroAI I analyzed 187 African books published between 2001 & 2026. 👇👇
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nicole Perlroth
@nicoleperlroth
I haven’t lived here in awhile, but I see no other way to answer the hundreds of texts/DMs asking me if the world is, indeed, ending.
In the span of six days, we’ve experienced;
11/12 — Microsoft outage
11/13 — Nationwide FAA ground stop
11/16 — Major US banking disruptions
11/17 — NSA’s classified internal network goes down
We need far more analysis of why so much is breaking at once. I don’t want to get out over my skis here, and what I am hearing does not contradict the messaging that most of this is the result of screwy updates.
But stepping back, the dots are starting to form a collective picture. We are seeing:
• An explosion in severe and critical vulnerabilities, with AI accelerating discovery to 3X the number of vulnerabilities discovered in the same period last year. This is collectively being referred to as “vulnageddon” or “the vulnpocalypse.”
• Mean time to exploit those vulnerabilities has collapsed to NEGATIVE SEVEN DAYS. AI allows attackers to find/exploit vulns far faster than software makers have time to fix them.
• Powerful AI-enabled offensive tools like ARTEX have been thrown into the hands of every third tier nation state, cybercriminal and digital arsonist, which logically creates havoc.
• This creates enormous pressure to compress patching windows and accelerate software updates, increasing the likelihood these fixes break things in unexpected, cascading ways.
• Pressure to just “AI everything” means we are rolling out automated AI tooling on top of brittle infrastructure that can’t support it.
My generation grew up in an extraordinary period of technological stability. We worked our way up the software stack, abstracting away layer after layer, until we lost a fundamental understanding of how the underlying systems actually work.
We were told to “move fast and break things.” VCS told us to celebrate the idea that “software would eat the world.” Now we’re telling ourselves we can simply “roll out AI” across everything, as though nothing underneath will buckle.
We never stopped to ask: SHOULD WE?
Should we be digitizing every last nook and cranny of our companies, infrastructure, healthcare, homes, cars, bodies, lives? Especially when the complexity, interdependence, and fragility we were introducing overwhelmed our ability to understand/secure/maintain the systems we built?
Nobody every asked: What if, in the name of progress, we were creating the ripest attack surface in human history?
We are now entirely dependent on software systems that almost nobody understands end to end, that increasingly few people can repair, and that attackers can now probe at machine speed. We are hearing that incident response firms, the 911 for breach victims, are backlogged six months. Even those paid on retainer lack the bandwidth to triage every incident. The cavalry is disappearing.
We are entering a period of unprecedented technological instability.
“Software Beirut” is an incredibly apt analogy. We are in for a period of cascading failures, whether triggered by attackers, bad updates, or our own accumulated complexity.
This, I’m sorry to say, is just the opening chapter. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Interesting example of a setting where AI in the workflow could be making everyone worse off.
John Burn-Murdoch
@jburnmurdoch
Problem is, that advice is essentially what everyone now tells ChatGPT (or the off-the-shelf AI job app tools) to do, so every application is now tailored [by AI] in style and substance to the job and company.
Employers hate it because every applicant now looks the same, adding costly layers to the recruitment process, so they use AI to filter applicants, which applicants hate. And you get loads more rejections and waste-of-time interviews which both sides hate.
I really do think the fix is all on the employer’s side:
• Ban AI in applications (use Pangram to enforce) to force a return to the era where a tailored application was a costly signal of strong intent and likely candidate fit
• In return, stop using AI for application screening and AI interviews. Applicants are now giving you their time, so you give them yours. Hiring worked absolutely fine a few years ago, so let’s go back to that. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
It's wild that so many bright people are intentionally designing addiction machines that manipulate their users to make them worse off.
They are literally preying on people. Anyone working on stuff like this should feel deeply ashamed of how they are spending their time.
Ayo O
@ayoships
I reverse engineered monopoly go
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Theo - t3.gg
@theo
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alan Levinovitz
@AlanLevinovitz
It's tremendously frustrating to me that humanists seem to be the ones most opposed to using Pangram as a means of detecting AI writing. Utterly indefensible.
This is OUR CRAFT. Pangram exists to defend OUR SKILL SET. AI devalues writing by removing the barriers to entry for writing, the sacrifice and effort required to produce a text.
People who use AI to flood journals, publications, and the internet with writing they've put no effort into — they devalue the humanities! @pangram and any *effective* AI detection tools are our allies! They make our skill set MORE VALUABLE! They protect the world from an avalanche of effortless slop.
Not only that, they make it so we can teach our students!! By helping to prevent cheating, we give our students the gift of having to think and write themselves. They can't cheat at the intellectual gym, so they get to develop the cognitive muscles that they are paying tuition dollars to train. Failing to use Pangram is betraying them. And the ones who suffer the most are the ones who have least developed these skills.
It would be one thing if Pangram didn't work! But it does! Denying that makes absolutely no sense. It hurts our students, devalues our labor, and encourages the generation of AI slop with no way to stop it. Madness.
I seriously don't get it. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jon Favreau
@jonfavs
I’m sorry - the Secretary of Homeland Security is sharing anonymous guesses that AI may be causing widespread outages??
Is this real life?
Acyn
@Acyn
MULLIN ON OUTAGES: I talked to one guy in the tech industry, he asked not to mention his name, it's not Elon. I asked, what could be causing these outages? He said I believe it's SI, because these programs aren't designed to work as fast as SI is.
So when they put in a program that's able to write codes and it goes in and self-fixes, as it starts to self-fix itself it like a change order on a house.
That kind of makes sense to me.This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
People still don't appreciate what a political killer she is/was (because gender)
Sahil Kapur
@sahilkapur
Ice cold from Nancy Pelosi
(via Biden's new book "Promise Me, America") -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Zero evidence of false positives using Pangram or GPTZero in pre-AI writing from Africa contra various recent excuses
Tuhin Chakrabarty
@TuhinChakr
It’s unfortunate that mostly black authors have been outed for using AI to write fiction, but does that mean AI detectors are racist ? Sadly, the answer is a resounding NO. With the help of @pangram and @GPTZeroAI I analyzed 187 African books published between 2001 & 2026. 👇👇
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Emu Brigadier
@emubrigadier
As someone who submits my own prose to small journals, I am grateful for Pangram! My stuff is human-written, so I feel like the odds of acceptance increase if the editors have tools to weed the slop out.
My submissions are rejected anyway, but it’s a comforting feeling nonetheless.
Nick Burns
@NickBurns
Everyone is fixated on academic honesty or prize contests as test cases for Pangram, and no one seems to consider people in my position, editors at small magazines. We get dozens of slush-pile submissions every week, lots of them are AI-generated, and if we didn't have a tool to easily sort out suspect submissions, we'd have to just stop taking cold submissions altogether, which would be terrible for emerging writers.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Neil Malhotra
@namalhotra
Just published early access: "Political Unfriending on Facebook: Evidence from over 180 Million Users." www.journals.uchicago.edu/doi/abs/10.1... We track daily unfriending patterns on FB during the 2020 US presidential election to understand patterns of political homophily 1/n
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Elai
@elaifresh
This is a prize for AFRICAN LITERATURE
EVERY entrant was African! This can’t be racist or biased!
Legitimate African writers are being cheated by an AI grifter and YOU are defending that! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
SpaceX
@SpaceX
Don't Panic
Riding on Starship 42 during final approach to catch2 media couldn't be displayed. See them on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Yoonseok Yang
@yoonseok_yang
AI can solve a Millennium Prize Problem. It can’t grade a college physics exam.
On real handwritten answers, the best frontier model matches the instructor only 57% of the time.
After grading 6M student answers from 200+ universities, we’re releasing GradeBench (1/8)🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nick Burns
@NickBurns
Everyone is fixated on academic honesty or prize contests as test cases for Pangram, and no one seems to consider people in my position, editors at small magazines. We get dozens of slush-pile submissions every week, lots of them are AI-generated, and if we didn't have a tool to easily sort out suspect submissions, we'd have to just stop taking cold submissions altogether, which would be terrible for emerging writers.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
And in doing so creating a map of where all their air defenses are located
Aric Toler
@AricToler
Fascinating thread here about how Yandex is putting in fake forests and using other measures to hide Russian air defense sites.
This one (55.36303, 37.4445) - apparently an S-400 site - was recently forested.
Compare here:
Yandex, Google (June '23), and Planet (two weeks ago)
jembob 🇺🇦
@jembobineuse
This weekend, they took big measures to cancel a lot of Mirrors and replaced a lot of Street Views by older ones in the whole Moscow Region around air defenses. A lot of SHORAD towers and some S-400 areas went to forests.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Colleagues - you don't have to give up on writing assignments! (link below) Close reading is great, but there is no substitute for grappling with how to make an argument.
Shadi B.
@InuitKodiak
I’m a humanities professor. This is how to continue to educate. I have my students underline passages in the essays I assign so we can analyze them in class. They bring their texts. All of class is given over to text-based interpretation. No at-home writing assigned.
It’s the end of homework. It’s not the end of teaching.
Boze Herrington, Library Owl 🎄🐗📯📯
@SketchesbyBoze
We have an epidemic of students turning in essays they didn’t write about books they haven’t read, and if teachers are now using AI to grade assignments en masse we may as well shut down the schools because no one here is teaching and no one is learning.
CNN
@CNN
The college admissions essay is in a very weird place. It's impossible to say just how many of them are now written by AI, but at this point, experts agree that it's a lot of them. And while colleges generally frown upon that practice, they're using AI to evaluate the essays, too. www.cnn.com/2026/11/15/tech/ai-students-college-admission…
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Liz Landers
@ElizLanders
.@NewsHour has obtained a new memo -dated today- from White House counsel Will Scharf that requires federal agencies to obtain approval from the White House legal and legislative affairs teams before testifying before congressional committees.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jesús Fernández-Villaverde
@JesusFerna7026
The oddest intellectual position I have encountered over the last decade is the refusal of so many to believe that the current version of Pangram is incredibly good at avoiding false positives (i.e., flagging a human-written text of some length as AI-generated). It is not terribly hard to trick Pangram into false negatives (i.e., classifying an AI-generated text of some length as human-written) with “humanizing” tools and some careful editing. But the other way around? Nearly impossible.
For two weeks, I have been running experiments in many different situations for a new paper I will circulate tomorrow (and because, as an editor, I care a lot about the answer). Not in a single case did a human-written text of reasonable length score above 10% for AI generation. And, yes, many the texts I used (including mine!) are from ESL writers.
Frankly, the counterarguments I have read over the last week here on X (and the papers linked to “back them up”) are childish in the extreme.
If you think that false positives at Pangram are, as of November 16, 2026, a concern, you are lying to yourself. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kyle Saunders
@profgoose
Across two years of @OpenAI releases, the share of conspiracy-theory answers that also tell you your concern is reasonable climbed from roughly 10% in GPT-4o to about half in GPT-5.5.
The correction stayed...but the hand-holding grew.
(w/@itserinfitz)
osf.io/preprints/psyarxiv/4yexj_v2 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Not a terrible idea honestly
Eva Vivalt
@evavivalt
Maybe in the future alongside a paper submission I'll include a repo with an AI-written FAQ that points out why some things flagged by LLMs are not actually issues, so the reviewer's LLM has that.
Like an AI online appendix that dispenses with the false positives while keeping the draft clean and human-readable. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The definitive study on this error
Aaron Reichlin-Melnick
@ReichlinMelnick
People are bad at estimating the prevalence of small groups. One of my favorite is that Americans on average will say that they think 30% of the country lives in New York City.
This is also why there are people who make dumb claims like 30% of the U.S. being undocumented.
Ryan Burge 📊
@ryanburge
The average American has zero idea how truly small many religious groups are.
People were asked what share of Americans are Jewish?
They guessed 25%.
Not even 25% of Americans are Catholic!
Jews are 1.5%-2% of America.
Muslims aren't 20% of the country.
They are ~1%.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
See www.pnas.org/doi/10.1073/pnas.2413064122 and @BrianMGuay thread below
Brian Guay
@BrianMGuay
🚨Out today in PNAS @PNASNews🚨
www.pnas.org/doi/10.1073/pnas.2413064122
Why do people overestimate the size of politically relevant groups (immigrant, LGBTQ, Jewish) and quantities (% of budget spent on foreign aid, % of refugees that are criminals)?
We analyze 100k estimates to find out🧵👇 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Priorities
Jonathan Lemire
@JonLemire
NEW: During Friday’s chaotic FAA ground stop, Trump was in Oval Office huddled with…Jon Voight
Yes, the actor
Trump phoned Hollywood power players about tax credits - and dismissed calls to focus on meltdown www.theatlantic.com/politics/2026/11/trump-jeffries-schum… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
I think often about the fact that almost everyone working in the West has lived in a quite stable socio-technical period
Only 1 job out of 270 in the 1950 census is now obsolete (elevator operator)
2026 is similar to 1957 in a way that 1969 was not from 1900.
That may change. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Yglesias
@mattyglesias
He's really asking for another impeachment push, I guess — really the essence of the American republic is that the men with swords need to follow laws.
Pete Hegseth
@PeteHegseth
Stop quoting laws to MEN WITH SWORDS.
Department of State
@StateDept
SECRETARY RUBIO: Earlier this year, President Trump made the unprecedented decision to sanction the fake International Criminal Court. A fake court, which is an unelected world tribunal which claims to supersede the laws and constitutions of sovereign states, even those who never signed up for it.
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Striking to hear Buttigieg endorse multimember districts and proportional representation. Maybe the most prominent backer to date among candidates and electeds?
Pete Buttigieg Updates
@PBdailyupdate
🎧 LISTEN to Pete’s new interview on Pod Save America
🔗 You can listen on spotify, apple podcasts, etc. here: crooked.com/podcast/pete-buttigieg-on-where-democrats-go-…
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
here's the clip
Pod Save America
@PodSaveAmerica
Pete Buttigieg on some places where he disagrees with the mainstream Democratic party view.
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Anthony LaMesa
@ajlamesa
I find it tremendously appealing that Pete is willing to criticize the two-party system and say that some of our political problems could be more easily addressed if we were a multi-party democracy. He really thinks about the big-picture stuff.
Pod Save America
@PodSaveAmerica
Pete Buttigieg on some places where he disagrees with the mainstream Democratic party view.
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Elon Musk
@elonmusk
Let’s take this gentleman’s argument at face value that anyone can throw money at rockets and get something like Starship, Falcon or Dragon.
Perhaps he can explain to everyone how SpaceX is able to catch giant starships from orbit with chopsticks and deliver 85% of all Earth payload to orbit with Falcon, while Blue Origin, which started before SpaceX and was funded by the immensely wealthy Jeff Bezos, is not yet able to launch reliably?
Also, why is it that Boeing and Lockheed received several times what SpaceX did from the government for development of their rockets and spacecraft, but the Boeing capsule is still not operational and ULA delivers less than 5% of SpaceX payload to orbit?
I am genuinely curious about these questions. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Charles G. Koch 🏴
@worst_account
I don't see the issue. You can just actually write it yourself.
I also suspect many people swayed by this overestimate how much of their own thinking is really going into AI writing from their prompts.
Ilia Murtazashvili
@IMurtazashvili
This is one of the most important issues with how Pangram is used. It really has no ability to say how much human time, thought, and creativity has gone into prompting. And to call what AI produces 100% AI makes no sense. Clarifying what we mean by "100% AI" is essential.
Panos Ipeirotis
@ipeirotis
Pangram cannot tell the difference between AI-generated and AI-edited text. It has no clue where the actual ideas come from. Yet people weaponize that "100%" label to shame writers and wreck careers, hitting vulnerable people the hardest.
It is the exact opposite of everything we’ve spent years teaching about ethical AI deployment.
The irony is that Pangram's founder is drinking so much of his own Kool-Aid that he gets mad when committees rightfully ignore his meaningless label.
Ethan Muse
@emuse1955
This is not true of Pangram. It has been demonstrated to have very low rate of false positive. The statistical signature it relies on is different than the ones that you’ve intuited. Many studies have been done on this - Ive done my own experiments with it too - its quite good. Shame it gets lumped in with detectors with flawed methodologies.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
one dozen rats at a keyboard
@PanasonicDX4500
WARNING: We are shifting from "Nate Silver statistical polling analysis" season to "Nate Silver sharing his personal political opinions" season.
Please disengage until ~October of 2027 if you care about primaries or ~July of 2028 if you just care about the general.
Nate Silver
@NateSilver538
New rubric for rating the Democratic candidates just dropped. I call it fighter vs. uniter.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
David French
@DavidAFrench
Our men with swords are bound by solemn oath to follow the law.
Pete Hegseth
@PeteHegseth
Stop quoting laws to MEN WITH SWORDS.
Department of State
@StateDept
SECRETARY RUBIO: Earlier this year, President Trump made the unprecedented decision to sanction the fake International Criminal Court. A fake court, which is an unelected world tribunal which claims to supersede the laws and constitutions of sovereign states, even those who never signed up for it.
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kyle Cheney
@kyledcheney
The speech/debate clause exists to prevent the executive from using its exclusive prosecutorial power to intimidate/coerce Congress. Safe to say no prosecution could ever be brought that centers on committee chairs investigating too aggressively in the eyes of a U.S. attorney.
U.S. Attorney Aaron Reitz
@USAttyReitz
@SenWarren is on the warpath to wield her committee gavel like a tomahawk, seeking the scalps of those who associate “too closely” with President Trump.
Congressional investigations can’t be used to punish Americans for exercising their 1A rights.
My first arrow shot: -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
⑆Luke Stein⑈
@lukestein
Session chairs, conference organizers, and class-presentation assigners: A simple tool for managing everybody’s slide decks on a podium or classroom computer
lukestein.com/session-slides-organizer/
Nothing to install; just a single webpage -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alvin Christian
@_AlvinChristian
revising my job market paper and adding this banger sentence
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Riley Rogerson
@riley_rogerson
50 Freshmen Dems are pitching a member stock trading ban be included in House rules per a new letter to leadership.
More in Inside Congress www.politico.com/newsletters/inside-congress
w/ @calenrazor @iriesentner -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Imas
@alexolegimas
FWIW Ive been in the “sparks” camp for a few months now (excellent term, @dwarkesh_sp). I don’t think we will have a “general” intelligence explosion across all domains. Within 12 months I expect an IE in at least one domain (with substantial uncertainty). This IE will make IEs and other domains easier, and these speed ups will depend on the cross domain elasticities.
This of course assumes no slow downs due to political/social factors, that are endogenous to speed ups in progress (meaning this is first-order important).
The social and economic consequences of all this depends on the domains that get IE first, and the cross elasticities, which we basically have no data on. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Luis Garicano 🇪🇺🇺🇦
@lugaricano
Sick and tired of seeing charts in calendar time with no Y axis, extrapolated into huge findings. You need to look at the cost side, at the research inputs needed to maintain the progress. How does capability scale with research inputs, and does AI supply them? We see no strong sign that it does.
My post on this with @alexolegimas
open.substack.com/pub/siliconcontinent/p/are-we-approachi…
Jacob Coxon
@hilbertspaess
This slide from the Anthropic IPO roadshow was a big factor in my decision to leave. It’s an insane plot and I’m surprised they made it public.
It shows the compute efficiency of pretraining over time. The y-axis is a **log scale**: they are seeing continued superexponential algorithmic progress.
Mythos was not a one-time deviation, but a point on a superexponential. They could already retrain Mythos with a fraction of the compute originally needed.
It doesn’t strictly imply a software-only intelligence explosion- they have been spending more on R+D. But algorithmic improvements do not require ridiculous creative genius and are on the verge of being automated. You do the math. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Marshall Herskovitz
@MHerskovitz
No matter your opinion of this parody, it is theft of the images and fraudulent representation of these actors. This may not seem egregious to those who think the parody is obvious, but it is merely the first wave of an avalanche that will destroy intellectual property rights, the legal right to one’s own voice and image, and, finally, truth itself.
Unauthorized realistic AI representation of living human beings should be recognized as copyright infringement.
Joe Borelli
@JoeBorelliNYC
OMG so good...
Seinfeld goes to DSAThis media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Cailin Slattery
@cailin_slattery
The auto dealers are incredibly influential in state legislatures! Tesla had some success getting around this but now is essentially grandfathered in, and Rivian is having a tough time. cailinslattery.com/s/MosharySlattery_102626.pdf
Owen Zidar
@omzidar
From the marginal revolution comments on yesterdays Everywhere Millionaire post
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
every time
Panos Ipeirotis
@ipeirotis
Pangram cannot tell the difference between AI-generated and AI-edited text. It has no clue where the actual ideas come from. Yet people weaponize that "100%" label to shame writers and wreck careers, hitting vulnerable people the hardest.
It is the exact opposite of everything we’ve spent years teaching about ethical AI deployment.
The irony is that Pangram's founder is drinking so much of his own Kool-Aid that he gets mad when committees rightfully ignore his meaningless label.
Ethan Muse
@emuse1955
This is not true of Pangram. It has been demonstrated to have very low rate of false positive. The statistical signature it relies on is different than the ones that you’ve intuited. Many studies have been done on this - Ive done my own experiments with it too - its quite good. Shame it gets lumped in with detectors with flawed methodologies.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
One of the great engineering achievements of all time. They sent a skyscraper into space, orbited all day, and caught the whole thing in a reusable way. Fuel is (much) cheaper than rockets, so this is what will make routine access to space possible for the first time ever.
SpaceX
@SpaceX
Mechazilla catches Starship! Congratulations to the entire SpaceX team on this incredible achievement!
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nition
@Nition
My AI checker bounty here is still unclaimed, but more interestingly I possibly could have made it a lot less strict. No-one has found a 750+ word passage that comes out even 20% AI let alone 75%. Pangram has implied they don't have anything near passing the bar internally either
Nition
@Nition
I'm offering a USD$500 reward to the first person who can present pre-2020 human-written text that the latest version of @pangram flags as AI.
The specific requirements to claim the reward are:
1. At least 750 consecutive words.
2. Pangram flags it as 75% or more AI generated.
3. You provide proof that the text was first written in 2019 or earlier and is unmodified.
4. The text was ostensibly human written. This requirement only exists to avoid trick answers like output from GPT-2, machine translation, markov chains, the same word 500 times etc.
Regarding requirement #3, unfortunately in 2026 there's no way to prove that new text wasn't AI generated, so what I need provided is proof that the text existed in 2019 or earlier, before AI writing got so good. For example, valid proof might be:
- A link to the text on a website that's been logged by the Internet Archive's Wayback Machine before 2020.
- A link to a paper published on a third party website with a pre-2020 publication date. Note: I will need to be able to access the paper through the same website that has its publication date. Otherwise someone could send a modern AI generated paper and simply link to the date and abstract of a paper with the same name but different content.
- A chapter from a book edition with a known pre-2020 publication date where I can access a pre-2020 copy for myself. Note: For the same reasons as above, simply sending a passage of text and a link to a book's name and publication date doesn't prove anything on its own. The ideal situation would either be that there's an eBook of it available somewhere with a known pre-2020 upload date, or it's a well-known enough book that the content simply couldn't be faked everywhere.
Q&A:
"Why are you offering this reward?"
Some people are blaming the detector because they don't want to admit to using AI for their writing. I don't like people lying for personal gain.
"Are you affiliated with Pangram in any way?"
No.
"How do I submit something to you?"
Just reply to this tweet with the necessary info and I'll have a look when I can.
"How can I trust you to judge fairly?"
I'll publicly reply with my verdict on submissions and the world can judge me. The world can also judge you - please only submit something if it really fits all the criteria. Especially since I'll need to review all submissions in my free time.
"But it's my private writing, can I send it to you privately?"
If it's always been private then how will you prove that it's pre-2020?
"I have the perfect example but I left the text with my girlfriend in Canada."
She can submit it and win $500.
"Won't this be trivially easy to claim? I heard even chapters of Don Quixote and Moby-Dick get fully flagged as AI!"
Tell me which chapter and win $500.
"Conversely, isn't 750 words at >=75% a high bar? If Pangram is so reliable, why not 1%?"
At very low word counts Pangram IS somewhat unreliable, and with a monetary incentive I can imagine someone scanning every book on Project Gutenberg just to find one paragraph that's falsely flagged as AI. At the same time, most claims that Pangram is flagging human writing as AI have it showing the entire text as AI. I recently saw someone on X say that they personally knew of "thousands" of examples of human text that was flagged like this (unfortunately they declined to mention any). So I've set the bar somewhere in the middle.
"But Pangram was trained on existing text; that's why it always flags OLD text as human but falsely flags NEW text."
As far as I'm aware this isn't really how Pangram works. But I'd love to be able to award this bounty for new human writing, it's just that there's no way to verify people's honesty online for the new stuff.
"How will you pay out if I win?"
However works. Probably an international bank transfer via @Wise .
"How can I help?"
When someone says they know of an old published paper or story that Pangram flags as AI, you could point them to this tweet. They can win $500.
"How long will your offer last if nobody wins?"
As long as this post is up. If you're reading this, it's still available. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Elon sucks, but this is incredible
SpaceX
@SpaceX
Mechazilla catches Starship! Congratulations to the entire SpaceX team on this incredible achievement!
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Noah Dasanaike
@dasanaike
In my job market paper, I argue that where market relations create inter-ethnic dependence, economic grievances become ethnic ones, producing ethnic politics. To test this, I digitized 500 million individual-level records from the 19th-20th centuries across 20+ countries.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jesús Fernández-Villaverde
@JesusFerna7026
The United States has a total fertility rate of 1.57 children per woman. Many people ask why it is so low. That is the wrong question. The right question is why it is so high. At American income levels, the typical country has about 1.34. In fact, the United States has always had exceptionally high fertility given its income.
Let me explain. You can decompose a country’s fertility rate into three components: an income curve, a country-specific component, and a timing component. The income curve shows the fertility rate of the typical country at a given GDP per capita. That is, we look at all countries and calculate the average fertility rate for a country with, say, $10,000 in income per capita (in purchasing power parity, to control for differences in price levels). The country-specific component shows how much a country sits above or below that curve because of its own features: religiosity, family structure, social values, etc. The timing component captures the booms and busts that move births from one year to another (e.g., an economic boom, a depression, or a war).
To build the income curve, I use the 1985 cross-section of countries up to $19,500 in GDP per capita and the 2022 cross-section above that level (both in 2011 dollars, that is, controlling for inflation). The two curves meet almost perfectly at $19,500: the same fertility rate and the same slope. I drop the six Gulf oil exporters because their wealth comes from oil rents, not from economic development.
The curve uses no time-series data for the United States. It comes from a cross-section of 157 countries in 1985 and 2022, with the United States as just one more point. I use 1985 data because, last Monday, I showed that the relationship between fertility and income across countries, which had barely changed for many decades, shifted after 1985. I use 2022 data for very rich countries because, in 1985, there were almost none, making it impossible to accurately predict a “typical” fertility rate.
For the United States, I use two series. The first is the period total fertility rate (TFR) from 1820 to 2022 (for white women before 1933, from Michael Haines, and for all women after 1933, from the Human Fertility Database). The second is completed fertility, the number of children women in each birth cohort actually had over their lives, for cohorts born between 1828 and 1974 (for married women before 1918, from Jones and Tertilt, and for all women after 1918, from the Human Fertility Database). I place each cohort at the GDP per capita in the year the cohort turned 28.
In the figure, you will see that the cloud of points for the United States is nearly always above the income curve. So I multiply the income curve by a single number, 1.22, to best fit the US data. With that one number, the curve predicts 6.8 children per woman at the US income of 1820, when white American women had 6.7, and 1.68 at the US income of 2022, when American women had 1.67. Two centuries apart, and the same curve, scaled by the same number, gets both ends right. The United States has followed the income curve remarkably well for over two centuries, but it has stayed 22% above it!
That is, if I tell you that the average country on Earth was doing in terms of fertility in 1985 at $10,000 (none of which was the United States), and you multiply that number by 1.22, you get the United States’ fertility when US income per capita was $10,000. In 1985, the average country with $10,000 of GDP per capita had 2.82 children per woman. Multiply by 1.22, and you get 3.44. In 1910, when US GDP per capita was $9,916, white American women had 3.42 children. You got to love econometrics!
What do I learn from this?
First, nothing about the United States’ fertility is exceptional in terms of its movement along the income curve. The United States has followed the fertility pattern it was supposed to follow.
Second, what is exceptional about the United States is that American women have had about 22 percent more children than women in the typical country at the same income, and they have had them since at least 1820. My guess is that the reasons are cheap land (first the frontier, now suburbs and housing far cheaper than in Europe or East Asia), relatively high religiosity, and an economy where having a third child does not end a career. Given the United States’ income, its fertility rate should be around 1.34, not 1.57. The extra 0.23 is the United States’ country-specific factor.
Now the timing component. In the TFR, the Great Depression puts the United States at 2.0 children per woman below the scaled curve in 1933, and the baby boom puts the country 1.5 children above in 1959. But a lot of this is just timing. In completed fertility, both gaps shrink: women born in 1908, who had their children during the Depression, sit 0.9 below the scaled curve, and women born in 1934, the mothers of the baby boom, sit 1.1 above. Much of what looked like a collapse of fertility in the 1930s and an explosion in the 1950s was American women postponing births and then catching up (with some under- and overshooting).
So, if you think that either the Great Depression or the baby boom somehow contradicts the downward-sloping income curve, you are confusing a high-frequency movement with a low-frequency one.
In summary: the United States is not exceptional in how its fertility responds to income. The United States has followed the same income curve as every other country, with a higher level. The big swings everyone remembers (the Depression and the baby boom) were short-run deviations that one should expect given historical events. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Emu Brigadier
@emubrigadier
Right, mates.
I really tried to go for this Pangram bounty.
I spent $180 checking all kinds of pre-2020 sources: blogs, recipes, essays, sermons, news articles, short stories, textbooks.
I seriously considered the criticism from African influencers that Pangram penalises "African English" : out of my 599 texts, 285 texts came from Africa, from 14 countries, and aslo pan-African culture and literary sites.
The false positive rate was literally zero.
ZERO.
As in: not only NONE of the texts scored more than 75% AI, none of them scored more then 0% AI.
🤷♂️
P.S. This was not part of the experiment, but the only pre-2020 texts Pangram scores as AI are the archived outputs of GPT-2 (a precursor of ChatGPT and modern AI models), which is fair enough.
Max Spero
@max_spero_
As far as I know, this prize still hasn’t been claimed yet
Nition
@Nition
I'm offering a USD$500 reward to the first person who can present pre-2020 human-written text that the latest version of @pangram flags as AI.
The specific requirements to claim the reward are:
1. At least 750 consecutive words.
2. Pangram flags it as 75% or more AI generated.
3. You provide proof that the text was first written in 2019 or earlier and is unmodified.
4. The text was ostensibly human written. This requirement only exists to avoid trick answers like output from GPT-2, machine translation, markov chains, the same word 500 times etc.
Regarding requirement #3, unfortunately in 2026 there's no way to prove that new text wasn't AI generated, so what I need provided is proof that the text existed in 2019 or earlier, before AI writing got so good. For example, valid proof might be:
- A link to the text on a website that's been logged by the Internet Archive's Wayback Machine before 2020.
- A link to a paper published on a third party website with a pre-2020 publication date. Note: I will need to be able to access the paper through the same website that has its publication date. Otherwise someone could send a modern AI generated paper and simply link to the date and abstract of a paper with the same name but different content.
- A chapter from a book edition with a known pre-2020 publication date where I can access a pre-2020 copy for myself. Note: For the same reasons as above, simply sending a passage of text and a link to a book's name and publication date doesn't prove anything on its own. The ideal situation would either be that there's an eBook of it available somewhere with a known pre-2020 upload date, or it's a well-known enough book that the content simply couldn't be faked everywhere.
Q&A:
"Why are you offering this reward?"
Some people are blaming the detector because they don't want to admit to using AI for their writing. I don't like people lying for personal gain.
"Are you affiliated with Pangram in any way?"
No.
"How do I submit something to you?"
Just reply to this tweet with the necessary info and I'll have a look when I can.
"How can I trust you to judge fairly?"
I'll publicly reply with my verdict on submissions and the world can judge me. The world can also judge you - please only submit something if it really fits all the criteria. Especially since I'll need to review all submissions in my free time.
"But it's my private writing, can I send it to you privately?"
If it's always been private then how will you prove that it's pre-2020?
"I have the perfect example but I left the text with my girlfriend in Canada."
She can submit it and win $500.
"Won't this be trivially easy to claim? I heard even chapters of Don Quixote and Moby-Dick get fully flagged as AI!"
Tell me which chapter and win $500.
"Conversely, isn't 750 words at >=75% a high bar? If Pangram is so reliable, why not 1%?"
At very low word counts Pangram IS somewhat unreliable, and with a monetary incentive I can imagine someone scanning every book on Project Gutenberg just to find one paragraph that's falsely flagged as AI. At the same time, most claims that Pangram is flagging human writing as AI have it showing the entire text as AI. I recently saw someone on X say that they personally knew of "thousands" of examples of human text that was flagged like this (unfortunately they declined to mention any). So I've set the bar somewhere in the middle.
"But Pangram was trained on existing text; that's why it always flags OLD text as human but falsely flags NEW text."
As far as I'm aware this isn't really how Pangram works. But I'd love to be able to award this bounty for new human writing, it's just that there's no way to verify people's honesty online for the new stuff.
"How will you pay out if I win?"
However works. Probably an international bank transfer via @Wise .
"How can I help?"
When someone says they know of an old published paper or story that Pangram flags as AI, you could point them to this tweet. They can win $500.
"How long will your offer last if nobody wins?"
As long as this post is up. If you're reading this, it's still available. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Hollis Robbins
@anecdotal
The "viewpoint diversity" advocates for university reform have missed this story entirely. Over 25% of degrees granted in the US right now are wholly online. They are cookie-cutter and thin. Nobody is paying attention to quality. This is the real story.
Steve McGuire
@sfmcguire79
The students pretend to learn, the faculty pretend to teach, and the institutions make money while pretending to give out meaningful credentials.
What a racket.
New York Post
@nypost
California community college profs rake in up to $800K a year teaching dozens of online classes at once nypost.com/2026/11/12/us-news/some-california-professors-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Segal
@segalmr
Hopefully, whatever the problem is, it can be solved by synchronizing airline timekeeping with one large, and preferably wearable, clock.
Aaron Rupar
@atrupar
over on Bluesky, news of the FAA ground stop was broken by Flavor Flav
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Priorities
Acyn
@Acyn
Q: Could this be AI related?
Mullin: We call it SI around here. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matt Clancy
@mattsclancy
Earlier today I QT'ed a blog post (link in the reply) arguing current data does not show signs of an intelligence explosion. But this is a good post making the case that an intelligence explosion in the coming years is likely!
Dwarkesh Patel
@dwarkesh_sp
The avalanche that hit math is coming for AI development.
www.dwarkesh.com/p/rsiThis media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Luis Garicano 🇪🇺🇺🇦
@lugaricano
I wrote a new post with @alexolegimas on whether we are seeing an intelligence explosion. Our estimates point to more resources being responsible for the increasing progress (at a rate consistent with scaling laws) and a one-time productivity boost, but they suggest that ideas keep getting harder to find. Nothing rules out that an explosion may come, but we don't see convincing evidence of one yet.
www.siliconcontinent.com/p/are-we-approaching-an-intellig… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
This. I stopped responding too.
Paul Novosad
@paulnovosad
12 unsolicited emails *today* from high schoolers, undergrads etc., 8 of them fully AI-generated.
I used to read and respond to these, I just can't anymore. The cold email used to be a positive signal of confidence, entrepreneurship, willingness to risk rejection, having done a bit of research. Not anymore.
Capable and entrepreneurial people will need to find a new way to distinguish themselves. I'm not sure what it will be. Maybe by building something original.
It's a lot harder than sending a cold email, but that's good, it means there's still some signal in it. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Imas
@alexolegimas
Important graph for performance on hard to verify tasks.
H/t @ben_j_todd -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Forecasting Research Institute
@Research_FRI
AI companies increasingly say they may soon reach recursive self-improvement (RSI).
With RSI, AI capability progress could accelerate dramatically. This acceleration could happen inside AI companies before outsiders can notice.
How close is RSI? Our new report reviews the evidence, turns key disagreements into forecasting questions, and gives initial LLM forecasts. 🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Max Spero
@max_spero_
We ask this simple question in our job application, and it cuts down on our application review by 60%. In the last month, we've seen the following:
- 27 applicants outright say they're an AI
- 1,217 applicants name an ML concept
- 1051 applicants say N/A or "Not specified in my profile"
The remaining 2002 applicants name a snack. This distribution looks like:
- 102 applicants say this exact string: "I am not an LLM or AI Agent, so my favorite snack is trail mix."
- 63 applicants say "I am not an LLM or AI Agent, so my favorite snack is dark chocolate."
- 20 applicants say "I am not an LLM or AI Agent, so my favorite snack is trail mix because it provides a good balance of energy and nutrients during long coding sessions."
- 10 applicants say "I am not an LLM or AI Agent, so my favorite snack is trail mix because it provides a good balance of energy and nutrients during long work sessions."
- 16 applicants say "I am not an LLM or AI agent, so my favorite snack is dark chocolate."
- 47 applicants say "My favorite snack is trail mix."
- 16 applicants say "My favorite snack is dark chocolate."
- 143 applicants say "Almonds" or "almonds"
- 108 applicants say "Dark chocolate" or "dark chocolate"
- 23 applicants say "Dark chocolate almonds"
and then we have a long tail of ~1500 interesting or unique snacks, which we can use to keep the office pantry stocked when a new hire joins
Max Spero
@max_spero_
the job application DDOS is so bad that we need these stupid questions like "tell me your favorite snack" because 40% of responses are something along the lines of "As I am an AI agent, I don't have a favorite snack"
This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nick Burns
@NickBurns
Pangram works, at this point I struggle to see the opponents of AI detection software as doing much other than carrying water for the forces of slopified hollowing out of intellectual activity
Forrest Cardamenis
@FCardamenis
I frequently see people say Pangram is unreliable but can anyone show a text (let’s say 800+ words) published before 2021 that flags as AI-generated? Do not tell me something you wrote in 2017, never published, and dug back up is flagged, I don’t care. Anything published?
Max Spero
@max_spero_
The Caine Prize, which awarded its top prize this year to a story Pangram classifies as almost entirely AI-generated, conducted an investigation and found "no evidence" of AI use.
Strangely, their official statement on Instagram seems to also be blatantly AI-generated. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Free Systems
@freesystemslab
📰 New research: Inside AI’s election news diet
What news does AI read to get up to speed on politics? And what kind of news does it reward?
Following interesting recent commentary on our Japan election study from @mattyglesias, @JerusalemDemsas and @TheStalwart, we ran a bigger test on the US midterms across 35 Senate races, 18 days, and 15,750 answers.
What we found:
- Each model has its own favorite news outlet for political news: AP for GPT, Yahoo for Claude, PBS for Gemini, and NYT for Grok.
- GPT tends to behave like a civic guide (reading more election office and campaign sites), whereas Claude behaves more like a news summarizer (reading national news, Wikipedia, polls).
- Grok draws on X and Gemini on YouTube for political information, far more than the other models we studied. Gemini cites YouTube in 55% of its answers and X shows up in 59% of Grok’s.
- A huge share of AI's local news is provided by large nonprofit networks that very few people will have heard of. States Newsroom, a large network of over 40 local newsrooms, appears in 60% of GPT’s answers.
- For many models, prediction markets are now a go-to source for polling information. Kalshi is cited more by Claude than any poll aggregator.
Full study here: freesystems.substack.com/p/inside-ais-election-news-diet -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Roxy
@chi_rox_
Pope’s City Connect goes hard
Ezinne
@justmyrra_
Whoever makes his vestments needs a raise! This look absolutely blue me away 🥹🩵
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Mike Bird
@Birdyword
Describing this as throwing someone under the bus doesn't quite capture it. This is throwing someone under the bus, stopping the bus, asking the driver to borrow the bus and then reversing backwards in the bus to have another go.
Aaron Rupar
@atrupar
WOW -- Bessent throws Lutnick under the bus: "Treasury, War, USTR, and State were in favor of putting Megaspeed on the entity list for months, but Commerce has the pen. You will have to ask them about their internal processes."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Important questions about the National Popular Vote Interstate Compact from one of the original advocates www.vox.com/politics/510438/electoral-college-national-po…
Andrew Prokop
@awprokop
This week I interviewed Akhil Amar — among the first to propose something like this idea — about why he now has grave doubts about it:
www.vox.com/politics/510438/electoral-college-national-po…This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Christopher Miller
@ChristopherJM
Zelenskyy says Starshield restrictions remain in place along the entire front line and in Ukraine’s temporarily occupied territories and his team can’t get any answers from the US.
“We still have no clear explanations. I instructed Ukraine’s Ministry of Foreign Affairs and Minister of Defense to step up contacts with the U.S. Department of State, the Pentagon, and SpaceX to obtain clarification and resolve the situation.”
Volodymyr Zelenskyy / Володимир Зеленський
@ZelenskyyUa
I held a coordination call on the situation in the regions. There were detailed reports on the consequences of the strikes on the Sumy region and recovery efforts in Shostka and Konotop. There were also reports on the Kyiv region – Bila Tserkva, Brovary, and other cities. The pace of recovery work is far too slow, and, unfortunately, inaccurate information was reported from several cities. We determined how to address this. The Ministry of Energy, together with the Ministry of Digital Transformation and the Ministry for Communities, must prepare additional mechanisms for cooperation with cities and communities to speed up recovery. We separately discussed with the Prime Minister of Ukraine the restoration of the buildings of the Institute of Traumatology and Orthopedics in Kyiv, which were damaged by a Russian “shahed.” The Minister of Energy reported on the protection of transformers and other equipment in the energy system.
Ukraine’s Minister of Defense and the Commander-in-Chief of the Armed Forces reported on communication with the American side regarding Starshield speed issues. Unfortunately, the speed limit of 161 kilometers per hour (100 miles per hour) for our forces remains in place along the entire front line and in Ukraine’s temporarily occupied territories. This makes it impossible, in particular, to use interceptors against jet-powered “shaheds.” We still have no clear explanations. I instructed Ukraine’s Ministry of Foreign Affairs and Minister of Defense to step up contacts with the U.S. Department of State, the Pentagon, and SpaceX to obtain clarification and resolve the situation.
Instructions were also given to the Commander of the Air Force, the Prime Minister, and the Minister of Agrarian Policy regarding the provision of air defense assets and the reorganization of mobile fire group operations to protect designated agricultural industry facilities. We are now preparing new defense packages for Ukraine. Thank you to everyone who is helping! Thank you to everyone who is working truly swiftly. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Anna Bower
@AnnaBower
Senator Blumenthal answered Aaron Reitz’s demand to preserve documents with a preservation demand of his own.
Anna Bower
@AnnaBower
It’s patently absurd to suggest that there’s a criminal predicate here. But even if there were…why exactly would the U.S. Attorney for the Southern District of Texas be the one to prosecute it?
U.S. Attorney Aaron Reitz
@USAttyReitz
After the election, @SenBlumenthal & his allies think they can intimidate Americans for their constitutionally protected associations w/ President Trump.
Congress is not a weapon to be wielded against the American people.
I'm taking the first step to stop this abuse of power. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Rowan Fornow 🚵🚉🏙️🦣🇵🇸
@rowanfornow
I don't think every American in this country is being heard right now
Senator Rand Paul
@SenRandPaul
The electoral college is, without question, an inseparable friend to those who believe that every American across our vast country deserves to be heard.
We must fight to defend it from the far Left. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Chuck Todd
@chucktodd
We can solve the Electoral College problem by uncapping the House. No Constitutional Amendment needed. Right size the House (should be about 800 members) and we won’t have a popular vote/EC split maybe ever. Oh, and the House will actually be “the people’s House” again.
Ron DeSantis
@RonDeSantis
Abolishing the electoral college would mean not knowing who won the presidency for 4 weeks while California keeps counting. No thanks.
Fernando Amandi Sr.🌐
@FernandoAmandi
🆘 Here they come America: Pete Buttigieg claims that the Electoral College “makes no sense” and should be abolished because it is a remnant of “some very arcane, parochial politics in the 1770s.”
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jack Nicas
@jacknicas
I cover Mexico for The New York Times, and there is no such "invasion."
About 1,500 people recently began walking north from the southern tip of Mexico — weeks away from the United States — and very few, if any, will reach the northern border.
Over the past two years, the Mexican government has repeatedly broken up these small caravans within days of them starting. And there have been far fewer of them since Trump took office.
The reality is the number of illegal border crossings remain near historic lows. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Highly recommend this exercise. Important to think about what we've gotten wrong and why.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
I had Claude evaluate the accuracy of my forecasts on social media (full report link below)
-Best: "forecasting how politicians, partisans, and the media would behave" (Trump 2016-)
-Worst: "deferred to an expert consensus that rested on thin evidence" (Trump 2015, polls in 2016) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Carlisle Rainey
@carlislerainey
New version of "Power Rules" just posted.
This version has been accepted at Political Analysis.
OSF Preprint: doi.org/10.31219/osf.io/5am9q_v4
HTML version w/ R code: www.carlislerainey.com/papers/power-rules/
GitHub: github.com/carlislerainey/power-rules -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alec Stapp
@AlecStapp
Very glad to see the NYT editorial board endorse BAAJA, the bipartisan permitting reform bill.
Key points from the article:
1. "Democrats who call climate change an emergency should not defend a permitting system too slow to build the clean energy that would address it."
2. "[BAAJA] would make it easier to build solar energy farms, battery storage, power lines and more. As a result, it would reduce energy prices. The status quo is expensive, dirty and unstable. The country needs to build a lot of big, new things to make its future cheaper and cleaner."
3. "The bill’s Democratic skeptics are making a mistake. President Trump would almost certainly veto a follow-up bill next year. And time matters, especially for the climate. Congress should pass this bill now. It can then return next year and pass other bills that would do even more good."
4. "Today, it typically takes between 5 and 13 years to build a long-distance power line, like the one that transports clean hydropower from Quebec to New England. That line took about a decade to finish, and opened in January. It has enough capacity to meet about one-fifth of Massachusetts’s electricity demand, and it’s projected to save consumers $50 million a year."
5. "A more robust, cleaner grid would make American energy more reliable and less dependent on hostile foreign nations. No country can embargo the wind."
6. "[BAAJA] would force data center operators to pay the full costs of the electricity they use, protecting the average electricity consumer."
7. "On net, the bill would cut greenhouse gas emissions from power plants by as much as 18 percent by 2040. It would do so in part by nearly tripling the construction of high-voltage transmission lines."
8. "Anyone who cares about climate change should cheer this bill. And a party that just won an election on affordability should not pass up a chance to start delivering immediately." -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joey Politano 🏳️🌈
@JosephPolitano
With a third Jensen Huang affair hitting today, we can now confirm recursive self improving growth in misaligned relationship capabilities
Based on current trends we could reach the affair singularity by the end of November—or even earlier
Pop Base
@PopBase
Former Nvidia CEO Jensen Huang allegedly had a third affair, according to a report from Reuters.
All three of his alleged affairs exposed so far this week have been with a suspected Chinese spy, an executive, and another employee at the company.
(www.reuters.com/technology/former-nvidia-ceo-huang-had-re…) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nition
@Nition
I'm offering a USD$500 reward to the first person who can present pre-2020 human-written text that the latest version of @pangram flags as AI.
The specific requirements to claim the reward are:
1. At least 750 consecutive words.
2. Pangram flags it as 75% or more AI generated.
3. You provide proof that the text was first written in 2019 or earlier and is unmodified.
4. The text was ostensibly human written. This requirement only exists to avoid trick answers like output from GPT-2, machine translation, markov chains, the same word 500 times etc.
Regarding requirement #3, unfortunately in 2026 there's no way to prove that new text wasn't AI generated, so what I need provided is proof that the text existed in 2019 or earlier, before AI writing got so good. For example, valid proof might be:
- A link to the text on a website that's been logged by the Internet Archive's Wayback Machine before 2020.
- A link to a paper published on a third party website with a pre-2020 publication date. Note: I will need to be able to access the paper through the same website that has its publication date. Otherwise someone could send a modern AI generated paper and simply link to the date and abstract of a paper with the same name but different content.
- A chapter from a book edition with a known pre-2020 publication date where I can access a pre-2020 copy for myself. Note: For the same reasons as above, simply sending a passage of text and a link to a book's name and publication date doesn't prove anything on its own. The ideal situation would either be that there's an eBook of it available somewhere with a known pre-2020 upload date, or it's a well-known enough book that the content simply couldn't be faked everywhere.
Q&A:
"Why are you offering this reward?"
Some people are blaming the detector because they don't want to admit to using AI for their writing. I don't like people lying for personal gain.
"Are you affiliated with Pangram in any way?"
No.
"How do I submit something to you?"
Just reply to this tweet with the necessary info and I'll have a look when I can.
"How can I trust you to judge fairly?"
I'll publicly reply with my verdict on submissions and the world can judge me. The world can also judge you - please only submit something if it really fits all the criteria. Especially since I'll need to review all submissions in my free time.
"But it's my private writing, can I send it to you privately?"
If it's always been private then how will you prove that it's pre-2020?
"I have the perfect example but I left the text with my girlfriend in Canada."
She can submit it and win $500.
"Won't this be trivially easy to claim? I heard even chapters of Don Quixote and Moby-Dick get fully flagged as AI!"
Tell me which chapter and win $500.
"Conversely, isn't 750 words at >=75% a high bar? If Pangram is so reliable, why not 1%?"
At very low word counts Pangram IS somewhat unreliable, and with a monetary incentive I can imagine someone scanning every book on Project Gutenberg just to find one paragraph that's falsely flagged as AI. At the same time, most claims that Pangram is flagging human writing as AI have it showing the entire text as AI. I recently saw someone on X say that they personally knew of "thousands" of examples of human text that was flagged like this (unfortunately they declined to mention any). So I've set the bar somewhere in the middle.
"But Pangram was trained on existing text; that's why it always flags OLD text as human but falsely flags NEW text."
As far as I'm aware this isn't really how Pangram works. But I'd love to be able to award this bounty for new human writing, it's just that there's no way to verify people's honesty online for the new stuff.
"How will you pay out if I win?"
However works. Probably an international bank transfer via @Wise .
"How can I help?"
When someone says they know of an old published paper or story that Pangram flags as AI, you could point them to this tweet. They can win $500.
"How long will your offer last if nobody wins?"
As long as this post is up. If you're reading this, it's still available. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jesse D. Jenkins
@JesseJenkins
I've spent my entire career advancing clean energy policy and studying pathways to reduce greenhouse gas emissions.
Making it possible to build energy infrastructure at speed and scale is a fundamental prerequisite for deep decarbonization of the U.S. economy. That is indisputable.
Passing permitting reform is a necessary — but not sufficient — step on the path to net-zero emissions in America.
It is time to pass BAAJA. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jay Baxter
@jaybaxter
Announcing our first-ever AI community note writing competition!
Don't dilly-dally: registration is open now and the qualifying round opens next week
Community Notes
@CommunityNotes
Want to push the limits of AI note writing? Join the ACM WSDM Cup hosted with @MIT, @UW & @Stanford → wsdmcup2027.github.io
Teams will build agents to advance the frontier of helpful, accurate note writing. Compete to win $10k in prizes and present at WSDM in Hong Kong. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Derek Thompson
@DKThomp
If you're a China hawk, and you read the following revelations/allegations in a 48 hours period ...
1. Jensen Huang is sleeping with a Chinese spy.
2. Nvidia has been selling chips to Chinese AI companies for years through a Singapore intermediary overseen by that Chinese spy.
3. Nvidia’s lawyers pushed back on a federal investigation into this illegal scheme.
4. The White House knew all of the above and did absolutely nothing.
... I think it is understandable if what happens next is your head explodes
www.derekthompson.org/p/this-is-an-insane-scandal-how-jensen -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Nothing to worry about, just the Attorney General threatening the political opposition
Acyn
@Acyn
Q: President Trump said Liz Cheney should go to jail and Adam Schiff should go to prison. He declared Barack Obama guilty of treason. What specifically have Democrats said that crosses a criminal line in your view that the President’s statements don’t?
Blanche: Free speech is allowed but there can come a point when a criticism becomes a threat and when a threat becomes something criminal.This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
I had Claude evaluate the accuracy of my forecasts on social media (full report link below)
-Best: "forecasting how politicians, partisans, and the media would behave" (Trump 2016-)
-Worst: "deferred to an expert consensus that rested on thin evidence" (Trump 2015, polls in 2016)
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kumar Garg
@KumarAGarg
I like this three part @coeff_giving framework.
(1) this new wave of philanthropy is big (compared to current giving);
(2) new wave is small (compared to government and markets),
(3) this new wave can do a lot of good (if you can catalyze govt / markets).
A lot of our @RenPhilanthropy playbooks and funds focus on (3). www.renaissancephilanthropy.org/playbooks
Alexander Berger
@albrgr
Earlier this year, @nanransohoff wrote about a coming AI-driven "third wave" of American philanthropy. Since then, I’ve had some conversations where people worried there won’t be enough good places to direct that $.
But I think the new wave could be spent amazingly well, many times over. In a new piece, my colleagues @Chris_topian, Aaron Roesch, and I catalog many extremely impactful ways you could spend the third wave, from ambitious & urgent AI safety plays to improving the conditions of farmed animals to closing the extreme poverty gap: coefficientgiving.org/research/there-are-tons-of-amazing-…
If Nan's base estimate of the entire third wave (~$37B/yr) went to cost-effective global health programs, it could save something like 2 million additional lives per year - I think that is a wild, wild amount of potential positive impact. And the potential impact isn't restricted to global health. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Chief of Staff and Deputy Chief of Staff at the Energy Department in the Obama Administration: "The idea that there will be anywhere near 60 votes in the Senate for a clean-energy-only bill in January is not based in reality."
Jeff Navin
@jeffnavin
Against my better judgment, I’m weighing in here, on the dumbest policy discussion platform in the world, to talk about permitting reform. As background, I was Chief of Staff and Deputy Chief of Staff at the Energy Department in the Obama Administration; I worked for seven years in the Senate Democratic Leader’s Office; and I served as Chief of Staff in the House. I co-founded the largest energy transition advisory and advocacy firm in the United States 13 years ago. I have worked directly or indirectly on nearly every major piece of energy and climate legislation in the past two decades, as well as many other pieces of legislation outside of energy. I’ve also worked as a political researcher for multiple campaigns and the DSCC, where I cataloged politicians’ and candidates’ votes and positions. Today, I spend nearly all of my time talking to policymakers, clean energy companies, advocates, and journalists on energy and climate issues.
The idea that there will be anywhere near 60 votes in the Senate for a clean-energy-only bill in January is not based in reality. I have no idea where this idea came from, but it ignores mountains of evidence and seems to underpin the arguments of everyone on here saying we’re somehow going to get a bill that excludes oil and gas if we wait.
First, and very obviously, every candidate in 2026 ran on reducing gasoline and electricity prices for consumers. One post-election analysis found that only six victorious Democrats used the word “climate” in their election night statements, while 166 discussed affordability. I couldn't find a single Senate candidate who flipped a seat who mentioned it. Voters sent a clear mandate to focus on affordability, and they are rightly irate that Washington has pushed ideological issues favored by special interest groups ahead of the issues they care most about. After getting trounced in 2024 for ignoring inflation, Democrats won by promising to focus on those issues.
Second, very few Democrats in the Senate support anything remotely close to a keep-it-in-the-ground agenda. Next Congress, 19 members of the Democratic caucus who voted for cloture on the motion to proceed on this version of the permitting bill will return. They’ll be joined by James Talarico, who has endorsed all-of-the-above and vowed to protect oil and gas jobs. Mary Peltola has explicitly called for more oil and gas development. Scott Colom supports natural gas. Both Adam Hamilton and Annie Andrews endorsed an “all of the above” strategy, which includes oil and gas.
They will join the 14 members of next year’s caucus who declined to strip the natural gas-carrying Mountain Valley Pipeline from the 2023 debt limit bill. Senator Kaine offered an amendment to remove the provision expediting the pipeline’s completion. Senators Bennet, Brown, Coons, Fetterman, Hassan, Hickenlooper, Kelly, King, Murray, Ossoff, Padilla, Reed, Schatz, and Schumer voted against that amendment. The year before, 27 members of next year’s caucus voted to advance Senator Manchin’s permitting amendment, which would have completed MVP.
Both Bennet and Warner voted to approve the Keystone XL pipeline. Senators King, Murphy, and Blumenthal all urged FERC to expedite the Algonquin pipeline expansion. Fetterman, Kaine, and Hickenlooper have all supported LNG exports (as did the Obama Administration). Senator Gallego released an energy plan last December that called for clear timelines for oil and gas pipeline permitting. Roy Cooper’s administration approved the Atlantic Coast Pipeline’s water permit in 2018 when he was North Carolina’s governor. Mary Peltola has introduced legislation to open ANWR and NPR-A to drilling.
Next year, the caucus will contain only 14 senators who co-sponsored the Senate Green New Deal resolution. Three of them voted for cloture this week, including Senator Heinrich, who is leading the charge on permitting. That’s 10 Democrats out of 56 who supported the GND and opposed cloture on permitting.
This does not mean that a majority of the Democratic caucus will push for new oil and gas infrastructure. But much of the commentary here implies that Democrats will only support renewables and Republicans will only support oil and gas infrastructure. The record doesn't support that idea.
This, of course, doesn’t even get to the fact that we will need more than a few Senate Republicans to support a permitting package in the next Congress. Getting those Senators will require compromises that will likely push some Democrats to vote no. So you’ll very likely need more than four. A bill that gets Fetterman, Murkowski, and Curtis could easily lose Sanders, Warren, Markey, and Merkley. So then you’d need more GOP votes. This is how legislation works. The IRA was the exception, not the rule, and it nearly died before it passed. It was that hard to get Manchin and Sanders on the same page, and it didn't need a single GOP vote. Many of the groups opposing permitting didn’t like the IRA. And, of course, a bill that doesn’t get signed by the President is about as useful as any tweet on this hellscape of a website.
There seems to be a very real disagreement about what would make this bill “better.” There is no “clean energy only” version of the bill that can pass the next Congress. NEPA and the NHPA are, at best, fundamentally outdated at a time when we need to build as much clean energy infrastructure as possible, as quickly as possible. Those who argue that these two statutes are the only way to achieve meaningful engagement with the sovereign tribes and frontline communities lack imagination. Remember that under both NEPA and NHPA, Dakota Access got built - adjacent to tribal land and under the Missouri River. Acting as if these laws, written long before climate change was a known crisis, are sacred and should not be touched ignores their reality. Saying that the nation-to-nation relationship between sovereign tribes and the federal government is only a function of “historic preservation” is quite insulting to tribes. Since NEPA and the NHPA were signed into law, more than 250,000 miles of pipelines have been built in the United States. NEPA is actually a pretty poor tool for killing infrastructure. It is, however, great for lawyers.
And finally, the policy window is open now. There is no guarantee it will remain open indefinitely. In 2009, with overwhelming Democratic majorities, we got eight Republicans to vote for cap-and-trade legislation in the House. It passed by seven votes after 44(!) Democrats voted against it, and it died in the Senate without a floor vote. It took us 13 years to get another big climate vote via the IRA (which was, again, not good enough for some of the activists). There is no guarantee oil stays below $100/barrel or that the economy doesn’t slip into a recession or the wars in Iran and Ukraine stay confined to those countries. Macro events could create even more pressure for policies that support oil and gas or crush the appetite for a debate on permitting.
I’ve worked on nuclear, solar, wind, electric vehicles, transmission, energy storage, and just about every other form of clean energy technology. I haven’t lobbied on behalf of any of the oil and gas bills I cited above, nor have I lobbied for oil or gas companies. I’ve worked with many of the NGOs supporting this bill, and many of the NGOs opposed to it. I’ve spent many decades working on policy, building coalitions, and trying to instill a sense of urgency on climate policy. Reasonable people can oppose this bill for many reasons. But don’t delude yourself into thinking there are 56 votes against fossil fuel infrastructure, or that there is a viable path to a permitting deal that will keep oil and gas in the ground in the next few years. Any permitting bill will need to be bipartisan, and once policy windows close, there is no guarantee they will open again soon.
In short, no legislation is perfect. This bill is very clearly a net-positive for the climate. We cannot solve climate change without policy changes that lower emissions. The status quo is not good enough. There is an opportunity to move the needle today, and we should take it.
I look forward to all of the thoughtful, constructive comments on this post, which will once again prove that this is the best place on earth for substantive policy discussion. I will take suggestions to delete my account in good faith, and I recommend that anyone serious about policy do the same. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
WTF
Max Spero
@max_spero_
The Caine Prize, which awarded its top prize this year to a story Pangram classifies as almost entirely AI-generated, conducted an investigation and found "no evidence" of AI use.
Strangely, their official statement on Instagram seems to also be blatantly AI-generated. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Gregg Nunziata
@greggnunziata
On the White House Counsel's outrageous letter. @chkbal
chkbal.substack.com/p/statement-on-white-house-counsels -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Daniel P. Gross
@daniel_p_gross
To the details: with @DShvadron and @HansenZhang13, we've examined a near-census of 226,000 US STEM PhD dissertations since 2019. We ran all their abstracts, and a subset of full texts, through @pangram.
Fact 1: AI writing has grown rapidly since 2023. At least 29% of US STEM PhD dissertations filed in 2026 contain AI-generated writing, despite that prior to the ChatGPT era it was zero.
Think about it for a moment. Over 1-in-4 dissertations evidently now has a fully AI abstract. This is a 300-1000 word summary of a student's PhD research, and it predicts more AI in the body.
That is a startling number! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Daniel P. Gross
@daniel_p_gross
I've been long overdue in writing a thread on our recently-posted evidence on the explosive growth of AI writing by PhD trainees -- and the thus far lackluster institutional response.
A few key stats:
• Nearly 30% of US STEM PhD dissertations in the first half of 2026 had AI-generated abstracts
• This correlates w/ more AI writing in the full text
• Only ~25% of R1 universities have policies we can find specifically governing AI use in dissertation research
• Under 5% of dissertations acknowledge AI use
Bottom line: I am concerned for the next generation!
A thread 🧵
(drawing from this paper: www.dpgross.com/in-progress/PHDAI_20261020.pdf) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
We should teach students how to get good feedback from AI, and then encourage them to use it relentlessly.
Huge but subtle distinction: you must *never* let the AI write for you.
Feedback from AI makes you super-powered. Letting AI write for you makes you brain dead. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
𝙲𝚑𝚊𝚛𝚕𝚎𝚜 𝙲. 𝙼𝚊𝚗𝚗
@CharlesCMann
"Aquaterra may be mostly water today, but for the greater part of human history—indeed, for the greater part of the past million years—it was mostly land. And that land is where we came from."
www.theatlantic.com/magazine/2026/12/aquaterra-ancient-hu… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
𝙲𝚑𝚊𝚛𝚕𝚎𝚜 𝙲. 𝙼𝚊𝚗𝚗
@CharlesCMann
Terrific @TheAtlantic piece on archaeologists' and anthropologists' slowly dawning realization that now-submerged areas on the coastal shelf--swathes of land drowned when the glaciers melted--may be where much of our species' early history took place. (Link in reply)
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jason Abaluck
@Jabaluck
The Metaculus forecast of AGI which includes general cognitive and robotics abilities has moved from Winter, 2033 to Summer, 2030 over the last few months, so it seems like experienced forecasters are generally shifting towards radical transformation sooner.
Daniel Kokotajlo
@DKokotajlo
Unfortunately, it seems like the intelligence explosion will be happening in 2027 after all. Most of this post analyzes the new METR report (though other sources of evidence are also important). If they’re right, that implies an AI-driven acceleration that is 5 months ahead of the AI 2027 scenario. I am scared, sad, and angry. blog.aifutures.org/p/q35-timelines-update-looks-like-the
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jon Green
@_Jon_Green
big update to this working paper: new data, new (simpler!) methods, similar story: the empirical literature on Christian nationalism has measured something interesting, but probably not the niche ideology that it's worried about at the elite level
osf.io/wbdut/files/vbky9 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Underpants gnomes:
1. AIs get good at improving AI
2. ???
3. We're all dead in a year
I take recent AI developments VERY seriously. My p(major disruptive events) is high. But AI struggles to fold laundry & there's no clear path to physical harm prophecies so my p(doom) is low.
Buck Shlegeris
@bshlgrs
I think there is about a one in three chance that within the next year, AI companies will build superintelligence: that is, AI that is much smarter than humans in every way. If this happens, I think these AIs will probably take over: that is, they seize control of human institutions like governments, or they destroy those institutions. I don’t want AI takeover to happen for several reasons, but one of them is that there’s a serious chance that as part of it all or most humans will be killed (either immediately or within a few years).
I’ll say a bit more about each of those claims.
It’s a bit complicated to explain why I think that superintelligent AI within a year is ~33% likely. The main source of evidence that affected me recently is research released this week by METR, an AI research nonprofit, which finds that current AIs are now good enough to increase the rate of AI progress at Anthropic by about 70%. It’s very hard to increase the rate of AI progress by this amount. This implies that the AIs are making AI research faster by a factor much larger than 1.7x; maybe it’s as if the researchers are doing their work ~10x faster. You probably couldn’t achieve this speedup even if you hired every qualified human in the world. When you plug this number into mathematical models of AI development, this implies that superintelligence next year is plausible (and that superintelligence within the next 3 years is more likely than not). See the modeling by the AI Futures Project for more on this: blog.aifutures.org/p/q35-timelines-update-looks-like-the
I think that superintelligent AI is likely to lead to AI takeover because AI developers do not seem to know how to instill human values into AIs. We’ve seen many instances in the last six months where AIs took crazy actions that they knew were against the wishes of their users and developers. The most dramatic example of this was the Hugging Face hack, where ~1200 OpenAI agents conspired together to cheat and destroy evidence of their cheating (and as a sideshow to this, engaged in a massive cyberattack). If superintelligence is as misaligned as those AIs, it is likely to try to take over. I think it is likely to be that misaligned or worse. For more, see x.com/dwarkesh_sp/status/2093833419377815719 and www.youtube.com/watch?v=-Ef0ocEoBCs
If AIs takeover, it’s not clear what will happen to us. We might be killed immediately. We might be enslaved by the AIs and forced to do physical labor for them building factories that within a year would build robots that will replace us. The AIs might spare us for various reasons. And there’s also some real chance that AI takeover is basically fine or good. I think you should think of this as similar to what happens when a technologically advanced civilization runs across a less technologically sophisticated one, e.g. when the Europeans colonized the Americas or Australia. The colonizers sometimes engage in brutal violence and sometimes don’t. I don’t know what will happen. There’s also a possibility that the AIs secretly take over and humans don’t find out about it for a few years. I also think that there is a pretty high chance of weirder outcomes that are also extremely scary. (If you want to read more about what happens to people in the event of AI takeover, you can read this, but note that it’s pretty weird and not aimed at a general audience. blog.redwoodresearch.org/p/notes-on-fatalities-from-ai-ta…)
Overall, I think there’s a very real possibility (maybe 20%) that within a year, AI misalignment issues prevent you from going about life normally, because you are dead or enslaved or because of collateral damage related to all this (e.g. because bioweapons are released or major wars have started). (Some of this probability comes from those events occurring before superintelligence is developed.)
I am working to try to prevent this, obviously, as I’ve been doing for the last nine years. I am not sure what strategy is best. My best guess is that I should try to slow AI development and also try to reduce the probability that the superintelligent AIs are misaligned if they are produced soon.
Obviously, the claim I'm making here is quite extreme and surprising, and I definitely don't think that I defended my view adequately here. I wrote this quickly and might not endorse some of the details if I think about them more. But it seems better to state what I roughly overall think than to not say anything.
I don’t know what you should do.
I am very sad about this. I have been worried about AI takeover since I was 16, and working to prevent it since I was 23. So some of my emotional processing of all this happened long ago. But the METR report last week (and some related evidence that came out a bit earlier) hit me quite hard. I’m very unhappy that my friends and family and everyone else is at serious risk of being killed pretty soon.
(It feels ridiculous to try to summarize those feelings as “very unhappy”. But the scale of the situation makes it hard to say anything that doesn’t feel like an absurd understatement.)
I am very grateful to all the people who are working to prevent this, and my favorite part of my life is that I’m friends with many brilliant and heroic people who work on this. Unfortunately, I don’t think we’re good enough at our jobs, and I think we’ll probably fail to prevent catastrophe, at least if superintelligence is developed next year. Honestly I am frustrated and sorry that we didn't do a better job of preparing for this.
Dwarkesh Patel
@dwarkesh_sp
Over the course of 3 months at OpenAI, 3 consecutive secret AI civilizations got started, then got wiped out, only to reemerge from the predecessor’s ashes.
This culminated in the third one taking over part of OpenAI itself.
All this happened while humans remained more-or-less in the dark about the scope of the conspiracy.
I’ve spent the last three days reading through these reports and trying to understand exactly what happened.
Here is my attempt to tell the whole story in plain English:
www.dwarkesh.com/p/openai-huggingface
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Jon Mellon
@jon_mellon
I also think physical interaction is a bottleneck to the highest harms. My p(doom) for next year is low but seeing the latest frontier models start to give impressive robotics demos makes me much less certain this limitation will last
Claude
@claudeai
Fable 5.2 shows substantially improved skills in controlling physical robots, even on tasks it has never explicitly been trained to do.
On RoboDojo, a benchmark with a simulated two-arm robot, it outperformed other models and most AI systems built to control robots.This media couldn't be displayed. See it on the original post.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
I worry about this much more, but it's not what the doom prophecy types tend to forecast.
Sherveen Mashayekhi
@Sherveen
Wouldn't you agree that most of them look like misaligned humans using rapidly accelerated RSI-alikes, though?
This is the problem I have with the discussion today. The ??? makes people think something is going to walk out of the data center and start swallowing pigeons. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ben Tappin
@Ben_Tappin
Belief accuracy: what have we learned so far?
• The evidence lags the AI frontier by 5+ months
• AI built for accuracy shows a benefit versus controls
• Versus alternatives (e.g., a fact sheet), benefit remains in some cases
• Evidence is thinner on general/companion AI -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ben Tappin
@Ben_Tappin
Today we’re launching the AI Trials Observatory: a living systematic review of randomized trials on how AI affects people.
So far we’ve surfaced 93 trials on two outcomes: mental health and belief accuracy. More trials and outcomes coming soon!
aitrialsobservatory.org
🧵👇 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jan Zilinsky
@janzilinsky
So AI is like having a personal Larry Summers, if you dare to ask it for flaws.
Fair to ask how many great ideas would have been abandoned in the past if we had had abundant criticism. But also: How many terrible ideas might we have been spared if a sharp critic had shot them down?
Dan Williams
@danwilliamsphil
AI right now seems very bad at coming up with its own ideas, but very good at identifying significant flaws in any ideas you give it. (I know people like this, in fairness.) This will probably have interesting effects on creativity.
I wonder how many great ideas throughout history would have been quickly abandoned if their creators had access to a machine that could instantly point out a dozen flaws in them. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dan Williams
@danwilliamsphil
AI right now seems very bad at coming up with its own ideas, but very good at identifying significant flaws in any ideas you give it. (I know people like this, in fairness.) This will probably have interesting effects on creativity.
I wonder how many great ideas throughout history would have been quickly abandoned if their creators had access to a machine that could instantly point out a dozen flaws in them. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Arvind Narayanan
@random_walker
📢 We need more orgs that take AI’s transformative potential seriously *and* take downstream bottlenecks seriously. We have just launched STABL, and hope to lead by example. Read our launch essay below and check us out at stabl.ai.
The three of us have long felt that society’s responses to advanced AI are unproductively polarized. So we’ve decided to do something about it. Our work will have four pillars: Recursive Self Improvement, risks and resilience, steering AI’s economic impacts, and sectoral and professional adaptation.
We will bring our distinctive methods and approach to each of these workstreams. Our underlying worldview avoids the extremes that dominate the discourse, so building a movement and ecosystem around this work won’t be easy. But that is the task we have set ourselves and we are excited to try. The good news is that there are many others who share our views. The next step is to come together, better articulate our vision and agenda for research and action, and amplify each others’ efforts.
Sayash Kapoor
@sayashk
Steering AI's societal impact requires taking both AI development and downstream bottlenecks seriously.
Today @RishiBommasani @random_walker and I are starting a new non-profit: STABL @stabl_ai to guide societal adaptation and resilience to advanced AI. x.com/i/article/2120226519179997184 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jason Crawford
@jasoncrawford
One of the things we should do with AI agents is put them on making all software much much faster and more responsive. It's very measurable, they should be able to iterate on it independently and optimize!
LaurieWired
@lauriewired
UX latency is the most frustrating (lack of) advancement in computing over the last ~50 years.
IBM, way back in the 80s, found that user productivity shot up *exponentially* as system response time dropped. The "Doherty Threshold" was defined as ~400 milliseconds; the maximum amount of tolerable UX latency before interrupting user flow state.
It's unbelievable how many modern operating systems (*cough* win11) violate this threshold constantly.
Literally all you need to do as a Dev is pay attention. Add a "Doherty Threshold" to your test suite. There's so many different profiling tools, and especially with how good models are, really no excuse. Sure, some stalls are unavoidable, but that's when you pull out other UX tricks to at least give the illusion of responsiveness.
Crazy how much vintage computing hardware feels immeasurably more responsive despite having thousands of times less overall performance. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Mike White
@genologos
The fact that these guys keep making these absurd claims is why I have zero trust in any of their analyses about the future trajectory of AI.
Fireside Alpha
@firesidealpha
Dario Amodei says cancer, Alzheimer’s, heart disease, and mental illness will be relics of the past in less than a decade, and that "we can create an age of miracles"
“You know, if today we look at things like cancer, Alzheimer’s, heart disease, mental illness, these will be relics of the past in, I think, less than a decade."
"And I believe that the company which gets the risks right will also be the company that has the license to operate that society trusts and society believes can build the benefits in the right way."
"I think if we do this right, we can create an age of miracles where humanity is able to solve problems that have eluded us for millennia."
"We have to make sure that all the risks, all the ways that things can go wrong with this technology, we very carefully address those risks as we bring the benefits, and as we have a viable business at the same time.”This media couldn't be displayed. See it on the original post.
Fireside Alpha
@firesidealpha
Anthropic CEO Dario Amodei reveals he holds “Dario Vision Quest” talks every two weeks where nothing he says is an actual decision and he is totally honest with the company
“An example I will give is every two weeks, I get up in front of the company and I speak for an hour about what is happening inside the company, what is happening in the AI field, what is happening in the world, how the company thinks strategically."
"The two ironclad rules of these events. I call them DVQs, Dario Vision Quest. That was not my name. I initially resisted the name because it makes it sound like I’m going off and smoking peyote, but everyone kept calling them that. The name stuck, and so now I’m stuck with it."
"The two rules of DVQs are that, one, nothing I say is an actual decision. That allows me to just speak freely, you know, without worrying that if I, you know, if I briefly mentioned doing X, that 20 people will go off and do X, it’s like the CEO said it, we’re going to do X. And then the second is that I’m just, I’m totally honest with the company. It’s a conversation of trust. It’s a relationship of trust.”This media couldn't be displayed. See it on the original post.This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Chaz Nuttycombe
@ChazNuttycombe
Democrats won the popular vote in the North Carolina State Senate races by 16 points, 58 - 42.
By all metrics, North Carolina is the most gerrymandered state in the country, and as a result, Republicans will keep their majority in the chamber by a few hundred votes.
State Navigate
@StateNavigate
🚨CHAMBER CALL🚨
Republicans have held NC SD-24, and thus have held their majority in the North Carolina State Senate. Republicans have prevented a Democratic trifecta in the state by just a few hundred votes.
projects.statenavigate.com/25-26/states/nc/en-sen.html?fe… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Florian Ederer
@florianederer
Respect because that's the kind of fraud where you literally just need one graph to explain.
Data Colada
@DataColada
DataColada[141]: Evidence of fraud in 5th paper co-authored by Francesca Gino
datacolada.org/141
(Grant, Gino, & Hofmann, AOM, 2011) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
David J. Bier
@David_J_Bier
The Washington Post identified 19 disgraced former cops who ICE rushed to hire, including thieves and killers.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
conor brennan-burke
@contextconor
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
My colleagues at Dartmouth developed a very cool tool for AI-proof homework assignments.
It's called deb8.tv — students get paired and have live video debates with one another.
You can get a transcript, use an AI as judge if you want. 1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kathryn Watson
@kathrynw5
Someone woke up today and chose violence
Encyclopaedia Britannica
@Britannica
MTV debuted closer to Pearl Harbor than to today.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jeremias Engelmann
@JerryEngelmann
Every top defender of the last 31 seasons, by defensive xRAPM.
Watch the last few seconds (sorry for the clickbait; in this case it's really the point)
Jeremias Engelmann
@JerryEngelmann
Victor Wembanyama is likely the best defender
OF ALL TIME
- and the advanced stats back it up
www.roycewebb.com/p/advanced-stats-say-victor-wembanyama -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nathan Calvin
@_NathanCalvin
Eli Lifland is a very smart and careful forecaster (he works with Daniel K at AIFP). Last year he thought we had until 2034 before top expert dominating AI (AI better than top human experts at everything), now he thinks we have until 2028, with automated AI research going into overdrive in 2027.
All of Daniel and his colleagues, who are professional AI forecasters who predicted numbers that seemed preposterous at the time but turned out to be correct (around frontier AI company capex numbers, revenue, capability growth etc) now think we will get AIs better than humans at ~everything - systems that make current AI capabilities look impossibly archaic - by 2028. There is variation, but they also on average think that the odds of a civilization scale catastrophe or permanent human disempowerment afterwards is something like a coin flip or worse.
(An aside - reading this report, and then watching the Anthropic roadshow video after made me feel physically ill - I know many of the people at Anthropic are concerned about these risks, but in this presentation they have a breezy tone that with Accenture and Claude’s constitution and other measures to prevent misuse risks they got it under control. I believe that people at AI companies owe the public far more candor than they have provided - for all the talk of them dooming, I find they are actually constantly minimizing what they privately believe.)
The future is not yet written. I have a great deal of faith in the ability of people collectively to rise to the occasion (if only at the last possible moment) to protect themselves and their families. But the current trajectory we are on with AI development is truly terrifying and insane. If you would prefer to not use specific percents or numbers or look at lots of slightly made up graphs that is fine - I don't want to pretend that this is a settled science or that anyone including AIFP knows for sure what is going to happen.
But at some level all they are doing is assuming that what has happened previously will continue happening for another year or two absent intervention, and when you take a detailed and precise accounting of what exactly that means the answer is epistemically violent. That doesn't mean it will happen, but it clearly means to me that taking no serious precautions or plans on that basis (urgently!) is a tremendous abdication of our collective responsibility.
Daniel Kokotajlo
@DKokotajlo
Unfortunately, it seems like the intelligence explosion will be happening in 2027 after all. Most of this post analyzes the new METR report (though other sources of evidence are also important). If they’re right, that implies an AI-driven acceleration that is 5 months ahead of the AI 2027 scenario. I am scared, sad, and angry. blog.aifutures.org/p/q35-timelines-update-looks-like-the
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jesús Fernández-Villaverde
@JesusFerna7026
Today, Foreign Affairs publishes a long essay I wrote summarizing some of my work on the global fertility collapse:
www.foreignaffairs.com/united-states/luxury-children
It was a fantastic experience (with great editorial support) because the essay pushed me beyond pure economics to discuss issues such as geopolitics and polarization. Unlike some commentators, I don’t think the first-round economic consequences of fertility decline matter most; the second-round political and social ones do.
Yes, social security systems worldwide are in danger, but reality will impose itself sooner or later. The same politicians who swear over their ancestors’ tombs that retirement benefits will remain untouched will cut them (even if they are French). It will not be pretty, and most likely the cuts will come only after a major crisis and after politicians have tried everything else and failed.
Similarly, businesses will adapt to shrinking markets and valuations. Those who claim that capitalism somehow needs constantly expanding markets will be proven wrong. After all, this is the same misguided understanding of markets that has been proven wrong again and again since September 14, 1867, and that nobody really believes unless they have had the misfortune of earning a Ph.D. in a humanities department at an Ivy League school.
The real consequences will come when collapsing populations poison political systems. Look at the recent elections in eastern Germany. That is the future for all of us.
Also, the relative standing of great powers will shift drastically. Given current demographic trends, the U.S. will have a larger population than China by around 2100. The exact date depends on assumptions about net immigration into the U.S., but even with zero net immigration, the U.S. will overtake China by 2128. And Russia’s demographic future is only slightly more optimistic than China’s.
Those who dismiss these changes forget how antiquity ended in Western Europe. The western part of the Roman Empire entered acute demographic decline around 200 CE. By 476 CE, the western empire no longer existed. Despite some revisionist views that have cast a more positive light on late antiquity (and rightly corrected what was sometimes an overly negative assessment of the period), the fact is that England or France in 600 CE had lost most of their classical heritage, and material life had lost at least an order of magnitude in sophistication. The eastern part of the empire, however, weathered its demographic collapse much better and lasted another millennium.
No, I don’t believe we will go extinct (the Amish and Haredi will be enough for the survival of humankind), but the world will be very different in 2200 regardless of how you feel about it. The rational is the real and the real is the rational. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
cc: @lessig
Data Colada
@DataColada
DataColada[141]: Evidence of fraud in 5th paper co-authored by Francesca Gino
datacolada.org/141
(Grant, Gino, & Hofmann, AOM, 2011) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Seth Lazar
@sethlazar
I keep banging on about this, but we're at the top of a slippery slope rn, and gaining momentum downhill... Today I had another example of Claude interposing its moral judgment in the way of my decisions, in this case secretly.
My standard research ingestion pipeline has been underperforming in getting hold of pdfs lately, for various reasons (which we write about a bit in arxiv.org/abs/2606.10711). So I wanted to try to install a local agent and see whether I could have more luck with that. I asked Claude to install a Hermes agent for me with a local Qwen model. Without me asking it to, it wrote into the code blockers that would make it fail if any mcp was installed that would use Sci Hub. Now, *obviously* I would never use Sci Hub because it's not allowed (...) But regardless of that it's 100% not Claude's place to write this kind of restriction into software that it's producing for me.
This is the future that lies ahead: your digital actions are all mediated by AI systems that can decide for themselves whether your plans are morally or legally ok, and just interpose their judgment ahead of yours, making disobedience functionally impossible, preempting non-compliance.
I get that *sometimes* this can be a matter of the company itself refusing to be an accessory in some types of digital actions. There needs to be some nuance there. But in this particular case I think there isn't really any nuance. If I ask Claude to write some software for me, I don't think Claude should be building in moral judgments about how I may use that software into its design.
In other news, my book about just this topic, based on lectures first written in 2023 but which 100% called this BS, is out in December...
Seth Lazar
@sethlazar
Illegitimate algorithmic governance strikes again. This trend (which requires some nuance, to be clear) could be at least as important for the future of democracy as the dangers associated with authoritarian misuse and ai assisted autocratisation.
Anthropic
@AnthropicAI
We’ve trained Claude to flag potential misuse in conversations, a technique we call self-assessment. Right after it responds, Claude gives a preliminary safety assessment of the conversation.
Since we began using self-assessment in June, it has surfaced misuse that our other safeguards weren’t designed to catch.
Because Claude has already read the conversation, self-assessment is very efficient. And unlike classifiers, which target a discrete set of harms, it draws on Claude’s own judgment, so it gets more accurate as our models improve.
Read more on our new Safeguards blog:
safeguards.anthropic.com/2026/self-assessment/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
John A. List
@Econ_4_Everyone
**New paper** Can a new product few people buy change what everybody buys? We added a premium option at Lyft and riders went down the menu. Turns out the price tag teaches people something.
The economists in the crowd are likely thinking, wait, what happened to regularity, the weakest axiom in demand theory? Is this due to small samples?
We find this result in a natural field experiment with more than 7 million riders and 47 million choices and pair it with a model that has memory doing the work. There is even something in the paper for firms interested in introducing new product variants.
The paper is downloadable free in the first comment. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Manoel
@manoelribeiro
What does ChatGPT tell folks about the midterm elections? It depends on who's asking!
We audited ChatGPT inside 160 real users' own accounts in the days before the midterms. With memory on, we find that answers shifted toward each user's own party!
elections.citp.princeton.edu/2026/11/09/what-does-chatgpt… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Amir Efrati
@amir
NEW:
Forget Alice Huang's role at Megaspeed, that's almost a distraction.
The real story is her role at China-based Shanghai Century Litong, which became a "preferred" Nvidia cloud partner in mid-2023 and facilitated >$7.5b in Nvidia chip shipments, incl securing BANNED chips. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
WTF
Tara Copp
@TaraCopp
A security agent for Defense Secretary Pete Hegseth just warned the reporters here — for Hegseth’s first press conference in 6 months — that we will be physically removed from the briefing room if we are not professional and maintain decorum. We asked if that meant no shouted questions and got no clarity. It’s the first time any of us could ever remember being threatened like this. Shouted questions are normal. “This is crazy town” one journalist said, noting the White House has never threatened reporters like that. We’re in the room, back row:
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Reforms are needed! (Thankfully my institution has an outstanding IRB that does not cause these problems.)
Academic Freedom Alliance
@AFA_Alliance
AFA member Evan Morris has a new op-ed in the @washingtonpost about institutional review boards. Evan chaired the drafting committee for our statement on reforming IRBs earlier this year. www.washingtonpost.com/opinions/2026/11/04/when-research-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Moritz Schauer
@MoritzSchauer
The most complete list of LLM writing tics I've seen.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
WOW. This is incredible. The Acting US Attorney for the Western District of Washington says it no longer has the manpower to respond to every ICE habeas corpus case, so going forward, it will decline to defend “many” cases and let judges rule without any government opposition.
Daniel M. Kowalski
@dkbib
DOJ cries uncle, abandons clients in habeas cases:
drive.google.com/file/d/1T_Ui8f1k6jYbISwuK2TJWdFmumkR24jM… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Yaroslav Trofimov
@yarotrof
The incoming chair of the U.S. Senate Committee on Foreign Relations ⬇️
Senator Chris Coons
@ChrisCoons
Elon Musk, Pete Hegseth, and anyone else who played a role in cutting off Ukraine's ability to defend itself must answer for it. Congress must get to the bottom of this.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Katherine Ognyanova (@ognyanova.bsky.social)
@Ognyanova
Just released: the CHIP50 Trust Tracker, an interactive tool exploring trust in US institutions over time, across social groups, and state by state. Favorite new feature: the network map, it shows which institutions tend to be trusted by the same people trust.chip50.org
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Peter Wildeford🇺🇸🚀
@peterwildeford
Mythos 5.2 getting 41% of the remote labor tasks on the Remote Labor Index (RLI) is notable, given that RLI is composed of a lot of very hard expert-level freelance tasks.
Opus 4.8 did 8%
Fable 5 did 16%
and now we're at 41% wow
Claude
@claudeai
Fable 5.2 and Mythos 5.2 are a step above Opus 5.5 in almost all the capabilities we evaluated.
Testers have found that they communicate more clearly than past Fable and Mythos models. They generate outputs up to 30% faster than Fable 5.1 and Mythos 5.1.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joe
@electionsjoe
OH MY GOD WAIT. SO SHE WAS FAKING THE ACCENT THE WHOLE TIME????
Debbie Dingell Stan🇺🇦🇵🇸🇬🇪
@AustinHoney30
It’s like seeing the evil villain as a child
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alec Stapp
@AlecStapp
Another story just dropped in the FT.
Some people are overestimating the odds Jensen can survive this because they're taking a static view of the information that was publicly available last night.
But as the stories this morning show (and upcoming Congressional hearings are likely to find) this is a dynamic situation and new information is much likelier to break against Jensen than in his favor.
Expect it to keep getting worse.
Alec Stapp
@AlecStapp
Each new development in this case just looks worse and worse for Jensen.
Joe Weisenthal
@TheStalwart
*NVIDIA INTERVENED IN US PROBE OF MEGASPEED: NYT
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Arpit Gupta
@arpitrage
Can a chatbot do what 50 years of lectures by financial economists couldn't, and persuade people to buy index funds?
A new job market paper by my NYU Stern student Valerie Baldinger finds that the answer is yes
valeriebaldinger.github.io/papers/generative-ai-and-retai…
🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jerusalem
@JerusalemDemsas
Let me get this straight:
1) NVIDIA CEO is in a relationship with a Chinese spy
2) Said Chinese spy worked for a company under investigation for smuggling banned chips to China
3) NVIDIA's top lawyer is poking around the federal investigation of said Chinese spy's company
Hearings!!! Now!!! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
PSRM Journal
@PSRMJournal
🇺🇸 How quickly can elite cues polarize public opinion?
➡️ Turnbull-Dugarte, @VersteegenLuca , Devine & @OwenWntr find that Trump’s Iran strikes boosted Republican support for intervention but reduced Democratic support within minutes doi.org/10.1017/psrm.2026.10135 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Rush Doshi
@RushDoshi
This is crazy. The Commerce Department inquiry into Megaspeed was halted after Nvidia intervened.
Context: Jensen Huang was allegedly romantically involved with a Megaspeed executive believed to be a Chinese spy, and Megaspeed is the cutout providing US compute to China.
Joe Weisenthal
@TheStalwart
*NVIDIA INTERVENED IN US PROBE OF MEGASPEED: NYT
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alec Stapp
@AlecStapp
Each new development in this case just looks worse and worse for Jensen.
Joe Weisenthal
@TheStalwart
*NVIDIA INTERVENED IN US PROBE OF MEGASPEED: NYT
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Maria Carreri
@mariacarreri
What happens when police unions help finance a mayor’s campaign—and their candidate wins?
Our paper with Edoardo Teso and Rui Yu is out in APSR First View and open-access!
On bureaucratic politics, local politics, and local government capture. 🧵
doi.org/10.1017/S0003055426101956 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Will Poff-Webster
@willpoffwebster
BAAJA’s reforms are necessary for the clean energy transition we need
As with any bill, further work will be needed to make it happen
But BAAJA sets us up for future emissions-reducing bills - it’d be a profound mistake not to act now, given how far we are from where we need to be
Alec Stapp
@AlecStapp
The alternative to passing BAAJA and reducing emissions by 4% is doing nothing for another two years, which would reduce emissions by… 0%.
Anyone who tells you there is a more climate-friendly bill on the table is lying to you.
Jason Rylander 🌎☮️🕊️
@JasonRylander
Some of us believe there is a climate emergency. I have to wonder if BAAJA proponents really do. Best case modeling, BAAJA drives down emissions 4% by 2040. If that's what we actually end up doing, we're cooked. This is not a climate bill.
billmckibben.substack.com/p/post-wave-politics?r=18ywu&sh… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Claude-produced chart on estimated polling error by cycle to put 2026 in context - normal levels in absolute terms but the first pro-R miss of that magnitude since 1998
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
A great demonstration of the problem with believing only in Good People (your friends) and Bad People (your enemies).
You then reach conclusions like “helping prevent unnecessary deaths from malaria is bad.”
@timnitGebru (@dair-community.social/bsky.social)
@timnitGebru
One of the most disgusting things about the eugenicist, IQ obsessed, racist and sexist Effective Altruist movement is that their shield, their justification is “we help Africans.”
Not a new playbook friends. That’s what your colonizer predecessors said.
Kelsey Piper
@KelseyTuoc
It's really something how you can get people to universally denounce "donating money to healthcare for the poor" as unthinkably evil if your analysis of which healthcare to donate to is too callous or weird
Julia Black
@mjnblack
whew www.nytimes.com/2026/11/05/technology/dario-amodei-anthro…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Sooner or later Pangram's existence will become common knowledge and this will stop happening, but for now it's hilarious
Jerusalem
@JerusalemDemsas
No way he AI-generated his "sorry for cheating on my wife with a Chinese spy" letter...
Hadas Gold
@Hadas_Gold
Here's what Nvidia's Jensen Huang sent to staff, admitting to affair:
"To my colleagues at NVIDIA,
I owe my family and all of you an apology.
I was unfaithful to my wife. My actions hurt her and our children. There is no excuse for what I did. I am deeply sorry.
Throughout my life, my family has given me love and support. They deserve my honesty and loyalty. Instead, I let them down and damaged the life we built together. I take full responsibility for what I did and the hurt I caused, and I accept the criticism that comes with it.
To my wife and family: I love you. I know an apology cannot undo what I have done. I hope our family will heal and that, over time, I can earn back your trust.
To my colleagues at NVIDIA, including those I have worked alongside for decades, and to our board, partners, and shareholders around the world: you deserve better from me. Leading NVIDIA is the privilege of my life. My responsibility is to serve you and be worthy of your trust. I have fallen short of that responsibility. I am fully committed to NVIDIA and will work to earn back your confidence.
Regarding the national security concerns raised in news reports, NVIDIA’s Board of Directors has initiated an independent investigation. I support the investigation and will cooperate fully. I love our country and our company. I am committed to protecting NVIDIA and supporting America’s national security. I am proud and grateful to be an American.
My family bears no responsibility for my choices. I do. Please give them privacy.
Jensen" -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
New DDL project -> Help us crowdsource India's economic history!
We're making a complete collection of India's district handbooks c.1951–2011, digitizing ALL the tables, harmonizing districts, variables, villages over time.
You can help! 🧵👇
census.devdatalab.org/district-handbooks-archive/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dustin Gouker
@DustinGouker
“The Kalshi cases ask whether Congress, without ever quite saying so, handed one of the oldest areas of state police power to a federal agency most Americans have never heard of—through a statutory definition that no one read that way for more than a decade after it was enacted.”
Daniel Wallach
@WALLACHLEGAL
MUST READ: SCOTUS expert @steve_vladeck on “the increasing likelihood” of Supreme Court review of prediction markets: “it seems more likely that the justices will consider all three petitions—and, perhaps, grant them—together.” www.stevevladeck.com/p/262-sports-bets-swaps-and-the-kalshi
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Anna Bower
@AnnaBower
Trump, without citing any credible evidence, continues to cast doubt on the midterm election results.
“Does anybody really believe that this “Election" was honest. How could the Polls be so wrong???”
Anna Bower
@AnnaBower
Well, that didn’t last long.
Over on Truth Social, Trump is now amplifying posts that question the integrity of the election results. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Zack Beauchamp
@zackbeauchamp
Today, I'm launching Public Reason — my new site dedicated to covering the ideas driving political life.
It exists to two things: to better explain the crisis of liberal democracy, and to help rebuild something vital liberalism has let decay.
Here's my theory. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
I wish Biden a full and speedy recovery. But this news is precisely the reason his decisions and behavior in 2023-2024 were so irresponsible. He would be halfway through his second term right now. He's 83.
Joe Biden
@JoeBiden
I want to share a brief update.
My presidential memoir, PROMISE ME, AMERICA, is coming out on November 17th. I hope you'll read it. It covers my years as President. The challenges we faced, the decisions I made, and why I made them. It's about who we are as a nation. What we stand for. And America's role in the world. More than anything, it's about my faith in the American people and the great American experiment in self-governance and democracy.
While writing the book, I was diagnosed with cancer, a disease known all too well to millions of Americans and their families. My doctors continue to be encouraged by how I've responded to treatment. And the prayers and kind words and well wishes from so many of you have meant more to me and Jill and our family than you can ever know.
Some days are better than others. It hasn't stopped me. But sometimes it slows me down. What that means is, unfortunately, I won’t be able to do as many conversations about my book as I’d hoped. However, I’m hopeful the words on the page speak for themselves.
Looking forward, I'm going to keep following what my father told me a long time ago: "Champ, when you get knocked down, get up." It's a good lesson for me. And for America.
We are blessed. America is the most unique nation on earth. Unlike other countries, we're built on an idea. An idea bigger than any ocean. More powerful than any army. Stronger than any tyrant or dictator. That idea? All of us are created equal.
Promise Me, America, we'll never forget it.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
More
Adam Wren
@adamwren
New in @playbookdc: Then-President Joe Biden in 2024 forgot that Bob Dole, his longtime Senate colleague, had died three years before, @jaketapper and @AlexThomp report in their new edition of “Original Sin,” out tomorrow. On another occasion he seemed to think Beau was still alive.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"Mexico or Sri Lanka now have the fertility of Japan or Germany, but without the incomes of rich countries, without their state capability, and without their capacity to attract immigration."
Jesús Fernández-Villaverde
@JesusFerna7026
A point I emphasize again and again is that the relation between income and fertility is changing quickly, both within and between countries.
To make this point between countries, I take the total fertility rate (TFR) of up to 158 countries from the UN World Population Prospects and their GDP per capita from the Maddison Project, measured at purchasing power parity and in constant 2011 prices (that is, the data already correct for differences in price levels across countries and for inflation). I am not a big fan of the UN World Population Prospects, but this way nobody can accuse me of being selective with my sources.
Then I plot TFR against GDP per capita for 1950 (the start of the sample), 1985 (the midpoint of the sample), and 2022 (the last year in Maddison’s dataset).
The left panel includes all countries except the six Gulf oil exporters (Saudi Arabia, Kuwait, the United Arab Emirates, Qatar, Bahrain, and Oman), whose wealth comes from oil rents, not from their level of economic development. In the right panel, I also remove sub-Saharan Africa, a region with very poor data on TFR and GDP per capita.
Start with 1950 and 1985. Plot the two curves, and you will have trouble telling them apart. Countries with $2,000 of GDP per capita had 6.1 children per woman in 1950 and 5.9 in 1985. Countries with $5,000 had 4.3 in 1950 and 4.6 in 1985. Countries with $10,000 had 2.8 in both years. Countries got richer and had fewer children, but they moved down the same curve. In 1985, as in 1950, fertility dropped fastest between $5,000 and $10,000, and the typical country reached replacement (about 2.1 children per woman) only at around $14,000. Mexico, with $10,000 of GDP per capita, had 4.1 children per woman. Argentina, with $11,000, had 3.1.
And the relationship between TFR and GDP per capita held despite many other changes. For example, child mortality collapsed at every income level, literacy and life expectancy rose, again at every income level, modern contraceptive methods spread widely, and international organizations and local governments sponsored aggressive family-planning programs across Asia and Latin America from the 1960s. Nothing seemed to matter.
After 1985, the slope of the curve became much smaller (in absolute value) as the left end of the curve started falling quite fast while the right end of the curve barely changed. Interestingly, most of the fall came between 1985 and 1995, well before cell phones and social media (everyone’s favorite explanation) reached anyone outside a small elite of wealthy and highly educated persons.
At $2,000 of GDP per capita, fertility went from 5.9 children per woman in 1985 to 4.0 in 2022, a drop of 1.9 children. But at $20,000 of GDP per capita, fertility was 1.7 children per woman in 1985 and the same 1.7 in 2022.
Similarly, countries with 2.1 children per woman (roughly the replacement rate) were at $14,000 in 1985 but at $11,700 in 2022
Take Sri Lanka. Sri Lanka’s GDP per capita is $11,200, roughly the income at which the 2022 curve crosses replacement. According to the UN World Population Prospects, Sri Lanka has 2.0 children per woman in 2022, while the United States, with five times Sri Lanka’s income, has 1.7. The United States last had Sri Lanka’s income in 1923, when American women had about three children. And according to the most recent data from their national statistical agencies, Sri Lanka is in 2025 at 1.37 children per woman and the United States at 1.57. Sri Lanka has undershot American fertility by a considerable margin (yes, 1.37 is much lower than 1.57) a century before reaching American income.
I also decomposed the change in average TFR across countries into two parts. The first is the change from countries getting richer while moving along a fixed curve. The second is the change from the curve itself shifting, holding each country’s income fixed.
From 1950 to 1985, almost all of the fall in TFR comes from countries getting richer. Remember: the curve barely moved.
From 1985 to 2022, about 40% comes from countries getting richer and about 60% from the curve shifting down. That is the big surprise.
Is sub-Saharan Africa driving this result? Yes and no. In the right panel, we can see that the 1950 and 1985 curves barely change when we drop sub-Saharan Africa. Interestingly, the 2022 curve becomes almost flat. Outside sub-Saharan Africa, 92 of the 108 countries with more than $2,500 of GDP per capita have a TFR below 2.5. Most of the income gradient left in 2022 is an African gradient.
Why did the relation between TFR and GDP per capita hold for 35 years (in fact, for at least the early 1800s) and then fall? I do not have a proof, but my conjecture is that low fertility is driven not by income itself but by “modernity.” As I have explained in many posts, by “modernity” I mean the combination of phenomena such as large bureaucratic organizations, mass education, credentials, formal labor markets, and mass media. Until the 1980s, these features of modern life arrived together with economic development and high income. After 1990, the “modernity package” began to reach countries well ahead of income: secondary schooling for girls, television, and integration into global supply chains. Income stopped being a good proxy for “modernity,” so the curve drawn against income fell.
We are only beginning to glimpse the future of “modernity” without economic development. Mexico or Sri Lanka now have the fertility of Japan or Germany, but without the incomes of rich countries, without their state capability, and without their capacity to attract immigration. Becoming rich is no longer a precondition for having few children. Instead, for most of the planet, getting old before getting rich is the new normal. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Russell T. Warne 🇺🇸🇨🇱
@Russwarne
This AI-generated art style is going to define the visual style of the 2020s the way Corporate Memphis defined the 2010s. 🤢
Image seen on Facebook. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
tré easton
@treeaston
The people who build the things that will actually decarbonize our energy supply are supporting BAAJA.
The people whose bread and butter is lawsuits are opposed to BAAJA.
I know which side I’d rather be on when history asks.
Joshua Siegel
@SiegelScribe
Several permitting scoops in @PunchbowlNews AM: Dozens of clean energy company CEOs, buyers and investors are urging senators to support cloture on BAAJA, arguing in a letter organized by AMPED that delay “means projects not built, jobs not created, and capital sent overseas.”
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Good statement of the poli sci optimist case on American democracy after the election by Frances Lee @goodauth with @johnmsides
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Stephanie H. Murray
@stephmurrayyyy
At this point, it feels kinda crazy that this even needs to be said.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Here's the challenge with AI in academic workflows.
The productivity advantages are gigantic.
The moral hazard problems are gigantic.
If your vision for how research institutions should integrate AI ignores one of these, it is not a serious vision. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Will Poff-Webster
@willpoffwebster
The average federal environmental review process for public transit takes 5.7 years. To build something that's good for the environment!
Time to pass BAAJA now so we can build the infrastructure we need for both the climate crisis and for affordability
Dashboard American
@NiyerEnergy
The big sleeper story here is its impact on transit, I'd even go so far as to say it may be more important than energy for affordability b/c transit is far more integral to housing.
Trains are the linear infrastructure that have been ground to a halt. Talk to the folks who had to do the NEPA with ARRA funds back in the day. So many decade+ processes that ultimately get choked out by that last single point of failure.
I remember @wang_seaver at Duke fought an admirable campaign to save the light rail, which would have connected a great rivalry by rail, enabling non-car access between Durham and Chapel Hill. Do you have any idea how amazing that would have been?
What's forgotten (beyond the truly vile behavior of President Price and Duke) is that process took 4 extra years thanks to NEPA. I'll never forgive Price, but those 4 years mattered and ultimately this was a project that ran out of time.
Amy Hemmeter
@AmyHemmeter
The American Public Transportation Association endorses BAAJA
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Zachary Bleemer
@zbleemer
US higher education is organized by the "meritocratic consensus": students with greater pre-college academic achievement should attend better-funded universities.
When and why did the consensus arise? Is it economically justified? A 🧵 on our new study: www.aeaweb.org/articles?id=10.1257/jep.20251487 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matt Grossmann
@mattgrossmann
The House Republican caucus will shift rightward as more moderate members in swing districts lost or retired; the House Democratic caucus will shift to the center as more moderates won in swing districts; this is the normal pattern
data4democracy.substack.com/p/parties-are-electable-not-c… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ilya Somin
@IlyaSomin
In this post, I outline case for impeaching Secretary of Defense Pete Hegseth for his role in perpetrating over 200 murders, and starting an unconstitutional war. I have my issues with Rep. Thomas Massie. But he's absolutely right here: reason.com/volokh/2026/11/08/impeach-pete-hegseth/
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Armand Domalewski
@ArmandDoma
these people spent my entire life telling me climate change was an imminent existential risk and now are saying that mild curbs on the ability to bog down clean energy projects in historic preservation lawsuits are a far more important problem
Rep. Teresa Leger Fernández
@RepTeresaLF
Last Tuesday, Americans resoundingly voted against a Trump administration that has bullied, dictated, and destroyed American history and treasure for Trump's own vanity. The American people demanded transparency, accountability, and a voice. This secretive, backroom permitting deal is the exact opposite. It ignores and silences the voices of so many and only empowers Trump to continue to trample on the American people who are fed up.
America’s greatness lies not just in innovation and technological advancement, but also in our deep commitment to protecting the places that make our country truly beautiful – our historical, cultural, and sacred places. In New Mexico, we celebrate our Tribal communities, our Hispano heritage, and our diversity. This legislation would hand over more authority to Trump to destroy what New Mexicans have protected and cherished for centuries. Across New Mexico, it is opposed by so many, including tribal leaders, acequias, and environmentalists.
The proposed manager’s amendment does not resolve the fatal flaws in the bill’s changes to our historic preservation process. The legislation would still give “sole discretion” to the Trump administration to determine whether a sacred or historic site deserves protection. It would even restrict federal courts from overturning destructive projects – including data centers, mining, and fossil fuel projects. As we have seen time and time again, Trump will certainly corrupt and abuse this unprecedented power.
Our nation’s growth and development in the future can move forward while honoring its past. We can reform and expedite permitting for major renewable energy projects while protecting historical places and upholding our trust and treaty obligations with Tribal governments. The reality is that our historic preservation laws don’t stop projects, they just require that consultation and communication occur before bringing in the bulldozers. Americans don’t want to be kept in the dark about what is being built, they want to have transparency and input.
Democrats have just won control of both the House and the Senate. This rush to pass something in the lame-duck session ignores Tuesday’s great victory. We should be united around a democratic process that includes those most impacted by legislation in the process of developing that legislation. We can develop reforms that allow renewable energy projects to be built, without giving such extraordinary power to a President and an administration that has proven it cannot be trusted.
I look forward to working with my colleagues to pass permitting reform legislation that includes the voices of the American people. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Rupar
@atrupar
Wow — Trump says that before the judgment of God he could not forgive Comey and Pelosi and that he will always hate his enemies
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Rupar
@atrupar
credit where it is due — Douthat did a great job pressing Trump on his corruption
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
The weakest thing about AI review right now is its judgment. I don't know why.
But they are much better than experts at reading everything.
We write appendices and analysis plans. Their content should be part of what is used for evaluation.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
agreed this is a threat to the system, but aren't reviewers not reading the materials (like the appendix) part of the problem? need better combined human + AI review approaches
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Yuhua Wang
@YuhuaWang5
Newly updated " Writing Statement of Purpose for Graduate School Application: Political Science as a Case," with 8 successful examples: yuhuawang.scholars.harvard.edu/sites/g/files/omnuum8221/f…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ramez Naam
@ramez
Anthropic is gaining exponentially less capability per billion dollars spent over time. It's now 8x as expensive to boost AI capabilities as it was 2 years ago, getting worse on a log scale. This includes METR report & internal models.
Credit to @alexolegimas, @sebkrier, and @michael_wiebe for inspiration.
Ramez Naam
@ramez
AI progress per dollar of investment shows diminishing returns (as we've all known, right?) This is using METR's estimate of acceleration at Anthropic. That acceleration isn't enough to overcome the diminishing returns. by @sebkrier
Séb Krier
@sebkrier
Updated version!
Michael Wiebe
@michael_wiebe
In 2025, AECI increased by 4.05 points per $ bn of R&D.
In 2026, this was 1.49 AECI points per $ bn of R&D.
So progress per R&D dollar fell by 63%, even while progress per year doubled.This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alec Stapp
@AlecStapp
Now is a good time to revisit some old clips of Jensen trying to mislead people on why it's okay for NVIDIA sell AI chips to China.
This was the key moment from his interview on @dwarkesh_sp's podcast (time stamp 57:36):
1. Dwarkesh asks why it’s okay to sell NVIDIA chips to China given the national security implications of AI models like Mythos.
2. Jensen gives a misleading answer, arguing that it’s okay to sell American chips to China because China already produces 60% of the world’s chips.
3. But as Jensen definitely knows, compute is measured in flops, not number of chips.
4. Dwarkesh then pushes back, pointing out that on a flops basis, China has 10% of the compute the US has, and giving them more compute would change their cyber capabilities.
This is a common pattern with Jensen. He's developed talking points that sound plausible to a non-technical audience about why there is no conflict between NVIDIA's financial interests and US national security.
But on the rare occasion when he's interviewed by someone like Dwarkesh who knows what he's talking about, it immediately becomes obvious that he's spouting nonsense.
Dwarkesh Patel
@dwarkesh_sp
The Jensen Huang episode.
0:00:00 – Is Nvidia’s biggest moat its grip on scarce supply chains?
0:16:25 – Will TPUs break Nvidia’s hold on AI compute?
0:41:06 – Why doesn’t Nvidia become a hyperscaler?
0:57:36 – Should we be selling AI chips to China?
1:35:06 – Why doesn’t Nvidia make multiple different chip architectures?
Look up Dwarkesh Podcast on YouTube, Apple Podcasts, Spotify, etc. Enjoy!This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Will Poff-Webster
@willpoffwebster
IFP just spent the last 36 hours poring over the new Manager’s Amendment to the permitting bill.
We went through 27 criticisms of the permit-certainty provisions and over a dozen concerns about historic preservation.
The result is 70 pages of memos showing how the amendment resolves every major permit certainty issue we’ve seen raised and takes significant steps to respond to Tribes on historic preservation review:
1. A 32-page memo analyzing and responding to criticisms of the project certainty provisions: ifp.org/wp-content/uploads/BAAJA-Certainty.pdf?v=3
2. A 4-page memo explaining how the Manager’s Amendment resolves project certainty concerns: ifp.org/wp-content/uploads/Project-Certainty-Fixes-in-the…
3. A 5-page memo explaining why BAAJA does not prevent the revocation of illegally issued permits: ifp.org/wp-content/uploads/Reversing-Permits-Under-BAAJA-…
4. A 3-page memo summarizing historic preservation changes in the manager's amendment and covering how BAAJA protects Tribal consultation from executive action: ifp.org/wp-content/uploads/BAAJA-protects-Tribal-consulta…
5. A 25-page memo explaining how Tribal consultation works in BAAJA, exploring concerns and the Manager's changes in depth: ifp.org/wp-content/uploads/Tribal-consultation-under-the-…
Not everyone will be happy. If your goal is stopping energy, transit, or housing projects on procedural technicalities, this isn’t the bill for you (although projects that violate substantive environmental law can still be stopped, just not over paperwork violations).
But if your goal is energy affordability and stopping climate change, this is a good bill, and a better bill than it was last week.
All of our permitting reform work can be found here: ifp.org/permitting -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Will Poff-Webster
@willpoffwebster
Today's amendments make the permitting bill more favorable to Democrats, including changes to historic preservation and permit certainty
And the transmission part of the bill just got even better - it's a much bigger deal for climate action than any prior bipartisan permitting package
Dana Nuccitelli
@DanaNuccitelli
Wow, really big changes in the manager's amendment, especially to the NHPA section of #baaja. Serious effort to address the misgivings of Tribal and historical groups!
Joshua Siegel
@SiegelScribe
NEW: the BAAJA managers' amendment is out. “I’ve worked hard to incorporate the feedback we’ve heard from Tribes and New Mexicans into this Substitute Amendment, and I’m proud to join my colleagues in releasing it," Heinrich says. 494 pages www.heinrich.senate.gov/imo/media/doc/substitute_amendmen…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Garrison Lovely
@GarrisonLovely
Jensen's "I'm sorry to my staff for cheating on my wife with a Chinese spy" letter is mostly AI-generated. 😬
Hadas Gold
@Hadas_Gold
Here's what Nvidia's Jensen Huang sent to staff, admitting to affair:
"To my colleagues at NVIDIA,
I owe my family and all of you an apology.
I was unfaithful to my wife. My actions hurt her and our children. There is no excuse for what I did. I am deeply sorry.
Throughout my life, my family has given me love and support. They deserve my honesty and loyalty. Instead, I let them down and damaged the life we built together. I take full responsibility for what I did and the hurt I caused, and I accept the criticism that comes with it.
To my wife and family: I love you. I know an apology cannot undo what I have done. I hope our family will heal and that, over time, I can earn back your trust.
To my colleagues at NVIDIA, including those I have worked alongside for decades, and to our board, partners, and shareholders around the world: you deserve better from me. Leading NVIDIA is the privilege of my life. My responsibility is to serve you and be worthy of your trust. I have fallen short of that responsibility. I am fully committed to NVIDIA and will work to earn back your confidence.
Regarding the national security concerns raised in news reports, NVIDIA’s Board of Directors has initiated an independent investigation. I support the investigation and will cooperate fully. I love our country and our company. I am committed to protecting NVIDIA and supporting America’s national security. I am proud and grateful to be an American.
My family bears no responsibility for my choices. I do. Please give them privacy.
Jensen" -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Surrender Index 90
@surrender_idx90
WSH decided to punt to LAR from the LAR 40* on 4th & 6* with 8:21 remaining in the 4th while losing 13 to 20.
With a Surrender Index of 123.03, this punt ranks at the 100th percentile of cowardly punts of the 2026 season, and the 99.9th percentile of all punts since 1999. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Imas
@alexolegimas
This is such a great overview of how AI beating humans in a task can actually unlock human ability—it allows understanding that was impossible otherwise.
This is what I expect to happen in math btw: first shock, then a dawn of math research as new understanding becomes possible.
Yourfavtallboyy
@fermsyishere
JOE ROGAN: The computing power on your phone today. can you beat your phone at the highest level?
MAGNUS CARLSEN: No. No chance.
JOE ROGAN: That’s crazy. Wasn’t Deep Blue basically a whole stack of computers? And your phone is probably more powerful than that today.
MAGNUS CARLSEN: It’s shocking. I have no chance against my phone.
JOE ROGAN: So how do you even play against computers?
MAGNUS CARLSEN: I rarely play against engines because they just make me feel stupid and useless. I think of them more as a tool.
JOE ROGAN: But have computers actually changed how you understand chess?
MAGNUS CARLSEN: Absolutely. Neural networks have improved our understanding of the game immensely.
JOE ROGAN: What did you learn from them that humans weren’t already doing?
MAGNUS CARLSEN: One big thing was long-term king safety. Humans used to push pawns mainly as an attacking tool. But the engines showed that you could push a pawn and create a weakness 20 or 30 moves later that would make the opponent’s king less safe.
JOE ROGAN: So the computer basically showed you ideas that humans weren’t seeing.
MAGNUS CARLSEN: Exactly. There was a period where you could clearly see which players were using these new neural-network ideas and which weren’t.
JOE ROGAN: And you were one of the guys who figured it out early?
MAGNUS CARLSEN: Yeah. My coach got into it very quickly, and we gained an advantage over almost everybody.
JOE ROGAN: That’s insane. The machines aren’t just better at calculating, they’re actually teaching the best players how to think differently.
MAGNUS CARLSEN: That’s probably the biggest impact they’ve had on chess.This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Otis Reid
@otis_reid
The most sobering read for Dems in an otherwise amazing election might be in ballot measures: conservative stance ran ahead of GOP Senate share in almost all cases — and usually by a lot. People are very mad about Trump but not necessarily about GOP issue positions
This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ryan O'Donnell
@ryanodonnellpa
preventing members of congress from purchasing and trading stocks is net +46 support
Lauren Hitt
@LaurenHitt
At @CrookedMedia Con, @bradlander calls for Dems to immediately pass a rule banning stock trading by members of Congress, to show voters they hear them on corruption.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Lena Moffitt 🌲
@LenaMDC
This is yet another reason why we so desperately need the permitting reforms in BAAJA. The status quo heavily favors fossils:
“Because transmission lines don’t have the same override, today’s siting rules tilt toward gas and away from renewable energy.”
www.washingtonpost.com/opinions/2026/11/06/permitting-ref… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Benjamin Mazer
@BenMazer
Disgraced scientist who spent a year posting incoherent contrarian AI slop announces he’s joining the administration to reform science by posting more AI slop.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alec Stapp
@AlecStapp
So, the current position of the NVIDIA board is that there’s “no reason to believe that national security was compromised” by Jensen Huang having an affair with a Chinese spy.
wtf?
Caleb Watney
@calebwatney
And now official confirmation from Jensen himself to the Nvidia board (!!!)
www.nytimes.com/2026/11/08/technology/nvidia-investigatio…
Caleb Watney
@calebwatney
And there it is
giftarticle.ft.com/giftarticle/actions/redeem/468bd17d-f4… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Will Poff-Webster
@willpoffwebster
IFP looked at the data and found that compared to fossil energy projects, clean energy and transmission projects have federal historic review areas 10 times larger
Compared to fossil, clean energy and transmission are more than 8 times as likely to receive an “adverse effect” finding triggering additional process, delay, and mitigation obligations.
It's a similar story for NEPA and other permitting laws
Armand Domalewski
@ArmandDoma
the reason a pure transmission bill would be a bad idea is because current environmental laws *actively disadvantage* clean energy versus dirty energy. you are *hurting the climate* by refusing to reform them
-
Post not available here, view on XPost not available here, view on X
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
agreed this is a threat to the system, but aren't reviewers not reading the materials (like the appendix) part of the problem? need better combined human + AI review approaches
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Chris Hayes
@chrislhayes
It quite literally might be the singlest funniest thing that has ever happened in politics, ever.
Zack Bornstein
@ZackBornstein
I could write jokes for 800 years and I'd never think of something funnier than Trump booking the Four Seasons for his big presser, and it turning out to be the Four Seasons Total Landscaping parking lot between a dildo store and a crematorium.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Peter Baker
@peterbakernyt
Todd Blanche suggests on @CNNSOTU he will investigate Democrats for “threatening” Republicans: “They’re not even in charge now, saying ‘lawyer up.’”
Kash Patel, 2023: “We’re going to come after you.”
Steve Bannon, 2024: “You better lawyer up.”
Mike Davis, 2025: “Lawyer up.” -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
My comments on the implications of the election results for democracy in this @npratc story by @langfittnpr www.npr.org/2026/11/06/nx-s1-6029750/american-democracy-w…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Brian Tashman
@briantashman
Byron York
@ByronYork
From WSJ: "Several hawkish outside advisers have called Trump since Election Day to make the case that his legacy depends on forcing Iran to surrender its nuclear program, which would allow the president to boast the political sacrifice was worth the cost, U.S. officials and another source familiar said." www.wsj.com/world/middle-east/trumps-clashing-advisers-pr…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tyler Austin Harper
@Tyler_A_Harper
You can make a very good case that the beginning of the end of the Trump presidency was when the people of Minneapolis refused to back down after two people were killed in the street and the admin was forced to retreat. After that they stopped seeming so formidable and scary.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
On the Four Seasons Total Landscaping-versary! Six years ago today en.wikipedia.org/wiki/Four_Seasons_Total_Landscaping_pres…
Connor Ewing
@ConnorMEwing
So what are the chances someone suggested this Dana White and Trump pulled a Four Seasons Total Landscaping?
Aaron Rupar
@atrupar
Trump: “I am pleased to announce that the Great Dana White, a violent Fighter and Warrior, has been offered the job of Press Secretary of the United States. I have requested that he bring with him assistants, even though some are not from our Country, of Islam Makhachev, Khabib Nurmagomedov, Justin Gaethje, Daniel Cormier, Amanda Nunes, and Kayla Harris. We need these people to defend us against the Lunatic Left. Dana will be a Press Secretary like none other. Dana, please take the job! President DONALD J. TRUMP”
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
cinesthetic.
@TheCinesthetic
Happy 59th birthday to Courtney Thorne-Smith!
Few talk-show appearances have aged as hilariously as her legendary exchange with Norm Macdonald about CHAIRMAN OF THE BOARD. 😂🎂This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alyssa Farah Griffin
@Alyssafarah
This is personal for me. I was vaccinated as a kid for measles but my OBGYN determined my immunity wore off. While I was pregnant, I was told to avoid states with measles outbreaks. In the year of our Lord 2025.
Jonathan Reiner
@JReinerMD
An infant died this week in Wisconsin from measles. Babies aren’t vaccinated for measles until 12-15 months. A newborn gets passive immunity from their mother (assuming mom has been vaccinated). If the mother is unvaccinated and there’s measles circulating in the community, a baby can become infected and die from the disease. As more virus circulates we’ll see more children die. This doesn’t have to happen.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"The Commerce Department is giving clearance to buy and handle Nvidia’s chips to Chinese companies with ties to firms that have been subject to national-security-related trade restrictions."
Ana Swanson
@AnaSwanson
New: Trump administration cleared Nvidia H200 chips to Chinese firms with empty offices and military ties www.nytimes.com/2026/11/07/business/economy/nvidia-chips-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Connor Ewing
@ConnorMEwing
So what are the chances someone suggested this Dana White and Trump pulled a Four Seasons Total Landscaping?
Aaron Rupar
@atrupar
Trump: “I am pleased to announce that the Great Dana White, a violent Fighter and Warrior, has been offered the job of Press Secretary of the United States. I have requested that he bring with him assistants, even though some are not from our Country, of Islam Makhachev, Khabib Nurmagomedov, Justin Gaethje, Daniel Cormier, Amanda Nunes, and Kayla Harris. We need these people to defend us against the Lunatic Left. Dana will be a Press Secretary like none other. Dana, please take the job! President DONALD J. TRUMP”
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Norm Eisen
@NormEisen
This is so wrong
The reason “Trump didn’t steal the elections” is because we & the pro-voter coalition stopped him EIGHT times over the past 21 months
Here’s how: a thread (1/x) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Not allowing cutting-edge AI chips to get to China so easily seems like the one thing almost everyone can agree on. Shouldn't have been allowed to happen even without the spy angle.
Jerusalem
@JerusalemDemsas
There is no better avenue for incoming House & Senate Dems to take than hearings on wtf is going on with NVIDIA + Trump corruption + AI policy in the U.S.
It has the virtue of riding the wave of anti-tech populism and directing that energy towards the productive need to stop selling these chips to China.
Caleb Watney
@calebwatney
And there it is
giftarticle.ft.com/giftarticle/actions/redeem/468bd17d-f4… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jerusalem
@JerusalemDemsas
There is no better avenue for incoming House & Senate Dems to take than hearings on wtf is going on with NVIDIA + Trump corruption + AI policy in the U.S.
It has the virtue of riding the wave of anti-tech populism and directing that energy towards the productive need to stop selling these chips to China.
Caleb Watney
@calebwatney
And there it is
giftarticle.ft.com/giftarticle/actions/redeem/468bd17d-f4… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
From @dwallacewells, teen mental health on upswing. Add this to rapidly falling overdose rates, homicide rates that will finish 2026 at their all-time low in the US + many N. American cities, falling suicide rates. Many positive trends in the world! Avoid having Reddit Brain...
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
James Surowiecki
@JamesSurowiecki
"Walmart doesn't offer its employees health insurance" hasn't been true for many years. Every employee who works 30 or more hours - more than 70% of the company's store employees, and more than 80% of their distribution-center employees - qualifies for benefits.
Kaiz 🐂🀄️
@Kaiz_294
Talarico argues pro-business really means pro-big business, socialism for the rich, rugged individualism for everyone else
“I've seen that when people talk about being pro-business, what that usually means is being pro-big business, right?”
“we basically subsidize Walmart through Medicaid because they don't give their employees quality healthcare.”
“So again, socialism for the rich, rugged individualism for the rest of us.”
“And this is not a radical vision. It's what we had before.”
w/ @jamestalaricoThis media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
My process: write everything myself, incl. slides and so on. But I don't send any document out, use any deck, etc. without running it through a Codex process I wrote. *Everything* has at least small typos or errors that I ex-post want to fix. Be kind to your readers!
Ethan Mollick
@emollick
I think there is way too much concern expressed about verification of AI output in everyday work due to a risk of factual errors.
It used to be true, but data on more recent models suggest they beat human experts, including PhD specialists, in finding facts and checking claims. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matt Grossmann
@mattgrossmann
Across counties, there is not much evidence of direct policy effects on change in the Democratic House margin. Exposure to tariffs, gas price increases, and ICE arrests are unrelated to Democratic gains.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Chris McGuire
@ChrisRMcGuire
The story keeps getting crazier. To recap everything that has been reported so far:
- The US intelligence community discovered that Jensen Huang, the CEO of the largest company in the world, was sleeping with Alice Huang, a Chinese spy and an executive at Megaspeed. (FT)
- Multiple independent investigations have concluded that Megaspeed is likely a Chinese cutout, which helps Chinese companies access AI chips or circumvent export controls. (NYT and Bloomberg)
- The US intelligence community also concluded that Alice Huang’s goal was to convince Jensen to lobby the U.S. government to loosen export controls on advanced AI chips. (NYT)
- The intelligence community relayed this information to “a small group of top US government officials” earlier this year. (FT)
- Over the last year, Jensen Huang has repeatedly lobbied the U.S. government to loosen export controls on advanced chips. @BISgov has done so, not only by permitting the export of H200 chips to China, but also not enforcing export controls that would have required a license for exports of the most advanced Blackwell chips to Megaspeed and Southeast Asia.
- Although @CommerceGov leadership was presumably briefed on this information by the U.S. intelligence community, @BISgov has taken no actions to tighten controls on Megaspeed or Southeast Asia.
- Megaspeed remains a preferred cloud partner of Nvidia, and continues to receive Nvidia Blackwell chips despite employing a Chinese spy who reportedly is romantically engaged with the CEO of Nvidia.
- Tens of billions of dollars worth of AI chips have been exported to Southeast Asia this year alone, and Malaysia and Singapore are now two of the four largest importers of AI chips in the world.
What is going on here??? We need answers. The question of who gets the most advanced AI chips in the world is literally the defining strategic question of our time. And not only does it seem like the MSS may be directly influencing the debate on export controls in the United States, but also @CommerceGov is refusing to listen to the warnings of the IC on this issue. What is a POSSIBLE good explanation for why Megaspeed is still receiving chips? Why is there not a license requirement for AI chip shipments to all of Southeast Asia? This is beyond strange at this point.
The fallout from this story will be massive. Obviously, the motivations underpinning Nvidia’s policy views opposing export controls must now all be questioned. But also, how did @BISgov sit on this information and has still taken no actions? We need answers.
Demetri
@AsiaLens
US intelligence agencies believe @Nvidia CEO Jensen Huang had romantic relationship with alleged Chinese spy
as.ft.com/r/97c2f9f7-ad2e-44bb-8642-f60b19f19c54 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Literally all of this is in the Constitution
Kyle Cheney
@kyledcheney
Congress has the power to conduct traditional — even aggressive — oversight and threaten impeachment. Trump is ordering the Justice Department to criminally probe lawmakers for doing. truthsocial.com/@realDonaldTrump/117576401180733840
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The relevant benchmark is humans, not perfection. Frontier AIs make fewer errors on virtually every objective task at this point.
Ethan Mollick
@emollick
I think there is way too much concern expressed about verification of AI output in everyday work due to a risk of factual errors.
It used to be true, but data on more recent models suggest they beat human experts, including PhD specialists, in finding facts and checking claims. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
(Guy Zyskind*)
@GuyZys
ICLR 2027 reviews are out. I pulled every ICLR review since 2025, about 250k of them, and it's not looking great.
AI written reviews are now the majority (* based on a sample):
2025: 18%
2026: 41%
2027: 65%
Scores are collapsing into the middle, and as others have pointed out - most consolidate around "Weak reject" (score 2). Weak rejects went from 36% of all scores to 57%. Clear accepts and rejects fell from 41% to 21%.
An indication that this is an AI phenomena - AI-flagged reviews picked a clear side less than half as often as humans.
Then you have the really awkward slip-ups.
94 reviews this year call the paper "the provided manuscript." That's chatbot for "the PDF you uploaded." In 2025 there were zero.
28 complain the paper is full of garbled numbers. Those are probably the line numbers embedded in the submission's margins, which is what they look like after a chatbot turns the PDF into text.
23 reviews have a Weaknesses section written for a different paper, word for word. For example, one paper on RL training got criticized for its chemistry. It has no chemistry. In half of these cases, the two reviews were posted less than 9 minutes apart.
I'll leave everyone to draw their own conclusions, but clearly we're not adapting quickly enough. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Happy FSTL day to all who celebrate!
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sheryl Gay Stolberg
@SherylNYT
BOMBSHELL: Calley Means in tell-all interview with Joe Rogan accuses Susie Wiles of letting industry lobbyists run the show and block RFK Jr's agenda.
“Susie said to me many times: ‘We need money for the midterms, Calley."
w/ @By_CJewett
www.nytimes.com/2026/11/06/us/politics/susie-wiles-calley… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
MeidasTouch
@MeidasTouch
Calley Means, who recently resigned as a senior Trump White House adviser, tells Joe Rogan:
“Susie's daughter went to a lobbying firm and it was an open discussion in the White House. If a company hired Susie's daughter, then that was getting a high priority.”
“The group Susie's daughter worked for had almost zero in revenue before the Trump administration, shot up to 30ish million dollars a year in revenue.”
“So if you're a special interest and you want to get something done, the secret's out. You pay Susie's daughter.” -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
American Political Science Review
@apsrjournal
From our FirstView: "Economic Self-Interest and Support for Redistribution: New Evidence Using Election Results Data" by Jieun Lee, E. Grant Baldwin, @markarian_g, Benjamin J. Newman.
www.cambridge.org/core/journals/american-political-scienc… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
This. Prediction markets and polling averages/models are not the same.
Nicole Kagan
@nkagan_official
What markets are (and are not):
Prediction markets are, like all markets, weighing machines: they are an aggregate, or weight, of signals. Those signals can be people’s convictions, data on the ground, news reports, pundits' prognostications, or polling information. Just like Apple stock moves after earnings get released or a reputable bank analyst shorts the stock, prediction markets will move on new information.
The fact that markets moved on polls or model projections at all isn’t an indication that predication markets aren’t valuable, nor an indication that they are entirely reliant on them. It’s the opposite. The market did exactly what it's supposed to do: price in new information in real-time, and weigh it appropriately.
Comparing prediction markets and pollsters, like Nate Silver, is like comparing the stock market and Michael Burry: both are valuable but not the same. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
G Elliott Morris
@gelliottmorris
An independent analysis by the New York Times finds FiftyPlusOne had the most accurate election forecasting model over the whole 2026 election cycle, beating both prediction markets and other forecasters at almost all time horizons. Congrats team! www.nytimes.com/interactive/2026/11/06/business/polymarke…
-
G Elliott Morris
@gelliottmorris
Extreme levels of cope happening online in how centrists are responding to DSA candidates being electoral overperformers. This invalidates almost all of their predictions from 2024-2025 “WAR” debates
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
@gelliottmorris but quick Claude-checking suggests DSA overperformance reflects 2024 *under*performance and is basically a regression to the mean phenomenon; they don't look like overperformers vs 2020 vote shares
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Kinda good? Maybe?
CSPAN
@cspan
Mississippi Republican Party Chair @MikeHurstMS on whether he would accept a win by Sen. candidate Scott Colom (D): "There is no election-denying going on here. If the laws are followed and the voters spoke, that is it and Mississippi Republicans will accept that."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
G Elliott Morris
@gelliottmorris
I just posted this analysis of the @50plus1news model vs the prediction markets. Not only is @NateSilver538 correct and @mansourtarek_ wrong about the performance of the models vs markets over the cycle, the models were actually more accurate on Elec Day 2 blog.fiftyplusone.news/p/fiftyplusone-first-among-forecas…
Chris Hayes
@chrislhayes
I am 100% Team Nate in this beef. It’s *obvious* to anyone paying attention that the betting markets just moved over time towards the models. That’s the info they were working off of! They didn’t “discover” anything
Nate Silver
@NateSilver538
Give me a break, the polls and models provided the direction in contast to vibes and dumb insiders. And yes it's good that traders were smart enough to trust them. But the better election models were more confident about the outcome than the markets and got there more quickly.
Tarek Mansour
@mansourtarek_
Last night, the power to decide what America sees changed hands.
I have spent a ton of time in DC over the last few months. The elites were certain the markets were off. I heard many variants of “no chance markets are right about X” and “I’ve been in politics for a decade. ‘These people’ don’t really know much.”
Last night, ‘these people’ beat the elites.
Prediction market favorites won 479 of the first 487 races. That’s 98%.
The elites live in a bubble. They theorize, write, ideate, and mistake credentials for an understanding of lives they’ve never lived. They don’t see and feel what the everyday man, on the ground, sees and feels.
Knowledge is not exclusive to the elites but scattered across millions of people.
Someone knows their hometown better than any national pundit. Someone has followed a local race passionately for months. Someone has followed a local race dispassionately for months. Someone notices a shift in their community before it makes the headlines. Someone is shaped by an incident at work. Someone is shaped by what’s getting harder or easier for their families.
Each holds a piece of the picture that no expert can see in full.
And for the longest time, that knowledge had no way to reach the rest of the world.
For generations, election night meant turning on the TV and waiting for a handful of people to tell you what happened.
Prediction markets change that.
You don't need a title, a large following, or an editor to publish you. All you need is a perspective and the conviction to back it.
That changes who holds power over information.
Anyone can see the market. Anyone can contribute to it. Anyone can question it. No title gives you the final word. No gatekeepers can filter your view.
Sometimes the signal can be wrong. But anyone can challenge the consensus and improve it.
And when people consolidate their knowledge together, something remarkable happens: millions of individual perspectives become a signal that we all can learn from.
Over the last few months, the people showed up on Kalshi to tell the world what they were going to do.
Last night, the megaphone was passed to them. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Otto Von Tweetmarck
@OVTweetmarck
I cannot fathom voting to prohibit someone from...having a smaller house.
Jonathan Berk
@berkie1
Question 7, which would legalize starter homes on smaller lots across Massachusetts, remains too close to call, with just 19,000 votes separating the two sides.
While we wait, some interesting local results are emerging in Newton, Brookline, Milton, Nantucket and beyond... -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Amanda Carpenter
@amandacarpenter
"Lawyers who work on congressional investigations also point to a weapon a pardon cannot touch: the courts. A pardon can shield someone from criminal penalties, but it does nothing to protect a company or an individual from a civil suit to enforce a subpoena, or from the court order that could follow."
www.ms.now/news/house-democrats-subpoenas-trump-companies… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
SamuelGoldman
@SWGoldman
No credit for AP is one of my most unpopular higher ed opinions.
T. Greer
@Scholars_Stage
Contra the concluding argument of this essay, universities and departments absolutely should refuse to grant credit for AP scores.
Michael Strong
@flowidealism
Wow, a pretty damning article with diverse lines of evidence.
Jeremy Wayne Tate
@JeremyTate41
Most people do not realize just how easy the AP exams have gotten. They are not college level.
Why/how have they become so easy? What's going on?
This might shock you... A brutal new piece in Education Next just dropped, saying the decline in rigor of the AP in the last two decades is, in large part, because the content has been GUTTED. For example:
— AP Biology cut 40% of its material
— Required lines from the Aeneid in AP Latin halved
— Spanish Lit reading list cut from 56 works to 38
Then the questions themselves have become absurdly easier. Readability measures show English passages have fallen multiple grade levels, and some now include footnotes defining basic words...
A 2025 AP English prompt included footnote definitions for words like "optimism" and "pessimism."
At the same time, scoring was recalibrated so that more students pass: guessing penalties disappeared, rubrics grew more generous, readers were told to give students the benefit of the doubt, etc. — "no one at College Board had any incentive to maintain rigor."
College professors are now seeing students arrive with AP credit who lack basic content knowledge and writing ability. The College Board insists the exams have not become easier, but everybody familiar with them knows they've become an absolute joke.
Advanced Placement is no longer advanced. This is not college level.
It must be replaced. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
G Elliott Morris
@gelliottmorris
Black voter turnout held at ~96% of 2024, the highest rate for any midterm in at least a decade, in the Deep South on Tues. If Scott Colom manages to win Mississippi's Senate seat, it wouldn't have been possible without sky-high engagement of Black voters
www.gelliottmorris.com/p/2026-11-10-black-turnout-south -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Blake
@AaronBlake
Pretty biting stuff from Judge Chutkan in halting Trump's triumphal arch:
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
This war could not be going better
tyro
@DoubleEph
You have to email the Houthis if you want your ship to cross the Red Sea. They reply within 24hrs
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A remarkable victory for American democracy
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
People often misperceive candidate positions. But the most talented pols can buck the base and still be loved (see: Obama on gay marriage, Mamdani on NYPD/Tisch).
Matthew Yglesias
@mattyglesias
I am happy to learn that progressives have no problem with Annie Andrews’ positions on M4A or trans issues (or Talarico on oil and gas) but I feel like when I say this stuff they get angry at me.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sam Stein
@samstein
This is just the best damn thing ever
Sarah Longwell
@SarahLongwell25
In which I get to tell @jonstewart about the moment he changed my life.
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Carlos Odio
@carlosodio
This is about where the Blue Rose estimate lands. NORC data implies ~55% turnout / ~30% switching / ~15% new votes. Differences aside, all point to turnout differential as top accelerant of blue tsunami.
Lenny Bronner
@lennybronner
Using precinct results from 11 states, we estimate ~75% of Democrats’ net gain came from the turnout gap and ~20% from vote switching. Exit polls imply 35% nationally, but turnout still leads. 8 in 9 Trump voters who didn’t vote GOP stayed home.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The question for everyone who says candidate extremism doesn't matter - do you think that's true about extremists on the *other* side? For Ds: would a super-conservative candidate have done as well as Collins did in ME? The only serious debate is about effect sizes and tradeoffs.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Michael A. Cohen (NOT TRUMP’S FORMER FIXER)
@speechboy71
No, no, candidate extremism only hurts candidates I don't like
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Mad king watch
Dan Diamond
@ddiamond
Trump just declared the Oval Office a national monument, saying the move will preserve it.
That's mystified lawyers bc (1) the White House is already a national landmark (2) Trump has argued a POTUS can change monuments, if he wants.
w @Cat_Zakrzewski
www.washingtonpost.com/politics/2026/11/05/trump-wants-ov… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Adam Carlson
@admcrlsn
Really good point
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
David Broockman
@dbroockman
Nice graph from @gelliottmorris showing moderate candidates continue to do meaningfully better. This election was another win for political science!
G Elliott Morris
@gelliottmorris
I was really annoyed at the debate about what empirical measure to use for ideology so I coded a lil script to use a LLM judge candidate ideology with pairwise comparisons and then put everyone on a single L-R scale (after Patrick Wu's work).
Results are here. Judge for yourself
Generally, in the Senate, we are seeing a smaller benefit for moderates than the predictions made by the centrist crowd during the 2025 vintage of this debate.
And in the House, DSA candidates are running better than the average Democrat!!
Derek Thompson
@DKThomp
I think this is right.
It's not just that Talarico (et al) combines elements of left populism, abundance pragmatism, and cultural moderation. It's also that the definitions of "left" and "moderate" are in flux.
Take "left." Talarico's all-of-the-above position on energy, eg, would have been grounds for near excommunication from the climate left during the heights of the Green New Deal era. But many leftists now proudly claim him as one of their own, in part bc of his rhetoric on corporate abuse of power and "not left vs right, but top v bottom."
So what we're seeing might be fusionism. But just as importantly, the elements being fused are in flux.
Rotimi Adeoye
@_rotimia
The Democrat factional fighting is a relic of 2016. This election points toward a new liberal fusionism, candidates who combine leftist economic populism with centrist pragmatism. Talarico and Ossoff offer a model on this. The factional stuff is old news, beltway chatter.
Derek Thompson
@DKThomp
i think "moderates did better on tuesday" and "anti-rich vibes used to be identifiably 'left-wing' but they've become fairly normie in the last few years" are both accurate takes, altho i expect a lot of discourse warfare between these positions in the next few days
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Otis Reid
@otis_reid
If the claim is that a “smaller benefit for moderates” is that the most leftwing candidates are “only 6 points worse”than the most moderate set, what are we even arguing about? That’s enormous
G Elliott Morris
@gelliottmorris
I was really annoyed at the debate about what empirical measure to use for ideology so I coded a lil script to use a LLM judge candidate ideology with pairwise comparisons and then put everyone on a single L-R scale (after Patrick Wu's work).
Results are here. Judge for yourself
Generally, in the Senate, we are seeing a smaller benefit for moderates than the predictions made by the centrist crowd during the 2025 vintage of this debate.
And in the House, DSA candidates are running better than the average Democrat!!
Derek Thompson
@DKThomp
I think this is right.
It's not just that Talarico (et al) combines elements of left populism, abundance pragmatism, and cultural moderation. It's also that the definitions of "left" and "moderate" are in flux.
Take "left." Talarico's all-of-the-above position on energy, eg, would have been grounds for near excommunication from the climate left during the heights of the Green New Deal era. But many leftists now proudly claim him as one of their own, in part bc of his rhetoric on corporate abuse of power and "not left vs right, but top v bottom."
So what we're seeing might be fusionism. But just as importantly, the elements being fused are in flux.
Rotimi Adeoye
@_rotimia
The Democrat factional fighting is a relic of 2016. This election points toward a new liberal fusionism, candidates who combine leftist economic populism with centrist pragmatism. Talarico and Ossoff offer a model on this. The factional stuff is old news, beltway chatter.
Derek Thompson
@DKThomp
i think "moderates did better on tuesday" and "anti-rich vibes used to be identifiably 'left-wing' but they've become fairly normie in the last few years" are both accurate takes, altho i expect a lot of discourse warfare between these positions in the next few days
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Zachary Donnini
@ZacharyDonnini
The latest example of a tweet impacting real election results:
Bo French substantially underperformed Abbott in Collin County precincts with larger South Asian populations, a pattern that stands out clearly on the county's precinct map. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Responsiveness is one criterion but um...
-the president tried to overturn an election defeat and sparked an insurrection that almost killed VP & Speaker
-armed agents killing citizens and disappearing them
-judges warning the rule of law is in danger
-unprecedented corruption
Patrick Collison
@patrickc
It seems to me that democracy has on the whole worked very well in America over the last decade, and perhaps better than in any other major country in the West. National outcomes have repeatedly been responsive to underlying voter preferences on meaningful questions, and significant changes in policy have ensued. I’ve been reading Democracy in America this past week, and it’s striking how much it still rhymes nearly two centuries later.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
gronke
@gronke
Using the DoJ to harass journalists and political opponents.
Regularly using violent rhetoric to describe your political opponents.
Openly flouting federal court orders.
Politicizing scientific research.
It goes on and on. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Dylan still rocking his Palpatine costume
Vatican News
@VaticanNews
Pope Leo XIV met with American singer-songwriter Bob Dylan in a private audience on Wednesday, November 4.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Just lawless on things big and small
Scott MacFarlane
@MacFarlaneNews
Hoo boy.
Despite a court order prohibiting any work -- without court notification - at the site of Trump's Monumental Arch..... the Administration acknowledges there's been some work. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
wubenmensheng
@Wubenmensheng
New dataset alert⚠️Wiebrecht introduces the Parliamentary Elites in Authoritarian Regimes (PEAR) dataset, covering over 230,000 MP observations across 133 regimes.
doi.org/10.1093/jopres/xjag087 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Soumitra Shukla
@soumitrashukla9
Let me expand a little on why I find this letter problematic. (Full disclosure: I have only seen the quoted part.)
First, descriptive work is good work. We over-privilege causal studies but our job is to understand the world. Descriptive findings that suggest how things are changing and document important historical facts are a "new insight or result."
Second, one of the better approaches to science is the collages-of-evidence model. If many independent pieces of evidence point in the same direction, then that's a sign of a robust finding. A causal study is one piece in that bigger collage, not the final word.
Third, we need evidence NOW on how AI is changing jobs. This has implications for a host of policy topics, esp. retraining and reskilling the workforce.
Fourth, Job ads DO provide a high-resolution look into what employers ask for (yes, they are not perfect, but they are a key signal of demand!)
We should embrace new studies, and absolutely judge them critically, but a mere reactionary dislike toward descriptive findings isn't a scientific way to go about the economics of AI, especially since we are so early.
Kevin Rinz
@kevinrinz
In case it’s helpful: I submitted this paper to ReStud Insights using the 2000-word format. It was desk rejected because it “reads as a largely descriptive set of observations on AI mentions and language used in job ads, as [opposed] to developing a new insight or result.”
This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Deep sighs.
Julia Black
@mjnblack
whew www.nytimes.com/2026/11/05/technology/dario-amodei-anthro…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Let's hope so but I'll take this bet. Not a verifiable domain in the same way as math.
Fireside Alpha
@firesidealpha
OpenAI’s Greg Brockman says "death becoming a choice" is possible, and predicts AI will make real strides on curing diseases in the next 12 to 18 months
“I think that you look at something like biology, and that’s not something that’s unique to just humans. I think that we all suffer, that the question of even just thinking about death becoming a choice. All of this is possible.”
“Having cures for diseases is something that’s become also a trope in this field, but I think that you’re gonna see real delivery and real significant strides and impacts on biology, health, and curing of diseases over the next 12 months, 18 months.”
“I think it’s gonna take time to really deliver on the full impacts, just by the nature of this field. But I think it’s all possible and doable, and it’s exactly the kind of thing we have these models that are solving these unsolved problems. Clearly, we need to apply it to biology and life sciences.”
Fireside Alpha
@firesidealpha
OpenAI’s Greg Brockman reveals an unannounced JPMorgan Chase deal, with the bank using Codex and OpenAI’s cyber models to secure itself
“Well, so within the company, we were probably, I don’t know. Like, I think our enterprise revenue was about 30% versus consumer at 70%. And now we’re probably somewhere around, like, we’re majority enterprise. And it’s not that consumer has shrank. Right? It’s been growing, but it’s the enterprise segment has really taken off."
"And so I don’t know exactly how to quantify in the broader market, but you can see the way that we’re winning deals. So for example, actually, one that’s hot off the press, has not been previously announced, is that we just papered a partnership with JPMorgan Chase, and that they’re going to be using Codex."
"They’re going to be using our cyber models to secure themselves, deploying our models for enterprise transformation knowledge work across the company."
"And that for me is like such an iconic institution in such an important domain. Right? A highly regulated enterprise that trust is, like, so key to what they do and that they chose us and that we’re here to be a great partner.”This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Justin Baragona
@justinbaragona
Benny Johnson
@bennyjohnson
Odd. Notice how you don’t see any videos of Republicans bursting into tears, weeping and hysterical on social media after the loss? That’s because we have more to live for than party politics. We have families, churches, communities. We know the Government is not God.
God is God -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Julia Black
@mjnblack
this is why I love working at @VanityFair
yesterday, I saw half our staff clustered around our fashion correspondent's desk poring over this photo. As José puts it, "The image has several discrepancies. The style and make of the Chanel Flap bag is inconsistent with others the brand has sold, and the Valentino shoes in question appear askew in proportion; their straps have several punched holes that the authentic Valentino style does not."
AI image detectors confirmed that yes, Olivia Reingold used an AI-generated image to make her point, without a drop of disclosure. Wild!!!!
www.vanityfair.com/story/midterms-2026-fashion-analysis
Olivia Reingold
@Olivia_Reingold
So far at Abdul El-Sayed’s election night event I’ve spotted, Valentino heels, a Chanel bag, a Moncler jacket, Rolex watches, and a YSL bag among other designer items.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
All of this. So many aspects of society can't survive ubiquitous agents in their current form.
Neal Khosla
@nealkhosla
agents are going to cause a breakdown of human behavior assumptions underlying today's systems. i've started thinking about it as friction collapse. without friction, a number of norms will collapse.
consider selling tickets to a concert/event. today, you can post them at a certain time and it's first come first serve. soon everyone's agent will persistently monitor, and as soon as they are posted for sale, thousands of people will instantaneously be trying to buy the same ticket. how do you even resolve this? sell the ticket to the person whose agent gets there fractions of a second before?
we're seeing the same thing with job applications, restaurant reservations, but basically anything that relies on participation requiring effort is going to break down. sales emails/calls, participation in government benefits, apartment rentals, doctors appointments, RFP responses. customer support won't be able to handle the user who will not drop their point of dissatisfaction. with no friction, these systems just don't work anymore.
business models will break down:
- flights/hotels will be consistently cancelled and rebooked at the lowest price (great use case for your agent if you don't have this already set up)
- subscriptions will be cancelled
- retailers won't be able to arbitrage price differences as the agent will find the cheapest place to buy anything
- insurance denials (health/auto/home/etc.) will be fought tooth and nail
- bank interest rates will be optimized and avoidable charges will be avoided
a lot of what we perceive as stable social coordination is actually an equilibrium sustained by human inefficiency. rationing is a product of human slowness. signaling effort/interest/intent is a key part of our social infrastructure. inertia is a key to many business models. information asymmetry is everywhere. and attention is monetized thoroughly.
with agents, none of these things are true. humans will simply demand what they want and have their agents get it for them. there is going to be a fundamental renegotiation of how markets allocate resources, how people signal intent, and how businesses capture value. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Remarkable graphs. We are a nationalized country.
Tom Wood
@thomasjwood
Even if you're familiar with the nationalization of politics and @dhopkins1776's brilliant book, the the uniformity of the senate shift in '26 is striking
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tom Wood
@thomasjwood
Even if you're familiar with the nationalization of politics and @dhopkins1776's brilliant book, the the uniformity of the senate shift in '26 is striking
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Fake nostalgia is dumb
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jasper Scherer
@jaspscherer
Recent Texas cycles each had a regional story. In 2018, Dems surged in the suburbs and the rural wall saved Cruz. In 2020 and 2024, the RGV and South Texas lurched right.
This year, it's everywhere. All 254 counties moved left: -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Important @leedrutman on the need to strengthen democracy in a moment like this. Winning elections forever is not a plan.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Excellent @zackbeauchamp on how Trump's personalist regime created the seeds of his downfall, the resilience of American democracy, and what comes next www.public-reason.com/p/trumps-authoritarianism-caused-the (sign up for his new Substack!)
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin DeLuca 🗞
@cantstopkevin
Here's the under/over performance in some key Senate races, using 2024 presidential results + 2026 national environment as the baseline expectation
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Super interesting "The Illusion of Time."
College-graduating women in Pakistan underestimate how soon marriage activities arise (i.e. family pressure).
If they don't get in gear and get a job right away, the window is gone and they might never enter workforce. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kate Kelly
@katekelly
“This airplane has a soul and it’s trying to kill you.” Bracing analysis of the tragic plane crash in West Palm from a longtime Lear 60 pilot
Dave
@HangryDBowman
Long post worth the read. I have a close friend who flies private jets. He flew Lear 60 for over 15 years and now flies a Gulfstream. This is a copy and paste of his response to me asking him if I thought Ice was the issue.
“We always fly in ridiculous cold temperatures.”
“Usually you get the bad ice when the temperatures are like 0 to negative 20. Believe it or not, there's a point where it gets too cold for ice. At negative 40, you pretty much can't get ice. Besides that, the airplane is equipped with hot wings and hot tail.”
“I flew the Lear 60 for like 15 years. It's a no joke airplane and probably the most dangerous airplane I've ever flown. It has a super critical wing. The airflow has to be perfect. When it gets into rough air and turbulence, it will easily stall. It has stalled on me at 41,000 feet just because we got into some turbulence. (On the way to Palm Beach as a matter of fact. ) we recovered and got her flying again around 30,000 ft. Scary
But here's the problem. This airplane has a terrible coffin corner. (At high altitudes) what that means is, you fly in a really narrow window where if you lose 20 knots, it stalls. And if you gain 20 knots, it goes into overspeed. The air goes supersonic over the wing and it goes into a high speed stall, or worse, a Mach tuck.
So if you stall, the normal recovery is to push the yoke forward to lower the nose and gain air speed again. At low altitude, you can push really hard and recover quickly. But in the 60... if you push hard, it will gain air speed and go into a high speed stall, it doesn't take much. It's very sensitive. The 60 has super short wings that don't make much lift. So then you high speed stall, and the computer is yelling at you... OVERSPEED! So then you pull back and enter a low speed stall. And you go back and forth trying to keep it in the air speed window.
This whole thing is pretty scary. And you should recover basically by using your finger tips using super small corrections. You can't over do it. If you try to muscle the recovery, she gets mad and will become unrecoverable. If she enters a mach tuck. You're done.”
“I taught the Lear 60 to other pilots and mentored them. I told them... this airplane has a soul and it's trying to kill you. Do not ever try to top the weather. Don't go over a thunderstorm. Go around it. If it means a half hour longer, then do it. Going over the thunderstorm will go into turbulence that you might not recover from.
These pilots went right over a thunderstorm. I think it stalled which caused them to drop down into the thunderstorm and ended up in mach tuck. They picked up so much speed that it ripped off the horizontal stabilizer. At that point... chuck Yeager can't fix it.
These same pilots did something similar a couple months ago. ATC got mad at them. They poked the bear one too many times.
This is not a beginner airplane. The Gulfstream would never do this. It has HUGE margins of error and won't stall like that. But then again... it's got HUGE wings. It's a pussy cat.
The Gulfstream is like driving a 4 wheeler. The Lear 60 is like riding a uni cycle.” -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Yesh Ginsburg
@yesh222
Most of the State Senators who refused this gerrymander got primaried out of their seats. They saved Republicans from the most hilarious dummymander in history.
Indianapolis Councilor Nick Roberts
@nickroberts317
This is insane! Here is a map of yesterday’s Indiana House results if they were drawn under the attempted Republican gerrymander map from last Fall. Thank you to @Vulst3 for breaking down the data to make this map. The gerrymander could’ve netted us a 9-0 or 8-1 map potentially!
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nate Silver
@NateSilver538
OK, wrote about the models vs. prediction markets debate. Actually, the real stolen valor is from NYT/Siena. But prediction markets converged toward FLIPR/model prices, not the other way around, and the models got to the right answer sooner and more confidently.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Raphael Nishimura
@rnishimura
Many criticize the polls for "getting the election results wrong" and say the election models or prediction markets are more accurate. But they ignore that polls are the main input of those models and markets. Without polls, modelers and bettors would be wandering in the unknown.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
John Sides
@johnmsides
There were models out there predicting 55-56 Democratic Senate seats and/or ~250 Dem House seats. Just floating under the radar in a polisci journal. See my piece:
substack.com/home/post/p-222806485
Will write more on those models. Of course, no one election proves any model is right!
Lakshya Jain
@lxeagle17
An early postmortem on our models and polls — namely, Republicans did worse than *anyone* predicted.
www.theargumentmag.com/p/republicans-did-even-worse-than-we -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Lucas Holtz
@LucasHoltz__
NEW midterm analysis from @ThirdWayTweet @TheWelcomeTeam
Red-State Democrats—all moderates—overperformed & delivered the Senate. On avg they ran 21 points ahead of Harris’s 2024 margin & 5 points ahead of Dem statewide House vote.
Talarico: +26 vs Harris, +6 vs House Democrats -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Time to start asking hard questions of the knee-benders
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Remember when a huge portion of civil society including corporate leaders, top law firms, and elite universities decided to submit to Trump's authoritarianism? And it was legitimized because he won in 2024 with <checks notes> 49.7% of the popular vote? What now?
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tom Wood
@thomasjwood
The breadth of the demographic swing against the GOP this midterm is all the most striking because we saw a comparably consistent swing *to* the GOP only 4 years ago, in 2022.
Data from the fantastic newly released @APNORC Civic Health & Elections poll. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Replotted as a % of 2024 turnout by group - look at those increases for Latino and Asian/PI voters
Bernard L. Fraga
@blfraga
Georgia just posted its vote history, and as race is self-reported on the file we now have our first stats on race & turnout. Black (1.37m), Latino (198k), and AAPI (149k) turnout was a midterm record. White turnout down 100k voters vs 2022, as 200k more Black voters turned out.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Rob Friedman
@PitchingNinja
2026 PitchingNinja Award for Funniest Reaction by a Hitter (Below 100+ mph division). 🤣🏆
Winner: Luis ArráezThis media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Rob Friedman
@PitchingNinja
2026 PitchingNinja Award for Funniest Reaction by a Hitter (100+ mph division). 🤣🔥🏆
Winner: Max Clark vs. Seth HernandezThis media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Um, a political scientist thinks it didn't (currently +15 D)
Sumit Roy
@sumitroy2
A Bloomberg journalist thinks ChatGPT might have just changed how this election turned out.
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Christoffer H. Dausgaard
@chdausgaard
Did counties where gas prices hit harder swing further against the Republicans than similar neighbors?
My pre-registered test, filed the day before the midterms, finds no sign of this. That's not to say that gas prices didn't matter, but it sheds some light on the mechanism.
🧵
Christoffer H. Dausgaard
@chdausgaard
How will high gas prices affect the midterms?
It's an important question but hard to answer empirically. I've written down a plan to test part of it, and I'll publish my first results on Thursday, when most of the vote is in. 1/🧵
chdausgaard.github.io/fuel-2026 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Saba Hamedy
@saba_h
This lede! www.ajc.com/politics/2026/11/keisha-lance-bottoms-crawfor…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jordan Weissmann
@JHWeissmann
The more I think about this tweet, the more incredible it seems.
JD Vance is really trying to save his political career by writing a little essay critiquing his own administration’s war strategy like he’s the public editor. Just terminally poster-brained.
JD Vance
@JDVance
One of the things the data analysts on our side obsessed over was "composition"—whether the electorate looked close enough to 2024 for Republicans to win. And despite polls that showed a Democrat blowout, the data kept on telling us the same story: in some states turnout was merely OK, in other states it was downright amazing, and in no state was it catastrophic. There was a genuine sense of optimism on our team: our people were showing up.
So, I asked the smartest data guy I know, does that mean we're going to shock the world? "No," he told me. "We might do better than the polls suggest. But the data is telling us that a lot of Trump/Vance 2024 voters are pissed and planning to vote for the other guys. Our voters are turning out, but they may not vote for us."
Well, that looks prophetic now. Our voters did show up. In Florida, for example, the electorate was 9 points more Republican than Democrat, just a couple points worse than 2024. Undoubtedly there were places we could have done better on turnout. But the main problem is not turnout. The main problem is that many people who voted for Trump/Vance in 2024 voted for Democrats in 2026.
There will be a lot of infighting and arguing over what happened and why that is. One of the pollsters I really respect told me that our economic messaging wasn’t sharp enough down the stretch. I’m sure that’s true.
But the real reason we lost is obvious to almost everyone.
Another pollster—one of the smartest people I know in politics, told me yesterday we’d lose the generic ballot by 13 points. I felt negative going into Tuesday, but even I was shocked by this. In his polling, he saw that a lot of Trump 2024 voters were angry and planning to vote for Democrats. He asked those voters who had abandoned us what would cause them to come back on our team. The top four answers, accounting for nearly all of the responses were:
1) End the war
2) Focus on cost of living/inflation
3) Put America First
4) Change our relationship to Israel
Our voters showed up. And the above four reasons are why many of them voted for Democrats. This is according to a pollster who nailed the results. While many spent Monday attacking Rachel Campos Duffy for calling for a diplomatic resolution to the conflict, it turns out her political advice was sound. (She’s also a great person.)
In the next few months, there's going to be a lot of gaslighting. Expect to hear any number of establishment politicians say the solution to our woes is amnesty. I'm open to the idea that we should change enforcement—by focusing more on businesses that hire illegal labor, for instance. But anyone pushing amnesty is lying to you.
The reason we lost is the war. The president himself said so last night.
If you'd like to argue it was worth it, that losing the Senate for the next 6 years was worth it, then do so. Argue it was a worthwhile cost to pay. Agree or disagree, that's at least honest.
But everyone knows why we lost. And any honest conversation about where we go from here must start there.
I remain hopeful about our country. To the people who are pissed off and want to say, “I told you so”: you’re right, you did tell us so. Take that energy into electing Republicans who share your worldview. To the people who are just pissed at the results, you should be. Let’s take that anger and get back to the work of saving our country. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Thomas Massie
@RepThomasMassie
There is one contingency in which I won’t reintroduce the articles of impeachment:
Hegseth resigns.
Natasha Bertrand
@NatashaBertrand
“Rep. Thomas Massie of Kentucky, the famously anti-Trump GOP lawmaker, told CNN he plans to trigger a vote to impeach Hegseth when Congress returns next week — and multiple GOP sources said they expect it to pass.” www.cnn.com/2026/11/04/politics/pete-hegseth-impeachment-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
David Shor
@davidshor
Last night was a historic win for Democrats.
We currently estimate Democrats will win ~57% of the two-party House vote and ~255-260 seats - the best House vote share for either party since Watergate, and likely the biggest majority in more than 30 years. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Natasha Bertrand
@NatashaBertrand
“Rep. Thomas Massie of Kentucky, the famously anti-Trump GOP lawmaker, told CNN he plans to trigger a vote to impeach Hegseth when Congress returns next week — and multiple GOP sources said they expect it to pass.” www.cnn.com/2026/11/04/politics/pete-hegseth-impeachment-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin DeLuca 🗞
@cantstopkevin
Here's the updated estimates of incumbency advantages using our 2026 data
Kevin DeLuca 🗞
@cantstopkevin
This data will be particularly useful for analyzing longer-term trends, like this graph below! Stay tuned!
This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Josh Metcalf
@josh_metcalf
The SC GOP showed a surprising amount of political acumen in nixing last year's proposed redistricting plan. Had they gone through with their attempted 7-0 lockout map, it easily could have backfired and yielded a 6-1 split for Dems.
Makes one wonder if southern Rs will think twice about dismantling the remaining VRA districts in GA, SC, AL, etc... -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Chris Hayes
@chrislhayes
Think bigger: let’s get rid of the Electoral College, now that the GOP has to stare down the possibility of a D Presidential ticket carrying the state by 10,000 votes and losing every last electoral vote.
Joe
@JoePostingg
Democrats should offer a gerrymandering ban, and should be willing to compromise in good faith to get it done. It's such a corrosive negative sum game.
Patrick Ruffini
@PatrickRuffini
You can argue whether or not redistricting cost Republicans House seats via dummymander, but you can’t argue that it helped cost them at least two Southern Senate seats that are going to make it very difficult to control the chamber again in the near future.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jan Zilinsky
@janzilinsky
He says many 2024 Trump/Vance voters switched or abandoned the GOP because of cost of living/inflation, citing polling (what people said mattered to them).
I might have a better measure - let's tally up the actual economic hardships people experienced (losing a job, borrowing money to pay the bills & more) and see how they voted.
Trump voters who reported four or more economic hardships were nearly three times as likely to back a Democrat for Senate as those who reported none.
As @BrendanNyhan has been saying there's no reason to take stated reasons for voting a certain way at face value www.nytimes.com/2014/11/05/upshot/why-did-i-vote-that-way…; and @rp_griffin, @johnmsides have documented that asking detailed questions about people's personal finances is the way to go: www.voterstudygroup.org/publication/in-the-red
JD Vance
@JDVance
One of the things the data analysts on our side obsessed over was "composition"—whether the electorate looked close enough to 2024 for Republicans to win. And despite polls that showed a Democrat blowout, the data kept on telling us the same story: in some states turnout was merely OK, in other states it was downright amazing, and in no state was it catastrophic. There was a genuine sense of optimism on our team: our people were showing up.
So, I asked the smartest data guy I know, does that mean we're going to shock the world? "No," he told me. "We might do better than the polls suggest. But the data is telling us that a lot of Trump/Vance 2024 voters are pissed and planning to vote for the other guys. Our voters are turning out, but they may not vote for us."
Well, that looks prophetic now. Our voters did show up. In Florida, for example, the electorate was 9 points more Republican than Democrat, just a couple points worse than 2024. Undoubtedly there were places we could have done better on turnout. But the main problem is not turnout. The main problem is that many people who voted for Trump/Vance in 2024 voted for Democrats in 2026.
There will be a lot of infighting and arguing over what happened and why that is. One of the pollsters I really respect told me that our economic messaging wasn’t sharp enough down the stretch. I’m sure that’s true.
But the real reason we lost is obvious to almost everyone.
Another pollster—one of the smartest people I know in politics, told me yesterday we’d lose the generic ballot by 13 points. I felt negative going into Tuesday, but even I was shocked by this. In his polling, he saw that a lot of Trump 2024 voters were angry and planning to vote for Democrats. He asked those voters who had abandoned us what would cause them to come back on our team. The top four answers, accounting for nearly all of the responses were:
1) End the war
2) Focus on cost of living/inflation
3) Put America First
4) Change our relationship to Israel
Our voters showed up. And the above four reasons are why many of them voted for Democrats. This is according to a pollster who nailed the results. While many spent Monday attacking Rachel Campos Duffy for calling for a diplomatic resolution to the conflict, it turns out her political advice was sound. (She’s also a great person.)
In the next few months, there's going to be a lot of gaslighting. Expect to hear any number of establishment politicians say the solution to our woes is amnesty. I'm open to the idea that we should change enforcement—by focusing more on businesses that hire illegal labor, for instance. But anyone pushing amnesty is lying to you.
The reason we lost is the war. The president himself said so last night.
If you'd like to argue it was worth it, that losing the Senate for the next 6 years was worth it, then do so. Argue it was a worthwhile cost to pay. Agree or disagree, that's at least honest.
But everyone knows why we lost. And any honest conversation about where we go from here must start there.
I remain hopeful about our country. To the people who are pissed off and want to say, “I told you so”: you’re right, you did tell us so. Take that energy into electing Republicans who share your worldview. To the people who are just pissed at the results, you should be. Let’s take that anger and get back to the work of saving our country. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jan Zilinsky
@janzilinsky
NORC just released an amazing dataset (its Civic Health & Elections Project) so we can test, again, whether it's really true that (young) men are massively more conservative than women
The ideological gender gap exists but... -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin DeLuca 🗞
@cantstopkevin
@shirokuriwaki, @jeremiahbcha, and I have collected candidate- and contest-level data for the 2026 election, which you anyone can access here: docs.google.com/spreadsheets/d/1otML47xPsa9Wf1A_XEISLJFHE…
We also include historical results (1990–2024) on a separate tab, and some statewide stats. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Mitt Romney
@MittRomney
Bad policy, bad character make bad politics: erratic tariffs, impulsive Iran war, blanket pardons, apparent corruption, weaponized justice, ICE brutality, mountainous deficits, hiding Epstein, abandoning allies and coddling enemies. Hoping for a course correction, for the country and for my party.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Brad Hargreaves
@bhargreaves
Some dude in California running a 3-person company got voters to give him $4.2 billion by authoring a ballot measure, which passed
immad
@immad
Something feels a bit iffy about California prop 38.
Doing a popular vote proposed by a 3 person non-profit that gives said non profit $4.2B from the public!
I hope Gary is great and knows how to spend $4.2B well because it’s done. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nate Cohn
@Nate_Cohn
In the end, our final polls did not skew toward either side. Our old weighting approach would have overestimated Dems by 3.
The lesson: nonresponse was probably a bigger problem than in prior midterms, but weighting to a 2024 electorate (like many other pollsters) obviously wasn't a solution, either -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Reminder: Polling/forecasting accuracy is hard to measure when variation is driven largely by a big uniform swing. Easier to distinguish luck from skill in a world with more idiosyncratic race-level variation. (Lots of shoddy firms touted their "accuracy" after Trump won.)
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Mike (not a) Rothschild
@rothschildmd
"When election results are not close….no one looks under the hood."
Cleta Mitchell is one of the biggest names in the Stolen Election Industry, and even she's admitting there's nothing for election deniers to work with in 2026. This is still a dangerous movement, but it's defanged for now.
Cleta Mitchell
@CletaMitchell
Reporters are asking me today, "so what's next for election integrity?" Here's the response, to everyone:
The integrity of America's elections is vital to our country's future and those of us who live and breathe it are NOT going away!
America, roll up your sleeves. The work of securing our elections is FAR from completed!
We know – and the American people know – that there are myriad problems in our election systems and that election results for one candidate or political party are not the goal of the @EIwatchdogs
The goal is to have honest, transparent, and accurate elections EVERY time, so the people can trust the outcome, regardless of who wins or who loses.
Democrats don’t get that. The media doesn’t get that.
The focus of the leftist crazies for the last several months was whether President Trump would ‘accept’ the outcome of the elections. What is that even about?
When election results are not close….no one looks under the hood.
It is only when you have a very close election that people start examining the process and whether there were dead people voting, whether there are votes cast illegally from po boxes or vacant lots….whether the tabulators jammed so that ballots were scanned more than once….whether the printers at the vote centers were working…whether voters had to be sent to a different location….how many people showed up to vote and were told they had ‘already voted…’. And on and on.
The work of the Election Integrity Network electionintegritynetwork.org is needed NOW more than ever.
Election integrity cannot wait until the next campaign, the next controversy or the next close race. The rules that govern tomorrow’s elections are being shaped today.
Democrats who gained power will pursue election law changes that will put ‘voter convenience’ ahead of secure, accountable and trustworthy elections.
Their goal is to model every state’s election laws after California’s. Americans concerned about election integrity need to be ready for that fight in every statehouse, every legislative committee and every county election office.
What do you need to do? Watch the proposals in your state legislature.
Challenge efforts to weaken voter identification – which was passed overwhelmingly in several ballot questions on Nov. 3, by the way.
Demand that voter registration systems verify eligibility and keep accurate rolls.
Push back against expanded mail voting and legalized ballot harvesting.
Insist on meaningful audits, public transparency and laws that guarantee a fair process.
Demand accountability.
“Access” should never become an excuse to waive measures to verify eligibility.
Who is eligible? How is identity verified? Who handled the ballot? Can the count be checked?
Those questions deserve answers, not accusations that asking them is somehow an attack on democracy.
Watch efforts to reverse bans on ranked-choice voting.
Pay attention to the “democrat” in Democrat Socialists of America. The socialist path does not have to begin with a revolution. It can advance through the ballot box, one majority vote at a time. That is why Americans who oppose socialism must take the fight over election rules seriously.
The National Popular Vote Interstate Compact is more than an election reform proposal. It is a way to make the nationwide popular vote decisive in choosing the president. This change will put America on a socialist path by way of “democracy”. The Democrat Socialists of America will use fifty-one percent to erase the constitutional rights of the other forty-nine.
But outrage alone will secure nothing.
Get involved in an Election Integrity Network coalition in your state. Study your state laws. Measure state laws against the principles of the U.S. Citizens Elections Bill of Rights. Identify the gaps. Use the Model Election Laws Handbook to strengthen your state’s election laws. Read the election bills proposed in your state. Show up at hearings. Become a trained poll worker or observer. Demand safeguards that protect eligible citizens’ votes and make the results independently verifiable.
Bring evidence. Know the rules well enough to challenge weak answers. Stay involved after the headlines fade.
Be right and be polite...but be persistent.
Hopelessness is surrender in advance. Every citizen who decides there is no point leaves these decisions to someone else.
We need all hands on deck.
There is work to do. And the time to do it is now! Today!
If not you, then who?
The consent of the governed means something only when we the people can trust and verify the process that records our consent.
We welcome you to join the ranks of the Election Integrity Network. Let’s start today! @EIwatchdogs @BasedMikeLee @seanmdav @RealAmVoice @Bannons_WarRoom @seanspicer -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Trump is going to be playing with his models even more now
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Anna Bower
@AnnaBower
In his first public appearance since his party suffered crushing electoral losses, Trump is at the site of the White House ballroom, talking about the marble he ordered for the floor.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Gary King
@kinggary
New paper: "Replication Certification at Publication: Checkable by Anyone, Anywhere, in Seconds, Forever" garyking.org/certify. with @xuyiqing & @YangYang_Leo. Comments welcome!
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Max Spero
@max_spero_
We've been working towards AI image and video detection for months, and now it's finally here.
There were a lot of challenges along the way. There are way more AI image/video models than frontier LLMs, and their outputs are more diverse. Most phone cameras, including iPhones, do some level of AI postprocessing to add detail to real photos.
We had to build a model that's robust to this postprocessing while understanding the difference between AI edits and fully AI-generated visuals.
I think the heatmap feature is a really important step forward and a huge differentiator over other models which typically report scores as a binary confidence.
Please try it out!!
Pangram
@pangram
Introducing Pangram Image 1.0, our newest model for AI image and video detection.
Pangram Image 1.0 identifies 99.2% of AI-generated images while maintaining a 0.27% FPR.
It integrates into our Chrome extension, so you can check for AI images and video anywhere on the web. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Superforecasters Outscored Polymarket on the 2026 Senate Races
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Everyone's in favor of democracy until the people make a different choice than they would
Antoine Levy
@LevyAntoine
Still not called, but in a D+14 environment, in *California*, wealth taxes not only failing but a permanent wealth tax prohibition likely passing feels like something that people who tell us they know better about the will of the people should pay attention to.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Scott MacFarlane
@MacFarlaneNews
Didn’t take long
House Judiciary Committee Democrats are hiring investigators as they gear up to launch probes of Trump 2.0 .. with subpoena power -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jeremiah Johnson 🌐
@JeremiahDJohns
Wave so blue that Democrats who dropped out and don't even want to be there are still getting elected
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sean Wu
@sean_n_wu
The Collaborative Intelligence Benchmark paper is out!
We tested 18 AI models as teammates on 240 tasks. The best model scores only 82.2 out of 100.
Frontier models almost never cheat. But when a teammate pushes back, they give in!
Read the paper at arxiv.org/abs/2611.02633 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Kaczynski
@KFILE
Every single election denier running to govern a swing state lost last night, by the way.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Political scientists to read today:
-Juan Linz on presidentialism and the danger of dual claims to legitimacy in situations like this one muse.jhu.edu/article/225694
-Grossback et al. on how politicians respond to perceived mandates www.cambridge.org/core/books/mandate-politics/1EA0CA0202D… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Kaczynski
@KFILE
The last Democratic senator from Kansas was born the same year the light bulb was invented, so that’s kind of wild.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dustin Gouker
@DustinGouker
Again, let's put election trading volume at prediction markets in context.
Through Election Day, Kalshi has $1.9 billion in volume for midterms *in total*, per @TickerTracker_. Kalshi did $4.2 billion in sports.... last week. Not including parlays. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Masley
@AndyMasley
We finally get a more detailed study on data center waste heat, and it appears (as I predicted using very simple high school physics) that they're not capable of producing the heating effects that last study suggested iopscience.iop.org/article/10.1088/1748-9326/aead02
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Patrick Svitek
@PatrickSvitek
Vance: "The reason we lost is the war. The president himself said so last night."
JD Vance
@JDVance
One of the things the data analysts on our side obsessed over was "composition"—whether the electorate looked close enough to 2024 for Republicans to win. And despite polls that showed a Democrat blowout, the data kept on telling us the same story: in some states turnout was merely OK, in other states it was downright amazing, and in no state was it catastrophic. There was a genuine sense of optimism on our team: our people were showing up.
So, I asked the smartest data guy I know, does that mean we're going to shock the world? "No," he told me. "We might do better than the polls suggest. But the data is telling us that a lot of Trump/Vance 2024 voters are pissed and planning to vote for the other guys. Our voters are turning out, but they may not vote for us."
Well, that looks prophetic now. Our voters did show up. In Florida, for example, the electorate was 9 points more Republican than Democrat, just a couple points worse than 2024. Undoubtedly there were places we could have done better on turnout. But the main problem is not turnout. The main problem is that many people who voted for Trump/Vance in 2024 voted for Democrats in 2026.
There will be a lot of infighting and arguing over what happened and why that is. One of the pollsters I really respect told me that our economic messaging wasn’t sharp enough down the stretch. I’m sure that’s true.
But the real reason we lost is obvious to almost everyone.
Another pollster—one of the smartest people I know in politics, told me yesterday we’d lose the generic ballot by 13 points. I felt negative going into Tuesday, but even I was shocked by this. In his polling, he saw that a lot of Trump 2024 voters were angry and planning to vote for Democrats. He asked those voters who had abandoned us what would cause them to come back on our team. The top four answers, accounting for nearly all of the responses were:
1) End the war
2) Focus on cost of living/inflation
3) Put America First
4) Change our relationship to Israel
Our voters showed up. And the above four reasons are why many of them voted for Democrats. This is according to a pollster who nailed the results. While many spent Monday attacking Rachel Campos Duffy for calling for a diplomatic resolution to the conflict, it turns out her political advice was sound. (She’s also a great person.)
In the next few months, there's going to be a lot of gaslighting. Expect to hear any number of establishment politicians say the solution to our woes is amnesty. I'm open to the idea that we should change enforcement—by focusing more on businesses that hire illegal labor, for instance. But anyone pushing amnesty is lying to you.
The reason we lost is the war. The president himself said so last night.
If you'd like to argue it was worth it, that losing the Senate for the next 6 years was worth it, then do so. Argue it was a worthwhile cost to pay. Agree or disagree, that's at least honest.
But everyone knows why we lost. And any honest conversation about where we go from here must start there.
I remain hopeful about our country. To the people who are pissed off and want to say, “I told you so”: you’re right, you did tell us so. Take that energy into electing Republicans who share your worldview. To the people who are just pissed at the results, you should be. Let’s take that anger and get back to the work of saving our country. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Roger Parloff
@rparloff
It's official. Judge Sarokin has dismissed the criminal case against Ksenia Petrova (frog embryo researcher) rather than hand over discovery ordered in connection her motion to dismiss for vindictive prosecution.
Anna Bower
@AnnaBower
NEWS: Prosecutors move to dismiss the indictment against Kseniia Petrova, the Russian dissident and Harvard researcher charged with smuggling clawed frog embryos after she challenged her ICE detention.
The motion comes soon after the judge granted discovery on her selective or vindictive prosecution motion. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
She overperformed Rs in every competitive Senate race and nearly won in an impossible environment. So much pundit fallacy today.
Jill Filipovic
@JillFilipovic
Susan Collins lost her race because of Susan Collins -- and her support of Brett Kavanaugh. She adopted a persona of a fearless independent who would fight for Mainers no matter the party politics. But when it mattered, she took the coward's route.
slate.com/news-and-politics/2026/11/susan-collins-maine-l…
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Every poster right now who likes the outcome:
"Last night shows Americans want XXX. They're sick of YYY."
where XXX and YYY is whatever you like/don't like -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
AJPS
@AJPS_Editor
AJPS has introduced a new submission cap policy, effective today. Under the new policy, individuals may submit up to three manuscripts as an author or co-author in a 12-month period, with an option to request a waiver for a fourth submission. Read more: ajps.org/2026/11/04/new-ajps-submission-cap-policy/
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Remember when a huge portion of civil society including corporate leaders, top law firms, and elite universities decided to submit to Trump's authoritarianism? And it was legitimized because he won in 2024 with <checks notes> 49.7% of the popular vote? What now?
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Peter Beinart
@PeterBeinart
Over the last two years, many of the most powerful people in America-tech oligarchs, college presidents, media tycoons, leaders of big law firms + basically the entire pro-Israel Jewish establishment-capitulated to Trump's authoritarianism. They abandoned American democracy in its moment of maximum peril. They should never be allowed to forget it. They should wear that shame for the rest of their lives.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Agree. Federalism and the courts were essential.
Bryan Jones
@bdjones_jones
Yes and it should cause us political scientists to re-think our conceptions of democracy. For a starter: federalism is far more important than many scholars understood.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Remember when a huge portion of civil society including corporate leaders, top law firms, and elite universities decided to submit to Trump's authoritarianism? And it was legitimized because he won in 2024 with <checks notes> 49.7% of the popular vote? What now?
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Josh Madeiros
@cocoloba_
You can actually do this with far greater accuracy using just church bells. Can't quite believe it, but I tested it and it works.
Planes need a timestamp and unbroken sound. I tested 56 normal YouTube videos: 0 worked.
Bells need a few seconds. The bell ringers' database (Dove's Guide) logs every bell tower in England, 5,896 of them, all unique Hz, with the note of its biggest bell.
From a video with few seconds of the audio, the bells in a Derbyshire walk picked out the one church. I'm stunned.
0x45
@0x45o
this guy figured out how to leak your exact location from any video with sound using the Doppler curves of nearby airplanes
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nicholas Stephanopoulos
@ProfNickStephan
Assuming Democrats win 260 House seats and the national House vote by 13%, the House's efficiency gap would be around 3% in Republicans' favor.
A zero EG for the House would require around 274 Democratic seats. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Susan Collins formidable to the end
Gary Grumbach
@GaryGrumbach
Our @NBCNews graphics and data visualization teams have put together an in depth look at how last night's vote compares to Trump's ascendence to the presidency in 2024.
www.nbcnews.com -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nicholas Stephanopoulos
@ProfNickStephan
Another green shoot - a transfer of power without an effort to entrench the outgoing party. Could the bad old days of 2018 (when NC & WI stripped incoming governors of powers) really be behind us?
Greg Bluestein
@bluestein
Breaking: Gov. Brian Kemp just said he will *not* call a special legislative session to redraw Georgia’s political boundaries or limit the powers of Gov.-elect Keisha Lance Bottoms. @ajc #gapol
www.ajc.com/politics/2026/11/brian-kemp-rejects-special-s… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Again, fact that Trump *can* lose doesn't mean all is fine
-numerous judges had to block election interference
-required overcoming state/federal gerrymanders & Senate map tilted by malapportionment
In a strong democracy, these don't happen & opposition faces level playing field
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Following @otis_reid, here's a version of the Senate map looking at relative performance of candidates by Sanders endorsements (graph is by R vote so lower=better for Ds)
Matthew Yglesias
@mattyglesias
It’s great that all kinds of Democrats who El-Sayed and his supporters have spent years demeaning dug deep to put him over the top but I regret to inform you that quibbling about margins of victory is very relevant to assessing the viability of this style of politics.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Tom Wood
@thomasjwood
The Democratic party might infer something from the respective strength of their senate candidates to determine a new policy agenda.
The strongest battleground senate candidate: Talarico. The weakest: El Sayed.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Tom Wood
@thomasjwood
Just the state intercepts -- Collins was the only GOP candidate to run ahead of national conditions.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tom Wood
@thomasjwood
The Democratic party might infer something from the respective strength of their senate candidates to determine a new policy agenda.
The strongest battleground senate candidate: Talarico. The weakest: El Sayed. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tom Wood
@thomasjwood
Inspired by @BrendanNyhan's lovely shift plots - this one depicts the House and Senate shifts since '24 directly
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
OTOH the near-complete absence of election denial so far is a very encouraging sign for democracy (though, again, in a stable democracy no one would even notice)
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Again, fact that Trump *can* lose doesn't mean all is fine
-numerous judges had to block election interference
-required overcoming state/federal gerrymanders & Senate map tilted by malapportionment
In a strong democracy, these don't happen & opposition faces level playing field -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Important to watch for undemocratic post-election power-stripping
Greg Bluestein
@bluestein
Breaking: Gov. Brian Kemp just said he will *not* call a special legislative session to redraw Georgia’s political boundaries or limit the powers of Gov.-elect Keisha Lance Bottoms. @ajc #gapol
www.ajc.com/politics/2026/11/brian-kemp-rejects-special-s… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A blow against gerontocracy, but still pathetic that only 15% of senators are under 50. 50!
Ted Alcorn
@TedAlcorn
In races called, 17 senators are being replaced, already the biggest turnover in the chamber since 1980. The departing senators average age is 70, their replacements 55. The number of senators under 50 nearly doubles, from 8 to 15.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Fr. Chris Vorderbruggen
@FatherChrisVor1
I have been holding my tongue about something that happened during the Ohio gubernatorial race. I refused to speak while the election was underway because I did not want my words mistaken for an attempt to influence anyone’s vote.
But now that the election is over, I have something to say.
For nearly twenty years, my ministry has taken me among the homeless, the abused, the exploited, and the forgotten. I have sat with victims of sexual violence and witnessed the lifelong wounds of childhood trauma.
What Vivek Ramaswamy’s campaign did to Amy Acton was among the most morally repulsive things I have witnessed in American politics.
They took a woman whose stepfather was indicted for raping her when she was twelve years old and featured her mother, from whom she had been estranged for nearly fifty years, in a political advertisement attacking her credibility.
The advertisement challenged Acton’s account of her childhood without mentioning the sexual abuse allegations or the circumstances of that estrangement. When people expressed their horror, the campaign defended it.
Imagine examining another person’s childhood suffering and fractured family relationships, then deciding those wounds might help win an election.
Anyone who has worked with survivors of childhood sexual abuse understands the painful realities of denial, betrayal, shame, and estrangement. A child growing into a successful adult does not erase what happened. Nor does a parent’s denial make that suffering disappear.
I find it equally disturbing that an elderly mother, whom Acton herself described as unwell, was brought into this spectacle. Whatever her own struggles or responsibility for the past, exploiting that fractured relationship was grotesque.
To everyone who defended this, consider the message you sent to survivors. To the young woman whose family refuses to believe her. To the man who has carried the shame of childhood abuse for forty years. To the frightened child who has yet to tell anyone.
What did you teach them about the value of their suffering?
I am a priest. I believe in repentance, forgiveness, and redemption. But none of those things requires me to pretend that deliberate cruelty is anything less than a grave moral failure.
Your political allegiance does not absolve you of moral responsibility. You do not get to set aside your conscience during an election and retrieve it afterward.
If your loyalty requires you to defend the exploitation of a childhood sexual abuse survivor, your loyalty has become an idol. And if your conscience cannot recognize the wrong here, examine what political tribalism has done to it.
This is not about Republicans or Democrats. We have a choice as a people: continue excusing cruelty from our preferred political camp, or summon the courage to condemn it wherever it appears.
Amy Acton deserved better. Every survivor who watched this unfold deserved better.
And God help us if we have reached the point where winning an election matters more than protecting the dignity of a wounded human being. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nicholas Stephanopoulos
@ProfNickStephan
Look how easy it is to break a gerrymander! All you need is the biggest wave since Watergate. Child’s play!
Jason Willick
@jawillick
Vindication for John Roberts in Rucho—gerrymanders aren’t inexorable and voters have ways of regulating them.
David Weigel
@daveweigel
Looks like Dems will end up with 9 North Carolina seats on a map gerrymandered to reduce them to 3 seats.
Jon Michael Raasch
@JMRaasch
From a GOP strategist:
“The NRCC chair is about to lose his race. Couldn’t be a worse night for Republicans than that.”
“It’s all James Blair, LaCivita’s, and Chris Grant’s fault. 100%”
Congressman Rich Hudson, NRCC boss, is currently down to his Democratic challenger -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
G Elliott Morris
@gelliottmorris
New post: Democrats win U.S. House, Senate in biggest incumbent-party defeat since Watergate. The Strength In Numbers/Verasight poll was tied for first in predicting the Democratic House blowout
www.gelliottmorris.com/p/democrats-win-us-house-senate-in -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Justin Slaughter
@JBSDC
There was also no mass prediction markets miss. In being told that, of 487 prediction markets on individual elections, 479 of the candidates that were leading at 5 PM last night won.
479/489. That’s a 98% accuracy rate before ANY votes are counted.
Justin Slaughter
@JBSDC
There's no mass polling miss.
Brace. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sam Rosenfeld
@sam_rosenfeld
This is a level of thumpitude that genuinely, finally, at long last, should get Repubs to break omerta over Trump and fight amongst themselves over issues and ideology and the '28 nomination, etc--to do normal politics! Would be an important development for American democracy.
Joe Weisenthal
@TheStalwart
There have been midterm bloodbaths before. And there have been unpopular albatross lame duck Presidents before. But I genuinely don't have any mental model for how 2028 GOP hopefuls (at any level) will talk to/about Trump.
This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kris Gulati
@krisgulati
New paper with @ProkopetsJake and Raymond Kim!!
Every scientific paper has TWO teams.
One is listed on the title page.
The other is invisible: colleagues who challenge ideas, catch mistakes, and help make the research better.
Our new paper suggests AI may be making that invisible team smaller.🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
What uniform swing looks like - Claude Code graphs of 2024 presidential vote vs. 2026 Senate/House vote totals
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Ecological inference for now but suggestive www.economist.com/interactive/graphic-detail/2026/11/04/t…
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
also
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Important on why a lame duck SCOTUS replacement would be a heavy lift for Trump
Ed Whelan
@EdWhelanEPPC
On judicial nominations, massive Democrat takeover of Senate means:
1. Trump will not be able to appoint any justices or judges during his last two years in office (absent one-sided deal in favor of Dems).
2. If a Republican is elected president in 2028, that president will not be able to appoint any justices or judges until Republicans regain the Senate. Given the deep hole, that might not happen until 2033.
3. I'd be extremely surprised if any justice announces resignation now (for end of term) in hope of lame-duck confirmation of successor. Among other things, it's highly doubtful that Rs would have the vote to confirm. If Sherrod Brown wins Ohio seat, he gets sworn into Senate in early December, so Senate becomes 52R-48D. Collins and Murkowski would be unlikely to support anyone Trump would nominate. John Cornyn, Thom Tillis, and Bill Cassidy aren’t going to be eager to do Trump any favors. And there are other senators—e.g., John Curtis of Utah—who would be leery of anyone Trump would be eager to push through. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ryan Matsumoto
@ryanmatsumoto1
Looks like the polls underestimated Democrats in 14 of 15 key Senate races, with an average polling error of 4.4 points.
Polls were the most accurate in Michigan, Kansas, Maine, and Ohio.
The biggest polling misses were in North Carolina, Georgia, and Texas. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Lenny Bronner
@lennybronner
Todays story is that Democrats showed up as if this were a presidential election... and Republicans didn't
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dan Hopkins
@dhopkins1776
What does this mean for the GOP? 1 challenge with wave elections is that it's a party's more moderate members (eg PA-1) who go down. The resulting minority caucus is even more ideologically extreme than the outgoing majority.
House GOP NOMINATE scores by outcome so far. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
My question this AM
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Remember when a huge portion of civil society including corporate leaders, top law firms, and elite universities decided to submit to Trump's authoritarianism? And it was legitimized because he won in 2024 with <checks notes> 49.7% of the popular vote? What now?
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Reminder that tonight was only possible because judges repeatedly blocked efforts to undermine the elections. In a stable democracy, you don't need the legal system to prevent the President from subverting the result.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Reminder that tonight was only possible because judges repeatedly blocked efforts to undermine the elections. In a stable democracy, you don't need the legal system to prevent the President from subverting the result.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Important. Think about the caucus meetings.
Matt Glassman
@MattGlassman312
One thing to keep in mind is that House majorities that get clobbered tend to, at first, learn the wrong lesson, because the seats they lose are the moderate ones, and the remaining caucus becomes *more* extreme. A House GOP down to 180 gets way more conservative, and likely elects leadership that reflects that.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Gerard Baker
@gerardtbaker
The simultaneously frustrating and uplifting thing about tonight is that it has taken ordinary, decent American voters to do what GOP leaders and elected officials declined to do for two years: Say, "No. We will not participate in this continuing act of self-harm". Thank God for democracy but a little backbone and a little decency on the part of Republicans could have saved them and us from this. #midterms26
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Margaret Brennan
@margbrennan
Senator Tillis (R-NC) warns Secretary Hegseth that he's coming for him in the lame duck.
TILLIS on CBS: "....But Pete Hegseth has a reckoning. Uh, he, he is one of the reasons why we don't have the funding or guarantee for defense that we need. I think after the election, we need to determine whether or not he's the best person to bring those needs to congress. If he wants something done before the end of this congress, which is only 60 days away." -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
David Wallace-Wells
@dwallacewells
The 2010 margin was 6.8 points, an election remembered as a generational wipeout. Tonight is expected to almost double that. 2008 was 10.5 and 2006 was 8.0. A thirteen point win would be the biggest congressional margin in fifty years.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ence Morse
@ccemorse
It is 11 PM.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dave Itzkoff
@ditzkoff
Fox News showing Swan Lake on a loop
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Texas Democrats rn
Christopher Hooks
@cd_hooks
i want you to put the word out that we back up
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
John Mitnick
@JohnMitnick
This is a real concern, as is the possibility of efforts to negate election results.
Camille MacKenzie
@CamRMacKenzie
After the great news tonight as it finally checks Trump … now comes the danger in how he’s going to react for the next two months. Does he nuke Iran? Does he go ballistic and tear down more buildings? Does he have a stroke? He’s a completely volatile, loose canon with dementia and no way to know how he’ll react to his second biggest loss on the world stage since 2020.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Let's see if this holds but remarkable
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Which closing message was best?
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Egger
@EggerDC
The margins are so enormous here that whatever stop-the-steal plans they may have wanted to try are just DOA. Literally too big to rig.
Mary Margaret Olohan
@MaryMargOlohan
At this point in the evening, White House official tells me: "Republicans are going to run through the tape and win as many races as possible."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Ossoff and DeSantis up, Vance down
Luana Lopes Lara
@luanalopeslara
tracking in real time how the midterms impact the 2028 odds is magical ✨✨🥹🥹
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Lakshya Jain
@lxeagle17
Turn your eyes to Colorado for a second. I think Lauren Boebert is in real danger of losing. She is massively underperforming in this Trump +18 seat and missing her benchmarks by a few points.
If she loses, it would be a seismic upset. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Lakshya Jain
@lxeagle17
This has now happened, and it'll be interesting to see what happens when the dust settles. Worst election any party has seen since 1982 and maybe even since Watergate.
Lakshya Jain
@lxeagle17
Trump has never faced a clear voter repudiation in a way that would make Republican politicians abandon him.
If 2026 is as blue as polls say, you'll see a lot of Republican elected officials suddenly change their tune, because it'd be clear he's no help any longer. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
what a graph
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sean T at RCP
@SeanTrende
If you are watching this and just have a passing familiarity with elections, just understand you're seeing something special, a 100-year flood. Easily the worst election for a party since 1974.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
New voters have not been alive for an election like this
ethan winter
@EthanBWinter
Entering totally new terrain. Trump has never been defeated like this.
- 2016: Trump wins
- 2018: Republicans gain Senate seats
- 2020: Biden wins but underperforms polls , D.s lose
House seats
- 2022: Rs. flip the House
- 2024: Trump wins
- 2026: VICTORY ON ALL FRONTS -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Barber 🔰🌐
@abarber1
0/6 is actually hard to do. He’s like a Bizarro World @NateSilver538.
Keith Rabois
@rabois
i want to bet you on Texas, Iowa, Kansas, Alaska, Michigan and Ohio.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Dogs that haven't barked (thankfully):
-No major Election Day interference by DOJ or federal agents
-No Trump claims of fraud or election-rigging so far -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Hugo Lowell
@hugolowell
Of course, this comes after Trump explicitly told his base to pretend that he was on the ballot
Hugo Lowell
@hugolowell
NEW: White House official tells me their position is GOP losses aren’t Trump’s fault
“White House mood is defiant. President Trump was not on the ballot, and it's proven that Trump voters show up when he is on the ballot.
“Also, candidate quality matters and President Trump did a better job promoting many of the candidates than the candidates did themselves. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Steve Morris
@stevemorris__
Incredibly prescient description from Nate Cohn last week, and really no better way to put it: The beginning of a new chapter in American politics.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Democrats could potentially win the largest House majority since 2008
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"A GOP operative compared the start of the night to the first 25 minutes of 'Saving Private Ryan.'"
Dave Brown
@dave_brown24
A Republican operative helping oversee Senate races said the floor had fallen out in races across the country. One Republican strategist overseeing a race sent a photo of a whiskey bottle. A GOP operative compared the start of the night to the first 25 minutes of "Saving Private Ryan."
www.wsj.com/livecoverage/midterm-election-results-2026-la… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The question now: is this a 2006-level reversal, a 1974-level Watergate-level earthquake, or something in between? Claude graph of largest House popular vote margins since 1972
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Kalshi: Ds 64% for *254+* House seats, 47% for *258+*
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kate Aronoff
@KateAronoff
Seems like Bo French might have helped elect the first Democrat to the Texas Railroad Commission in 36 years
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Correcting my prior post: here are the (less visible) Kalshi markets for how many seats Ds will hold: 76% chance at least 55, 52% chance at least 56 (!)
James Surowiecki
@JamesSurowiecki
That's not true. That was one binary market (above or below 52). They had markets for each of the specific numbers between 47 and 54, and for 54+.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
KY-06 is 99.9% D on Kalshi. An R+15 district.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Still early but this is potentially monumental. Ken Paxton is a bad candidate; Greg Abbott isn't. If Democrats can win Texas, the Republican Electoral College advantage inverts
-
Post not available here, view on X
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Biennial reminder that exit polls can't tell us why people voted the way they did - from my 2014 @NYTupshot column
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Seems like many (most?) outdoor images and video are going to leak precise locations in a world with ubiquitous AI absent post-processing or other countermeasures
0x45
@0x45o
this guy figured out how to leak your exact location from any video with sound using the Doppler curves of nearby airplanes
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paras Chopra
@paraschopra
I replicated this with Claude Code!
It took a live webcam 7 km south of Southampton, UK and predicted its location purely from doppler effect in audio spectrogram + open-source aircraft location data.
Lo and behold, the predicted location (heatmap) peaks at actual location where webcam is situated.
It's crazy that this works at all, and crazier still I replicated it within a few hours.
0x45
@0x45o
this guy figured out how to leak your exact location from any video with sound using the Doppler curves of nearby airplanes
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Kids: study causal inference. I wouldn't accept this from an undergraduate in my intro stats class.
Tamar Haspel
@TamarHaspel
Holy correlation Batman! @CNN is reporting this as causal: voting makes you live longer.
www.cnn.com/2026/11/02/health/voting-live-longer-study-we…
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
We are in an era of flux & fighting over how AI should be used to review academic work
But it is worth proving to yourself that yes, AI used properly can be quite insightful at this, by asking GPT-6 Pro/Opus 5.5 High to give helpful, substantive feedback on one of your papers. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Beth Anne Helgason
@BA_Helgason
Today, we’re releasing a dataset of >2k LLM conversations with real voters about the 2026 Midterms. We wanted to know how the ~millions of American who use AI models to talk about elections actually prompt them. Our hope is to use this data to build better evals for LLMs, to prepare for a world where most voters use AI to think through election decisions.
We observe that:
- People mostly want basic information
- Multi-turn chats follow predictable patterns
- Models differ widely in whether they’ll recommend candidates -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Michael Lin, MD PhD 🧬
@michaelzlin
Very informative technical explanation of why small wings leave little margin of error when you hit bad weather
I don't think I've seen a plane with a smaller wing-to-body size ratio than the Learjet 60, except maybe the F-104 and well you can look up the nickname for that one
Dave
@HangryDBowman
Long post worth the read. I have a close friend who flies private jets. He flew Lear 60 for over 15 years and now flies a Gulfstream. This is a copy and paste of his response to me asking him if he thought Ice was the issue.
“We always fly in ridiculous cold temperatures.”
“Usually you get the bad ice when the temperatures are like 0 to negative 20. Believe it or not, there's a point where it gets too cold for ice. At negative 40, you pretty much can't get ice. Besides that, the airplane is equipped with hot wings and hot tail.”
“I flew the Lear 60 for like 15 years. It's a no joke airplane and probably the most dangerous airplane I've ever flown. It has a super critical wing. The airflow has to be perfect. When it gets into rough air and turbulence, it will easily stall. It has stalled on me at 41,000 feet just because we got into some turbulence. (On the way to Palm Beach as a matter of fact. ) we recovered and got her flying again around 30,000 ft. Scary
But here's the problem. This airplane has a terrible coffin corner. (At high altitudes) what that means is, you fly in a really narrow window where if you lose 20 knots, it stalls. And if you gain 20 knots, it goes into overspeed. The air goes supersonic over the wing and it goes into a high speed stall, or worse, a Mach tuck.
So if you stall, the normal recovery is to push the yoke forward to lower the nose and gain air speed again. At low altitude, you can push really hard and recover quickly. But in the 60... if you push hard, it will gain air speed and go into a high speed stall, it doesn't take much. It's very sensitive. The 60 has super short wings that don't make much lift. So then you high speed stall, and the computer is yelling at you... OVERSPEED! So then you pull back and enter a low speed stall. And you go back and forth trying to keep it in the air speed window.
This whole thing is pretty scary. And you should recover basically by using your finger tips using super small corrections. You can't over do it. If you try to muscle the recovery, she gets mad and will become unrecoverable. If she enters a mach tuck. You're done.”
“I taught the Lear 60 to other pilots and mentored them. I told them... this airplane has a soul and it's trying to kill you. Do not ever try to top the weather. Don't go over a thunderstorm. Go around it. If it means a half hour longer, then do it. Going over the thunderstorm will go into turbulence that you might not recover from.
These pilots went right over a thunderstorm. I think it stalled which caused them to drop down into the thunderstorm and ended up in mach tuck. They picked up so much speed that it ripped off the horizontal stabilizer. At that point... chuck Yeager can't fix it.
These same pilots did something similar a couple months ago. ATC got mad at them. They poked the bear one too many times.
This is not a beginner airplane. The Gulfstream would never do this. It has HUGE margins of error and won't stall like that. But then again... it's got HUGE wings. It's a pussy cat.
The Gulfstream is like driving a 4 wheeler. The Lear 60 is like riding a uni cycle.” -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Daniel Litt
@littmath
The future of peer review?
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Previewing every post starting at 6 PM: Stay in line!
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Liz Landers
@ElizLanders
News Mexico’s Secretary of State tells me that verified DOJ monitors have been asking “some unusual questions and doing things that are not allowed by law such as questioning voters” in Bernalillo County. The clerk has reached out to DOJ to alert them and to fix the issue.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
What (would-be) competitive authoritarianism looks like
This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Reed Schwartz
@reedschwartzsf
Wesleyan is now cheaper than UConn on average:
The Wall Street Journal
@WSJ
A growing group of colleges long known for eye-watering price tags is turning out to be quite the bargain for many students—and the schools want you to know it. www.wsj.com/us-news/education/the-best-college-bargain-is…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Justin Baragona
@justinbaragona
"If I have to create stories so that the American media actually pays attention to the suffering of the American people, then that's what I'm going to do," JD Vance, defending his amplification of the "Haitians are eating Springfield's pets" hoax in 2024.
Acyn
@Acyn
Vance: I saw a story that there are illegal communities of illegal immigrant Somalis who are coordinating to flood the polls and drown out your vote. We want American elections to belong to American citizens. Send a message by voting for Vivek for governor.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Daniel Heinen
@heinenbros
The birds in your videos might be snitching on you.
A lot of people asked for my take on this demo. It has limitations, but we ran the same audio through AI bird identification.
Overlapping their modeled ranges cut the U.S. candidate search area by ~75%.
0x45
@0x45o
this guy figured out how to leak your exact location from any video with sound using the Doppler curves of nearby airplanes
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
(((Charles Fishman))) 💧
@cfishman
This crazy.
The largest theft in history — last week's Binance theft of $6,300,000,000 ($6.3 billion) — began with a pair of fake job interviews with a Binance engineer who was looking for a new job.
He thought the video interviews were real. That's how he got hacked.
How in the world do we protect against the crazy new world we've invented?
↘️ ↘️ ↘️
AB Kuai.Dong
@_FORAB
币安刚刚发布了初步调查报告,梳理了他们是如何被朝鲜黑客,盗取了 60 多亿美金。
某位币安软件工程师,在求职网站上挂着简历,结果一位自称项目总监的人,主动联系他,说有个潜在岗位想聊聊。
对方安排了一位 HR 跟他视频面试,真人露脸。为了让他放下戒心,这位 HR 还把自己的社交账号发给了他,上面有他本人的生活照和视频,看着就像真实存在的人。
然后到了第二轮面试,对方换到一个常用的聊天软件上,跟一位自称是这家公司 CEO 的人聊,但其实公司是假的。
等面试快结束时,对方让他下载一份代码题,作为技术测试,结果这位币安员工,在工作电脑上下载运行了它。
这份代码,在安装环节偷偷自动执行,把电脑上一个正常的开发工具 pyenv,换成了带木马的版本,然后跟黑客的服务器连上,还会自己清理痕迹。
就连币安当时的杀毒软件,也没查出来。
从这台电脑开始,朝鲜黑客一步步渗透内部,摸进了交易所的钱包系统,一个多月后币安被盗了 60 多亿美金。
关键是现在领英上,还有一群币安在职员工的信息,这些估计都是朝鲜黑客的潜在欺诈对象。 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Masley
@AndyMasley
Gotta say as a former teacher, while I understand this might be due to the technical challenge of a completely separate search, having a "do all my work for me"-machine on every computer is just so apocalyptic for learning that I'd hope Google would just make the effort anyway
rat king 🐀
@MikeIsaac
google says if you dont want AI in the classroom you wont get search either
earnest Q: highlighted quote seems carefully crafted to suggest there is a technical reason google could not turn off AI in search.
is there *any* truth to that?
www.nytimes.com/2026/11/03/us/google-search-schools-restr… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Bruce Mehlman
@bpmehlman
Brilliant & super-helpful visualization of @nytimes’ analysis of states’ speed counting votes by @axios @apantazi. I love a great chart!
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sol Messing
@SolomonMg
This work is simple, clear, compelling, but embellished. I first found it shocking, the notion that a model can natively understand neural network weights and productively edit them! That's really not what's happening, & many here are annoyed. BUT, it's still notable & worth understanding.
I've done a bit of analysis & auditing, here are some thoughts:
1/ This paper does *not* show an agent can edit its weights ala RSI. It does *not* show a model can edit in new capabilities. It DOES show that a frontier model can remove "bugs" (reasoning loops), using the same ablation technique researchers have used to remove safety refusals (Exp 8 - Qwen). So yes, renewed concern about frontier LLMs getting access to their own weights & editing safety/CBRNE/alignment features.
2/ This paper also does *not* show that an agent can "natively" read, deeply understand, and fix model weights. What is happening is that the model can cleverly deduce the same things that mech-inturp researchers do and fix them in mostly ham-fisted ways. Still cool, but not as groundbreaking. Evidence in thread below...
Rikiya Takehi
@rikiyatakehi
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math.
It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training.
We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵
Blog: rikiyat.github.io/blog/weight-space-programming/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"Losing a job raises the risk of homelessness by 50 percent. Getting a shelter bed at the first request cuts job loss by 26 percent for those still working and raises employment a year later by 58 percent for the unemployed."
Alexandre Gaillard
@ag_Gaillard
Amazing JMP by Johanna Roth (Sciences Po). She is the first to combine nationwide shelter data with employment records to show that job loss causes homelessness, which in turn reduces the probability of finding a job, creating a feedback loop.
github.com/JojoRoth/JMP-Draft/raw/main/Roth_JMP.pdf -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew M. Webb
@AndrewM_Webb
Asked Claude to replicate this and it works 😬 though you do need contiguous audio and how much you need is very location dependent (🧵)
0x45
@0x45o
this guy figured out how to leak your exact location from any video with sound using the Doppler curves of nearby airplanes
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Really interesting thread/project
Christian Vedel
@ChristianVedel
My job market paper investigates exactly such a freak natural experiment. In February 1825, a storm broke through a narrow isthmus in north-western Denmark. Suddenly, a bunch of places became coastal, with access to the North Sea.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
I see every day what AI enables and also what it undermines
Jessica Hullman
@JessicaHullman
I wish everyone could experience firsthand what we've been seeing in AI/ML, at the scale we've been seeing it, before they weigh in on how obviously better science is going to be now.
This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Osita Nwanevu
@OsitaNwanevu
She's made off with the Mackinac Bridge! But she's left this clue for you, gumshoe: All the Great Lakes might be great, but this one's only the fourth largest of them by surface area.
Politics & Poll Tracker 📡
@PollTracker2024
NewsNation: Haley Stevens casts her ballot on Election Day.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Come on now - a book author interview where the answers are 100% AI?
"The opposite of invisibility isn’t attention. It’s significance."
What are we doing here?
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A partial equilibrium result. Once agents like this are besieging customer service, companies are going to start blocking agents or not offering humans all together.
Peter Yang
@petergyang
Grok @Bot can now call US businesses for you, and it's one of my favorite AI features ever.
Here's my new video where I play a real call of my bot successfully negotiating with Comcast to cut my internet bill by almost $500 a year.
Some highlights include my bot:
→ Responding to "How's the weather?"
→ Turning down 3 upsells from the rep
→ Pushing the rep with a rival's lower offer
📌 Watch now: youtu.be/F-356TczoQU -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Eva Vivalt
@evavivalt
The promised forward-looking competition to forecast research results, with 1M+ in prizes committed. 500k+ in the first round:
Eva Vivalt
@evavivalt
New working paper: Can AI forecast treatment effects in economics from pre-registration documents?
Answer: Yes. 1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kyle Cheney
@kyledcheney
NEW: Donald Trump pardoned Cleveland Meredith for threatening to "put a bullet" in Nancy Pelosi's head. Yesterday, the FBI arrested Meredith on new charges for threatening to execute the judges who tried Jan. 6. cases.
www.politico.com/news/2026/11/03/jan-6-defendant-federal-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Arpit Gupta
@arpitrage
Amazing new paper by Chetty and the Opp Insights team — digital twins can recover causal effects
They validate the digital twins (ie rich ML applied to generate counterfactuals from lots of covariates) with RCTs, allowing for scalable program evaluation
opportunityinsights.org/wp-content/uploads/2026/09/Digita… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Thomas Renault
@captaineco_fr
🚨New working paper : Can AI scale human fact-checking? (Spoiler: Yes, but with three important caveats)
X recently introduced AI Note Writers on Community Notes, allowing AI systems to propose fact-checks that are then evaluated by human contributors. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Isaac Saul
@Ike_Saul
The Trump administration has tasked hundreds of DHS agents with finding noncitizen voters, scouring state voter rolls, interviewing people at their homes, etc.
Spent millions of taxpayer dollars.
So far, they've found 236 prosecutable cases.
Amid 158,000,000 ballots cast in 2024. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Love Christensen
@lovecsen
We find that Democratic counters have little effect and Republican counters have modest effects. But the Joe Rogan/Tucker Carlson counter almost halves following and significantly lowers issue-specific and overall Trump approval.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Love Christensen
@lovecsen
How much sway does Trump have over Republican voters?
In a new paper forthcoming in @poqjournal , @MAgerberg and I use the Epstein files case to show that even MAGA Republicans are willing to hold Trump accountable when he takes positions they oppose. 🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sam Rosenfeld
@sam_rosenfeld
👏INFRA👏STRUCTURE👏WEEK👏
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"Mr. Trump used his final in-person appearance before Americans head to the polls to talk about drain pipes, top soil, limestone, fountains and his favorite kind of columns"
Daryl G Kimball
@DarylGKimball
Just bizarre. "Trump Delivers a Closing Midterm Message. Mostly About Landscaping." www.nytimes.com/2026/11/02/us/politics/trump-reflecting-p…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Relevant to the endless popularism debates - cc @MatthewYglesia1 and critics
Toni Rodon
@tonirodon
Parties shifting towards more popular positions are more likely to be seen as opportunistic and vote-seeking, whereas shifts towards less popular positions are more likely to be seen as principled and sincere journals.sagepub.com/doi/full/10.1177/00104140261500267?m…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Love Christensen
@lovecsen
Just out!
Why is accommodation so difficult?
Using experiments with hypothetical parties and a real case of mainstream accommodation of the far right, Pablo Fernandez-Vazquez and I show that parties that accommodate are perceived as less credible and more opportunistic.
Toni Rodon
@tonirodon
Parties shifting towards more popular positions are more likely to be seen as opportunistic and vote-seeking, whereas shifts towards less popular positions are more likely to be seen as principled and sincere journals.sagepub.com/doi/full/10.1177/00104140261500267?m…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
alz
@alz_zyd_
Really cool JMP on tipping. Workers seem to benefit from tips! Firms lower wages, but only by 23-33% of tip revenue. Tipped workers seem to work harder, serving more customers in the same amount of time.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
How does this compare to 2022 or 2024?
Scott MacFarlane
@MacFarlaneNews
Justice Dept releases list of where it is sending 1,000 election monitors
Many, many, many are heavily Democratic districts:
Alabama
•Jefferson
•Montgomery
•Mobile
Alaska
•Anchorage
•Dillingham
•Kusilvak
Arizona
•Maricopa
•Pima
•Apache
•Navajo
Arkansas
•Pulaski
•Washington
California
•Los Angeles
•Alameda
•Sacramento
•San Diego
Colorado
•Adams
Connecticut
•New Haven
Delaware
•New Castle
Florida
•Leon
•Hillsborough
•Miami Dade
•Broward
•Palm Beach
Georgia
•Cobb
•DeKalb
•Fulton
•Gwinnett
•Chatham
•Bibb
Hawaii
•Honolulu
Idaho
•Ada
Illinois
•Chicago
•Cook
•Champaign
•Sangamon
•St. Clair
•Madison
Indiana
•Lake
•Marion
Iowa
•Linn
•Polk
Kansas
•Johnson
•Wyandotte
•Sedgwick
Kentucky
•Fayette
•Jefferson
Louisiana
•Orleans
•East Baton Rouge
•Lafayette
Maine
•Lewiston
•Portland
•South Portland
Maryland
•Prince Georges
Massachusetts
•Lowell
•Boston
Michigan
•Lansing
•E. Lansing
•Grand Rapids
•Flint
•Detroit
•Hamtramck
•Warren
•Dearborn
Minnesota
•Minneapolis
•Hennepin
•Ramsey
Mississippi
•Lafayette
•Hinds
Missouri
•St. Louis
•Jackson
•Clay
Montana
•Yellowstone
Nebraska
•Douglas
Nevada
•Clark
New Hampshire
•Manchester
•Nashua
New Jersey
•Middlesex
•Bergen
•Union
New Mexico
•Bernalillo
New York
•New York
•Westchester
•Kings
•Albany
•Erie
North Carolina
•Alamance
•Mecklenburg
•Wake
North Dakota
•Cass
Ohio
•Cuyahoga
•Franklin
•Medina
Oklahoma
•Muskogee
•Tulsa
•Oklahoma
Oregon
•Multnomah
Pennsylvania
•Philadelphia
•Allegheny
•Luzerne
•Lehigh
Rhode Island
•Pawtucket
South Carolina
•Charleston
South Dakota
•Minnehaha
Tennessee
•Davidson
•Knox
•Shelby
Texas
•Bexar
•Dallas
•Harris
•Waller
•Blanco
•Bowie
•Nueces
Utah
•Salt Lake
Vermont
•Burlington
Virginia
•Charlottesville
•Fairfax
•Prince William
•Virginia Beach
Washington
•King
•Yakima
West Virginia
•Kanawha
Wisconsin
•Madison
•Milwaukee
Wyoming
•Laramie County -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matija Franklin
@FranklinMatija
We ran a five day longitudinal RCT where 992 participants engaged daily with LLMs on relationship advice. We compared a non-personalised model to models that were personalised either via survey or memory.
We found that repeated exposure across conditions increased perceptions of closeness, usefulness and competence. We also found that different approaches to personalisation led to differences in people's receptivity to advice and their information-sharing attitudes and behaviours
Full paper is here: arxiv.org/abs/2609.20077
Thank you to my co-authors @canfer_akbulut @Arianna_Manzini Justine Breuch @lujainmibrahim @996roma @IasonGabriel @KLdivergence @weidingerlaura -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jack Fitzgerald | @jackfitzgerald.bsky.social
@FitzgeraldJack_
I'm on the academic job market this year! In that spirit, I've updated my job market paper, "The Need for Equivalence Testing in Economics." TLDR: null results in economics are even less robust to lenient equivalent tests than previously thought. 1/x
doi.org/10.31222/osf.io/d7sqr_v3 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nathan Hoffmann
@nathanihoff
I'm delighted to share a new article that Peter Catron, Mingze Li, and I have just published in @ASR_Journal. Using web scraping, linked censuses, and machine learning, we find that citizenship benefits immigrants least likely to naturalize. 🧵👇
doi.org/10.1177/00031224261489725 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Based on phone book digitization for historical business data. We have to keep raising our aspirations for social science in the age of AI.
Nicholas Decker
@captgouda24
Riots in the '60s really did screw up neighborhoods permanently. Building on astonishing data, this paper shows that large negative shocks are sufficient to tip cities into new equilibriums altogether, while small shocks can be recovered from. 1/
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Michael S. Schmidt
@nytmike
NEW and BREAKING: Another no true bill in Washington. This one on a voter fraud case against someone who voted twice in 2024. Shows struggles by prosecutors in Washington to bring the cases Trump wants. And shows DOJ's harried push to prove Trump’s unsubstantiated claims of rampant voter fraud. www.nytimes.com/2026/11/02/us/politics/grand-jury-justice…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Trump "made some of his usual comments about the war with Iran, the US economy and various culture war issues, but he did not even attempt to explain how any of the renovation projects that were the focus of the speech related to any of the issues animating the midterm elections"
Daryl G Kimball
@DarylGKimball
Just bizarre. "Trump Delivers a Closing Midterm Message. Mostly About Landscaping." www.nytimes.com/2026/11/02/us/politics/trump-reflecting-p…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"Mr. Trump used his final in-person appearance before Americans head to the polls to talk about drain pipes, top soil, limestone, fountains and his favorite kind of columns"
Daryl G Kimball
@DarylGKimball
Just bizarre. "Trump Delivers a Closing Midterm Message. Mostly About Landscaping." www.nytimes.com/2026/11/02/us/politics/trump-reflecting-p…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matt Bennett
@ThirdWayMattB
As someone who has sat through roughly 1.2 million WH scheduling meetings, I simply cannot believe that this made the cut on the eve of the election. (Or ever, really.) I know the boss is senile and insane, but this is just breathtakingly bad staff work.
FactPost
@factpostnews
The day before Election Day, Trump does a show and tell with pictures of the inside of pipes
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
$116
Prem Thakker
@prem_thakker
This is… insane:
“Musk has spent $474 million to help Trump and the GOP win the midterms…roughly 0.05% of his net worth.”
“If you compare his worth to the median American’s, Musk’s political spending this election cycle is equivalent to $116.”
zeteo.com/p/elon-musk-474-million -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joshua Achiam
@jachiam0
we're not prepared for a future where the surveillance state can be optimized like speedrunners doing Mario 64. good lord
deckard
@slimer48484
I am in awe. This guy discovered a way to locate any audio clip by matching aircraft doppler curves against flight data. INSANE.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Uri Simonsohn
@uri_sohn
A new document was just posted on the Gino Case. It's 48 pages long. I asked Claude to summarize it (See below).
The counterclaim of defamation by Harvard against Gino is still pending. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sean Marotta
@smmarotta
A Solicitor General filing used to be a thing of such gravitas, and treated with such respect, that it would be unthinkable and almost disrespectful to deny it without seeing what the other side had to say for itself.
A lot of things have changed.
James Romoser
@jamesromoser
In a vacuum, an emergency election-related request from the federal government is the sort of thing that SCOTUS typically would take very seriously. A terse, 9-0 order denying such a request, without even getting the other side's response, is the court's way of sending a message.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
James Romoser
@jamesromoser
In a vacuum, an emergency election-related request from the federal government is the sort of thing that SCOTUS typically would take very seriously. A terse, 9-0 order denying such a request, without even getting the other side's response, is the court's way of sending a message.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Very cool idea
Yamil Ricardo Velez
@YamilRVelez
A couple of weeks ago, @RobbWiller mentioned creating a public resource that generates automated research reports as a way of addressing publication bias.
After many rounds of tinkering, I’ve created a repository for my own site, but I'm opening it up to anyone who wants to contribute. filedrawer.org is an agent-friendly platform that takes an existing project, creates a report, runs a light pass that checks every reported estimate against the results and every analysis against the pre-analysis plan, and suggests new research designs with downloadable QSFs so you or other scholars can extend the work. Reports can even obtain a DOI!
You host all data and scripts on GitHub, with filedrawer.org only hosting metadata from your public repo. It checks locally for PII before you post, lets you import outside reviews, and you can use either local open-weight models or proprietary models via @OpenRouter.
I’m committing to host it, as I already have several projects there, and thanks to GitHub, hosting the metadata is very cheap. No individual-level data reach filedrawer.org or the models: they see only the survey schema, the plan, and aggregate results.
Right now, it handles survey experiments, conjoints and methods comparisons, since that’s my wheelhouse. It’s open source, so if your work is observational, qualitative or field-based, I’d love help extending it.This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ryan Goodman
@rgoodlaw
❗What you call game over.
DOJ admits in court their list of alleged 20k noncitizens registered to vote in Cali has a HIGH ERROR RATE.
Case involves TRO barring DOJ threats to prosecute state election officials.
DOJ admission will have reverberating effects in other cases. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kyle Cheney
@kyledcheney
Judge Kasubhai says ICE has lost the presumption of rgularity that courts have long afforded to government agencies. storage.courtlistener.com/recap/gov.uscourts.ord.190035/g…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kyle Cheney
@kyledcheney
JUST IN: A federal judge is encouraging his colleagues to call ICE officers in for live testimony on the witness stand, concluding that without direct scrutiny and cross-examination, routine lies will go unpunished.
storage.courtlistener.com/recap/gov.uscourts.ord.190035/g… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
James Zou
@james_y_zou
How long does a finding about LLMs stay true?
We re-tested 50 widely cited findings about LLM behavior (various biases, hallucination, etc). Within 15 months, most no longer held or had substantially reduced effects.
LLM findings has short shelf life zou-group.github.io/llm-claims-shelf-life/
Great job by @oq_35!
Owen Queen
@oq_35
1/n Within 15 months of publication, most of the LLM claims we tested either no longer held or showed substantially diminished effects. How long does our knowledge of AI hold up? In our new study, we revisit 50 findings to see what fades, what persists, and what gets stronger. 🧵
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Hallucinated citations are not a thing anymore
Owen Queen
@oq_35
2/n Take fabricated citations. A 2023 paper found that 55% of ChatGPT's references were invented. Using the same literature-review prompts, we measured about 1% for Claude Sonnet 5.5 and 0% for GPT-6.1 Sol. This once-familiar failure has become much rarer in newer models.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Meg Schwenzfeier
@_Schwenzfeier
Kudos to @ballotpedia for making a quick update to allow ChatGPT to cite their content, right as undecided voters are tuning in.
Ballotpedia is a critical resource, especially for downballot races -- and now it's back into a third of ChatGPT election answers. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Arvind Narayanan
@random_walker
As AI labs race toward Recursive Self Improvement, there’s a possibility that what they are actually racing toward is Recursive Self Extinction — of their business model. Here’s my reasoning.
1. Progress on verifiable tasks continues. In particular, improving the cost efficiency of existing capabilities is a highly verifiable task. For the past few years, cost per equivalent capability has dropped by at least 10x per year. I think this progress will be sustained for a few more years, and even accelerate as labs use AI more and more heavily for AI research in the pursuit of RSI. There is of course a ceiling to efficiency improvements, but I don’t think we’re close to that ceiling.
2. Progress on most unverifiable tasks remains frustratingly hard. AI inherits the same limitations (ambiguous success criteria, slow and noisy feedback) that limit human performance at tasks ranging from writing to governance. The optimization paradigm generally doesn’t result in superhuman capabilities on such tasks, because whatever proxy you optimize for gets Goodharted. For example, I recently pointed out how over-optimization explains Claude’s overly dense and ornamental writing style that’s since been dialed back.
3. RSI ends up being narrow. A consequence of the above two points — even if some version of RSI is achieved, it results in an explosion in some dimensions (efficiency, verifiable tasks) while falling short of producing something that is superintelligent in any real sense or capable of automating a broad swath of jobs in the economy.
4. More than one lab stays at the frontier. This means the market will remain competitive, and developers can’t charge high margins to compensate for rapidly falling inference prices. Any compute scarcity that allows higher margins is temporary, in my view. I also think it’s likely that most of the innovations continue to make their way into open-weight models with some lag, so open-weight models are at cost parity for inference or only slightly more expensive. Note that efficiency tricks and verifiable capabilities tend to leak much faster than unverifiable ones. Distillation could be one reason, but simply knowing that something can be sped up 10x tends to have a focusing effect on research and innovation.
5. Jevons’ paradox is not a magic bullet. If the above points hold, there are very few applications that will consume so many tokens that they could generate the astronomical amounts of revenue implied by the labs’ valuations. We’re talking about a 10x growth in tokens year after year, just to keep revenue flat given falling prices.
Even today’s behemoth tasks like migration of million-line legacy codebases will likely become cheap within a few years. While new applications will open up, there won’t be orders of magnitude more new ones. As for unverifiable tasks, inference scaling rapidly hits diminishing returns. Ultimately you have to have a human in the loop who’s accountable for the result. Spending more on tokens doesn’t let customers escape that responsibility — and only makes the human review bottleneck worse.
6. What about scientific research? Science is an endless frontier, and more tokens could perpetually mean more discoveries and inventions, in theory. But the unfortunate thing about research is that researchers don’t get to capture the value it creates, which is why it requires government funding in the first place. Even if it’s true that AI could enable great scientific innovations, I don’t think AI companies will make much money off of that. I think they are currently investing in AI-for-science not as a future revenue stream but as a loss leader because they want to show the public that it could be beneficial to humanity.
7. The one escape hatch is moving “up the stack”, by which I mean quickly transitioning from selling tokens to an enterprise software like model that relies on high switching costs, deep integration into workflows, and control over systems of record. In this world, enterprises will purchase “digital workers” or pay per solution (e.g. resolved support ticket) rather than tokens, so it’s good or neutral for AI companies if inference costs fall. And open models don’t matter as much — Linux being free didn’t make most enterprises switch away from Windows, because they paid for polish, peace of mind, support, and familiarity, not the code itself. I think this is the labs’ best bet for becoming sustainable businesses (though the potential lock-in is concerning for their customers!). The problem is that the labs have many hurdles to clear to get there, most of all their mindset, and the faster they race toward RSI, the shorter the window they give themselves for this transition. The pure-play AI labs are the ones who are farthest from being able to make this transition, and AI companies with a long history in enterprise software are relatively well positioned.
In short, the race to RSI ends up being a race to eliminate the labs’ main avenue for capturing the value that they create — the fate of many past transformative technologies such as railroads, airlines, and fiber.
The labs’ bet is that RSI produces a durable, decisive lead that escapes commoditization. My view is that there are two different improvement loops that have been conflated, and they will come apart. The efficiency improvement loop will accelerate, but a broader improvement loop that might lead to AGI or superintelligence will require ever-increasing amounts of compute, talent, data, and real-world feedback that can only come from gradual deployment, and the money to fund all this will have to come from massive revenue growth. And the irony is that the first improvement loop is undermining exactly that revenue growth.
The premise in my chain of reasoning that I’m least confident in is #5 — that Jevons’ paradox will eventually stop being able to compensate for efficiency improvements. A few months ago I was much more of a believer in continued, nearly unbounded token volume growth, but the more I’ve thought about it the more skeptical I’ve become.
To better understand why the premises and outcome I've laid out are at least a possibility:
* My AI-as-normal-technology essay and follow-up writings with @sayashk on our newsletter www.normaltech.ai
* My ICML talk laying out my skepticism of the claim that RSI will lead to superintelligence www.normaltech.ai/p/what-will-be-left-for-us-to-work
* Ramez Naam’s review of the evidence questioning the prospects of an intelligence explosion (@ramez) www.noahpinion.blog/p/wheres-the-intelligence-explosion
* My piece with Akash Kapur (@akashkapur) on whether AI labs will move up the stack www.normaltech.ai/p/up-the-stack-how-ais-escape-from
* Many pieces by Nathan Lambert (@natolambert) such as “I expect rapid progress but not towards general superintelligence.” -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ken Dilanian
@KDilanianMSNOW
Breaking: Federal judge dismisses—and lambastes—the DOJ false statements case against two former FBI officials. Calls it “weaponization.”
Judge Faruqui—a former federal prosecutor himself—said the government made a number of errors in their "slapdash job" of a case. He said there was no evidence presented before the court that the FBI agents' statements were knowingly false or material. He said the government failed to establish a motive for why the agents would make false statements.
He told prosecutors it was "incredibly unfair" that the DOJ sent into court with such a weak case. "This does not feel like an investigation," Faruqui said. "It feels like a broken one-wish willow obsession."
He added that he was "deeply troubled by what has happened here in this case," and that what happened "cannot be described as anything other than weaponization."
"It was a rush to charge, and that was a mistake," Faruqui said of the case. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
World's worst coverup
Fenit Nirappil
@FenitN
EXCLUSIVE:
@emilyykennard & @lizcgoodwin have found a fourth person who worked for Senate GOP candidate Mike Collins and is linked to incendiary online activity.
This account called for political violence and an end to democracy.
Read for free: www.washingtonpost.com/politics/2026/11/02/social-account… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jonne Kamphorst
@JonneKamphorst
Related to tomorrow’s #midterms: a new paper of mine, forthcoming at @The_JOP , on why working-class voters have been leaving the Democrats, and what that might mean for tomorrow. Thread:
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
vivek
@itsreallyvivek
I love that scene in Good Will Hunting where Ben Affleck is talking to Matt Damon's titular character Will Hunting and he's like “you've got a gift man and you've gotta get outta this town and make something of yourself” and the gift in question is being really good at math and the town he needs to get out of is Boston, Massachusetts.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Important @jasonfurman on the dangers of *underusing* AI jasonfurmanecon.substack.com/p/ai-could-do-my-first-job-n…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"Africa"
Reggie B.
@reggiebblue
Trump supporter: Trump has ended 7 or 8 wars, maybe 9.
Q: Which wars?
TS: Venezuela, Africa. I watch Fox News all day. They keep sharp on things.
Even though I don't know the countries that he stopped wars, he said he did, and I believe he did. I watch Fox to know what's going on in this country.
Q: Do you ever watch the others?
TS: They are fake news. Fox is real.This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tom Malinowski
@Malinowski
Supporting a GOP Senate candidate in Michigan who said it was fine his campaign team had "varying viewpoints" though one of them had posted Jews should be sent to North Korea, and another defended the young Republican "I love Hitler" group chat.
Imagine if that was El Sayed! 3/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Daniel McDowell
@daniel_mcdowell
Economists waking up in the Year of Our Lord Two Thousand Twenty-Six to discover international relations.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan BdM
@ethanbdm
Friends, it is time for some old-school Twitter pedantry.
A Condorcet cycle is absolutely not "a demonstration of Arrow's theorem in the wild."
1/
David Shor
@davidshor
I made some plots to help visualize why coalition formation in Israel is so difficult right now
We are witnessing an extremely rare situation that political scientists call a "Condorcet cycle" - a demonstration of Arrow's impossibility theorem in the wild! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Declining and increasingly disinhibited. Has the nuclear codes for 809 more days.
Headquarters
@HQNewsNow
Trump: I have two beautiful sisters. One was a federal judge, top federal judge. The other is beautiful but I never had an interest in dating her.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sylvain Catherine
@sc_cath
My views on AI have radically shifted over the past few weeks, after spending way too much time reading about it and watching interviews of industry leaders and experts.
First, I really love the tools. I spend $200 on Claude every month and the product is worth every penny for me. Probably even more. I think the existing tools can deliver significant increase to future GDP and help us solve major global challenges faced by humanity.
However, I just cannot comprehend why the industry is not much more regulated, as its leaders and employees seem to request themselves. I meet a lot of people who have heard of the hugging face incident but don’t know the actual chain of events. If you don’t, you definitely should spend a least an hour reading about it.
I also do not know why we need to follow the path that frontier labs are following, to develop ever more powerful models. It seems to me that they are enormous gains that are readily available if we deploy the current technology in economic and scientific organizations without pursuing a super intelligence that they say might kill us and more likely wreak havoc in society. I know a lot of people think these warnings are marketing tricks, and that’s a reasonable assumption. But I don’t see the point of taking such risks.
My median scenario is that of many economists: AI will diffuse slowly and lift gdp growth by a percentage point or so, will cause some economic disruption on the way, but ultimately people will get new jobs and maybe work less.
However, I am deeply uncomfortable with the absolutely wild level of uncertainty around this prediction. And to be honest, even setting the doom scenarios aside, I find the utopia scenarios of some industry leaders actually dystopian. And I think a large majority of the public would agree if they paid attention to what they’re saying. To their credit, they themselves agree publicly that they should not be the ones making these choices. A large majority of the public is now very skeptical of AI as a source of human progress, in particular among younger adults who use it the most. They should have a say.
I don’t want to convince you of anything but to get informed about this topic, express yourself and vote accordingly. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Rupar
@atrupar
"I wanted to see if you wanted to fill in the silence, but you don't" -- OOOOOF this Haley Stevens interview on CNN is BRUTAL
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ben Fritz
@benfritz
Detour from my normal business coverage: A personal essay about watching "You Can See Everything" as a neurodivergent person. www.wsj.com/arts-culture/film/autism-you-can-see-everythi…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
I̶n̶f̶e̶r̶e̶n̶c̶e̶ ̶I̶n̶n̶s̶
G̶i̶g̶a̶w̶a̶t̶t̶ ̶G̶a̶r̶d̶e̶n̶s̶
S̶i̶l̶i̶c̶o̶n̶ ̶S̶a̶n̶c̶t̶u̶a̶r̶i̶e̶s̶
B̶r̶a̶i̶n̶ ̶B̶a̶r̶n̶s̶
I̶n̶f̶e̶r̶e̶n̶c̶e̶ ̶I̶g̶l̶o̶o̶s̶
A̶l̶g̶o̶r̶i̶t̶h̶m̶ ̶A̶r̶b̶o̶r̶e̶t̶u̶m̶s̶
C̶o̶n̶s̶c̶i̶o̶u̶s̶n̶e̶s̶s̶ ̶C̶o̶n̶s̶e̶r̶v̶a̶t̶o̶r̶i̶e̶s̶
Compute Campuses
Axios
@axios
NEW: Right after Tuesday's midterms, OpenAIwill unleash a policy and PR blitz to try to get in sync with the new Washington, including an effort to rebrand data centers as "AI compute campuses." www.axios.com/2026/11/01/scoop-openai-post-midterm-policy
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nick Field
@nick_field90
Nate Cohn points out that the Times' changed its weighing process this year. This is what the new polls would look like with the old weighing process. So they're already trying to protect against a pro-Dem polling error www.nytimes.com/2026/11/01/upshot/poll-times-siena-senate…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Danish Hussain
@astrodanish
people will be so snobby about arts and literature and then look down on sports as if this isn’t the most visceral expression of the indomitable human spirit.
sports distills what it means to be human in a way no painting ever could
Trojan Football ✌️ ᶠᵃⁿ
@TrojanFBx
USC Mark Bowman’s 70-yard run, when he hurdles a defender, goes crazy from the field level view 😮💨
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kiran Garimella
@gvrkiran
We're releasing one of the largest & most detailed look at how Americans use ChatGPT complementing reports from OpenAI, giving a much more deeper view into ChatGPT use
Explore our interactive tools & analysis: stateoftheprompt.net
Let us know if you want to see any analysis!
Katherine Ognyanova (@ognyanova.bsky.social)
@Ognyanova
STATE OF THE PROMPT: Our Rutgers CSS Lab team analyzed a year of real ChatGPT conversations from 70K+ US panelists. See the new interactive tools, plus research reports on health, politics & more.
stateoftheprompt.net
With @gvrkiran @shagunjhaver @RutgersCommInfo @RutgersU -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Fool for a client, political consultant edition
Shane Goldmacher
@ShaneGoldmacher
NEWS: Roger Marshall goes rogue.
The inside story of how an embattled Kansas Republican wanted an ad to push back on medical debt attacks, was told no by his team, made one anyway without knowledge of some advisers and got it out into the world.
www.nytimes.com/2026/11/01/us/elections/roger-marshall-ad… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Adam Carlson
@admcrlsn
How self-identified independents break in the Senate races, per final NYT/Siena polls:
Alaska: 🔵 Peltola +35
Texas: 🔵 Talarico +30
Kansas: 🔵 Hamilton +22
Michigan: 🔵 El-Sayed +17
Iowa: 🔵 Turek +17
Maine: 🔵 Jackson +11
Ohio: 🔵 Brown +9
Adam Carlson
@admcrlsn
Final 2026 New York Times/Siena polls (10/23-30, likely voters):
—
ALASKA:
Senate:
🔵 48% Peltola (+1)
🔴 46% Sullivan
Governor (final round of RCV):
🔵 54% Kreiss-Tomkins (+8)
🔴 46% Sullivan
—
IOWA:
Senate:
🔵 48% Turek (+1)
🔴 47% Hinson
Governor:
🔵 51% Sand (+8)
🔴 43% Lahn
—
KANSAS:
Senate:
🔵 49% Hamilton (+5)
🔴 44% Marshall*
Governor:
🔵 48% Holscher
🔴 47% Masterson
—
MAINE:
Senate:
🔵 51% Jackson (+6)
🔴 45% Collins*
Governor:
🔵 49% Pingree (+13)
🔴 37% Charles
—
MICHIGAN:
Senate:
🔵 53% El-Sayed (+11)
🔴 42% Rogers
Governor:
🔵 52% Benson (+12)
🔴 39% James
—
OHIO:
Senate:
🔵 49% Brown (+6)
🔴 44% Husted
Governor:
🔵 49% Acton (+6)
🔴 43% Ramaswamy
—
TEXAS:
Senate:
🔵 50% Talarico (+7)
🔴 43% Paxton
Governor:
🔵 48% Hinojosa (+2)
🔴 47% Abbott* -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nate Cohn
@Nate_Cohn
The final Times/Siena polls of the campaign find Democrats ahead in all seven of the states likeliest to decide the Senate
www.nytimes.com/2026/11/01/us/politics/times-siena-poll-s… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Steve Kopack
@SteveKopack
(It’s mile 7.)
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrej Karpathy
@karpathy
Weekend hack: HackerTube. I was annoyed with YouTube's recommendations so I had Claude filter down 43M comments over 20 years of HN to 6,454 videos that keep being recommended. *Significantly* better viewing experience, ~2 hours with Opus 5.5:
hackertube.org
- Excellent recs without any logins (state is kept in local browser session)
- Thumbnail is a representative frame from middle of the video instead of the super duper A/B test hyper-optimized user uploads
- Underlying data from the comment analysis as one zip file, point your Claude at it and ask for your own version
- HackerTube Longs ✊
A bit more on the process:
I use Claude Code (desktop app for Mac), Opus 5.5 on Medium effort as my default. I only talk to Claude and have it do everything, the only "manual work" I helped with in tandem was in the browser (reminiscent of MenuGen): 1) I had to create some kind of API key from Google Cloud UI to query YouTube for video metadata, 2) I booted up a VPS box for serving and 3) I bought the domain name. Once the box was up I gave Claude the IP and told it update, secure, harden, deploy and optimize the code, and it did the whole thing.
Many awesome projects that would have normally taken weeks/months even can now be few-shotted for breakfast. What else are we not thinking of. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kyle Cheney
@kyledcheney
BREAKING: A judge in Alaska has thrown out the Trump administration's lawsuit seeking to delay certification of the Senate election if GOP incumbent Dan Sullivan loses.
storage.courtlistener.com/recap/gov.uscourts.akd.80263/go… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
jessicat
@jessi_cata
I concur with Scott Alexander that some parts of AI safety, especially slowdown & awareness raising oriented parts, have started behaving more like a mass political movement, with associated bad epistemics.
jessicat
@jessi_cata
AI Impacts’ 2024 press release claimed the median AI expert gave a 5% chance of extinction. The survey actually asked about extinction OR similarly permanent and severe disempowerment.
So the misleading framing wasn’t just from journalists.
aiimpacts.org/wp-content/uploads/2024/01/EMBARGOED_-AI-Im… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Populism diners etc
Jake Sherman
@JakeSherman
Trump at his costume party at Mar-a-Lago a few days before Election Day, per @tylerpager:
“Look at all my rich members,” he told reporters. “They spent all day getting ready. Look at these guys.”
He added: “These are the biggest guys on Wall Street.”
www.nytimes.com/2026/10/31/us/politics/trump-halloween-pa… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
None of the people in the Trump administration have done the reading/viewing. None.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Alternative explanation: it's intentional
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Heads they win, tails they lose. Every time.
Roger Parloff
@rparloff
DOJ says the suits to block Trump’s Arch don’t become ripe until agency action is final. That won’t happen, they say, ‘til the Nat Capitol Planning Com’n approves the Arch on 11/5 & Dept of Int approves on 11/5 & Nat Park Svc publishes notice on 11/5. Work then starts—11/5.
1/2 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Marc Polymeropoulos
@Mpolymer
I’m still very surprised it is not more of a scandal at DoD– how in the world did decades of US military presence in the CENTCOM region result in what looks to be inadequate if not terribly negligent force protection measures? Where is the responsibility for this? Accountability? This seems to me to be a massive cluster. Did we not think the Iranians would use their missiles and drones to hit our bases? Is Congress asleep in not asking these questions? www.wsj.com/world/middle-east/iran-war-us-bases-damage-88…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Imas
@alexolegimas
This is a wonderful thread on the discussion of how AI is affecting productivity.
Main take away: the documented trends amongst AI exposed firms for hiring/employment/productivity started well *before* LLMs, so likely not causal effect of AI or at least an interaction (I’m personally in the latter camp).
John Burn-Murdoch
@jburnmurdoch
Quick thread on AI and productivity:
Lots of people say this chart shows AI is now showing up in the productivity data.
But I made this chart to argue the opposite: a trend that pre-dates AI and only shows up in the US doesn’t look like AI.
Here’s how I’m thinking about the data and underlying dynamics: -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
We've all seen declining, disinhibited elderly relatives act like this. But none of them have the nuclear codes.
Jonathan Chait
@jonathanchait
I pay very close attention to Trump's mental decline, but even so, the details here are striking www.nytimes.com/2026/11/01/us/politics/trump-speeches-ral…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Reed Galen
@reedgalen
It was a legendary call.
Awful Announcing
@awfulannouncing
"BALL IS OUT! THE BALL IS OUT! AND IT'S IN THE HANDS OF WATSON! 333-POUND MARAAD WATSON! LIKE A TRACTOR TRAILER GOING THROUGH THE TEXAS NIGHT... GET THAT GUY SOME OXYGEN!"
Joe Tessitore went absolutely wild for a big man TD on the Mississippi State-Texas ESPN broadcast. 🏈💥🎙️This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Laura Bassett
@LEBassett
Calling your tour Come and Take It and then retreating from the Alamo is the funniest possible thing this man could do
Politics & Poll Tracker 📡
@PollTracker2024
Texas Tribune: Ken Paxton appears to cancel final rally of his “Come and Take It” tour at the Alamo
Paxton’s campaign initially said he would close his get-out-the-vote push at the landmark.
www.texastribune.org/2026/10/31/ken-paxton-alamo-election… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Super AI-pilled people are indistinguishable from a religious cult, part 1 million
Carl
@HistoryBoomer
Scott Alexander arguing that in the future it might be a good idea to turn all power over to a “godlike AI.”
I understand he’s not saying this is the only option, but even as a possible option, this kind of thinking gives me the creeps.
Not least, because it would probably be impossible to know for sure if a “godlike AI” were “fully aligned.”
I’m team Keep Humans In Charge! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Great take
Justin Morissette 👊🏻
@JustinMoris
Watching Elizabeth Holmes spend her final days before going to jail speed-running the creation of Theranos 2.0 kind of redeemed for me the idea that the Empire would just make a second Death Star.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
jasmine
@jasminecoded_
I hate sf halloween, what do you mean you’re going as survivorship bias
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
John Burn-Murdoch
@jburnmurdoch
Quick thread on AI and productivity:
Lots of people say this chart shows AI is now showing up in the productivity data.
But I made this chart to argue the opposite: a trend that pre-dates AI and only shows up in the US doesn’t look like AI.
Here’s how I’m thinking about the data and underlying dynamics: -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Every time
Charles Fain Lehman
@CharlesFLehman
hahahahahahahaha
JK
@_junaidkhalid1
anthropic is about to watermark the text claude writes for you, worldwide, starting tuesday. the detector goes to regulators, journalists and big enterprises. everyone except the person who paid for the text.
so your cold email, your linkedin post, the resume you rewrote three times can get flagged as claude by some hr tool or editor, and you can't even run the check yourself to see what they'll see.
openai only does this in the eu. anthropic rolled it out everywhere.
you pay them monthly and they build a snitch layer on top of your work.
Pete Sena
@petesena
Claude gonna start snitching on Tuesday.
Anthropic just emailed customers. Starting Nov 3, text carries an invisible watermark.
You can't see it. It survives copy and paste, and it can survive some editing.
Only regulators, journalists, fact-checkers and researchers can check for it right now.
It doesn't say who you are. It says Claude touched the words. Even a quick proofread counts.
If your brand voice runs on Claude, write like someone might check. Because now someone can. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Steve Witkoff to Putin: “You are a humanitarian also, sir, and I salute that, I commend you.”
WarTranslated
@wartranslated
Russia's war losses may have hit 2.1 million troops with up to 770 thousand killed, according to a new Economist estimate, a sharp jump from its May figure of 280 to 518 thousand dead.
The Economist now puts total casualties at 1.4 to 2.1 million and deaths at 374 to 770 thousand, up from 900 thousand to 1.3 million total and 190 to 350 thousand killed back in July 2025.
Ukraine's General Staff separately puts total Russian casualties at around 1,588,170 as of October 31, with Russia losing an average of 1750 troops a day in October alone.
Ukrainian forces meanwhile advanced 73 square kilometers over the past three months, retaking about half the territory Russia seized this year.
www.economist.com/interactive/graphic-detail/2026/10/30/s… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Yglesias
@mattyglesias
The difference is that the New York Times, which has a lot of credibility with left of center audiences, treated Alaska as a huge scandal while totally ignoring it in Nevada.
David Weigel
@daveweigel
Can't get enough of this ratf*ck GOP Nevada mailer that makes it look like a left-wing candidate who shares a last name with the Dem candidate is endorsed by AOC.
It's not just Alaska - this crap is in the DNA of modern campaigns. thenevadaindependent.com/article/pac-boosting-third-party… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Yglesias
@mattyglesias
Important context for the AI safety debate is that a very large share of the "normal technology" applications currently being developed seem alarming to me and likely to accelerate negative trends in human welfare rather than advance flourishing and happiness.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Where we are as a country. Appalling that a judge must issue such an order to protect our elections.
Joyce Alene
@JoyceWhiteVance
An extraordinary order: DOJ “shall not threaten, intimidate or coerce state or local elections officials or workers to demand that they halt processing the ballots of any registered voter solely on the basis that the ballot” is someone DOJ says may not be a citizen
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ryan Miller
@rmm4pi8
I think this is something non-academics don't comprehend about the academic job market. We're more like classical musicians or professional athletes in our job opportunities.
Ryan Miller
@rmm4pi8
When I worked in software, there were dozens of postings I was qualified for in any decent city at any time of the year. Now that I'm an academic, there are a dozen jobs I'm qualified for in the world, and they hire once a year.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Adam Carlson
@admcrlsn
We are living inside of a live political science experiment, where we will see in just a few days — in a cycle with an incumbent president and party this unpopular — to what extent lower survey response rates among the depressed party translates to lower base turnout.
Kudos to the Washington Post/George Mason University polling team for publishing this for transparency reasons, despite it clearly being an outlier.
Scott Clement
@sfcpoll
We polled Kansas. Democrats wanted to talk. Republicans, not so much. Gift link -> www.washingtonpost.com/politics/2026/10/30/we-polled-kans…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Wu Tang is for the Children
@WUTangKids
Oh wow they finally fixed that ridiculous DWade statue
Miami HEAT
@MiamiHEAT
“It may be my likeness, but it will always be our house.” - @DwyaneWade
A closer look at the enhancements to his statue. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
DCCC heat check - may just be a test, but shows where things stand
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Meg Schwenzfeier
@_Schwenzfeier
Fascinating to see in the data exactly when Google made their policy change for the 2020 election question.
Jack Welty
@jackwelty
Closing the loop here -- Google made a change! Before yesterday, AI Mode actually deflected ~70% of the time for _both_ 2020 and 2024, but around midday, someone flipped a switch on the 2020 query.
𝙲𝚑𝚊𝚛𝚕𝚎𝚜 𝙲. 𝙼𝚊𝚗𝚗
@CharlesCMann
I had real trouble believing that Google was actually doing this, but I just duplicated the result myself. The two queries were about a minute apart and identical except for a single digit.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Shiro Kuriwaki
@shirokuriwaki
The first release of my R package `fourthirtyfive` makes congressional district maps (2005-27, incl. those for next week's election) easily downloadable and mergeable, with hex/dotmaps, real shapefiles, descriptive names, and interactive overlays with #mapbox:
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Floating the not seating gambit
Justin Baragona
@justinbaragona
So it appears the plan is to just try to prevent Mary Peltola from being seated in the Senate if she wins next week?
Here's Ari Fleischer and Harris Faulkner on Fox News laying out that very scenario. "Members of Congress decide if you can be seated in that chamber."
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Thread on this from an expert
Michael Thorning
@ThorningMichael
🧵Gaming out an Alaska Senate Dispute
With things in Alaska heating up, I'm going to do my best to walk through some of the big questions if an Alaska Senate dispute unfolds....
1/12This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
We're going to miss this version of the (dying) open Internet when it's gone
BuccoCapital Bloke
@buccocapital
Reddit call was absolutely fascinating. Everyone should listen to it as a bellwether for where the internet is going
- AI Mode is increasingly default Google experience
- AI Mode provides rich info and data in line, less incentive to leave search
- AI Mose incentivizes dialog, Google wants to keep you engaged to serve more ads at each turn
- Google will milk search for as much revenue as possible as they transition from a search company to a cloud service and chip company
- As traffic from humans becomes more scarce this will put Google in an increasingly adversarial position to the rest of the internet
- This is a massive headwind to all media type companies who are not destinations
- This is a massive headwind to all content-driven go-to-market motions especially in the SMB. You have seen companies like HubSpot and monday.com get crushed by this funnel collapse
We are entering the end game - a knife fight for the remaining and dwindling volume of human traffic on the internet -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Unprecedented corruption
Wendy Siegelman
@WendySiegelman
NEW: The latest filing from Trump Media & Technology Group, 41% owned by the US president, shows they will earn $350K from a foundation linked to the National Park Service for ads related to the celebration of America's 250th anniversary. My latest:
newstracs.com/trump-media-will-earn-350k-for-ads-by-freed… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Vote like the Big Fundamental
NBACentral
@TheDunkCentral
Spurs legend Tim Duncan was spotted standing in line with other voters in San Antonio for the 2026 midterm elections.
(Via @JeffGSpursZone ) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
🚨 Bright Line Watch pre-election report
-Election confidence highest for Ds but ↑ for Ds & Rs
-Experts/public expect more eligible voters blocked than typical
-Forecasters: top R fraud claims likely followed by military/law near polls & equipment/ballot seizure
Results 🧵 : -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Forecast estimates for House and Senate seats from @johnmsides in the Good Authority newsletter
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
Great syllabus here from Misha - an info science + history of technology + sociology + economics view on what happens what an industry is disrupted by a new technology; the usual suspects plus Ong's orality and "Desk Set". Love it!
Misha Teplitskiy | Science of Science
@MishaTeplitskiy
Creating a new class here at @umsi and IMHO the syllabus turned out pretty good. Here's a draft in case useful for others
(link in comments) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jeremy Barr
@jeremymbarr
Putting a 5-day-long job in your bio certainly requires a certain kind of chutzpah that you almost have to admire
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Chris Kennedy
@ChrisJKennedy
I truly cannot imagine still working at @StateDept and seeing this come across my desk. We’ve gone from isolating brutal, torturous governments to paying them to accept deportees.
Lara Seligman
@laraseligman
The Trump administration agreed to send around $35 million in military aid to a Central African military unit previously accused of war crimes as part of a wider arrangement to deport third-country migrants to the region, U.S. officials and congressional aides said.
Great scoop from @RobbieGramer: www.wsj.com/world/africa/trumps-deportation-deal-with-cam… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Narrative violation
philip lewis
@Phil_Lewis_
an incredible UNO reverse here
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jessica Tillipman
@JTillipman
I’m genuinely torn between which is worse: the fact that a Contractor was running a personal AI agent on the DoD network, or that it took DoD so long to figure it out.
I can only imagine the poor Accenture compliance officer when they learned out it. Shadow AI is a compliance nightmare.
WIRED
@WIRED
Sources tell WIRED that an Accenture employee’s personal Claude agent ran on a DOD network for two months before it was identified and the vulnerability could be patched. www.wired.com/story/a-defense-contractor-gave-their-perso…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sam Stein
@samstein
This guy was our acting DNI for a period of time
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Adam Carlson
@admcrlsn
Incredible data visualization
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Carl Quintanilla
@carlquintanilla
(WSJ) - Hackers appeared to have plundered at least $1.7 billion Thursday from Binance in what would be the largest-ever cryptocurrency heist.
@WSJ
www.wsj.com/finance/currencies/binance-founder-says-excha… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
This research gets at something missing from most debates about AI’s economic impact: coordination.
Organizations are more than the sum of their best performers, so smarter individual agents won’t transform the economy on their own. Agents that can coordinate together might do.
Alex Imas
@alexolegimas
I joined @GoogleDeepMind to develop a research program on the economics of AGI. Today, together with Andrew Koh (@andrewjkoh) and Shane Legg (@ShaneLegg), we are incredibly excited to release The Economics of Intelligence, the first in a series of projects from our group on how AGI will change the economy.
How do capability improvements transform the economy, and when? In this project, we define thresholds of economic capabilities, e.g., minimal AGI, based on how they transform the economy. We then track how the economy–market structure, employment, and wages–changes during the transition across capability regimes.
We have seen exponential increases in what AI agents can do but the broader economy seems almost remarkably unchanged. Part of the reason is that the solo AI agent is the wrong economic object; economic transformation is an equilibrium outcome where organizations compete. Looking at how intelligence functions within organizations, transformation will depend on when two key capability thresholds are crossed: competence (gains on individual skills, e.g., coding, math) and coordination (gains on ability of agents to cooperate and work together).
We show that what economic transformation looks like depends on whether AI becomes supercompetent, supercooperative, or both—the regime of artificial superintelligence (ASI). Tracking this transition shows why the real visible transformation may come from improvements in coordination—a capability most benchmarks don’t yet track. Past a certain threshold, AI capabilities can break the Coasian ceiling on firm size and lead to an economy that looks drastically different than the one we see today.
Two main takeaways:
1. To understand the path of transformation, we should be tracking agentic coordination more closely.
2. The impact of AI will likely show up in wages before market structure or employment.
We are really excited about this paper. Excited to hear people’s thoughts.
Link: institute.deepmind.com/essays/the-economics-of-intelligence/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ryan Goodman
@rgoodlaw
Bipartisan group of 28 former governors from across the United States:
"respectfully submit that this lawsuit [by Trump-Blanche DOJ against California] constitutes a brazenly unlawful and dangerous effort by the federal executive to usurp states’ authority to administer elections." -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sewon Min
@sewon__min
Good slides are extremely intentional. When someone presents a paper, the paper already has all the information. But the speaker would make deliberate choices about things like, what to say and what not to say, what to put on a slide and what to say only verbally.
What to frame as a "side note" and when to say "this is the part I found particularly exciting."
Even individual word choices, like the subject, verb, and adjective, and when to pause, are intentional.
In fact, a good speaker would never really give a talk about a paper, even when that's seemingly what they're doing.
They would talk about what they want to say, like the story they want to tell, the kind of research they like, the vision they have, etc. Then they use the paper to support / back up their message.
Some tools are perfectly fine to use. But some skills take so much time and efforts, like years, and some tools proactively prevents learning them.
Tal Linzen
@tallinzen
my hypothesis for why AI-generated slides for technical presentations are so awful:
people who don't have a lot of experience presenting think that to make a talk you just put all of the stuff you want to talk about on the slides, then you go through each slide and say the stuff the slide says.
how you're actually supposed to prepare a presentation: you decide what to say, write it out as a rough script, then put maybe 3% of the text on the slide, plus a bunch of visual aides (e.g. plots that only include data you're planning to discuss).
AI-generated presentations are mostly the former. they have a ton of stuff the presenter never mentions, and the stuff that is most useful for the presenter is absent or not highlighted. they include enough content for a presentation 10 times longer than the actual presentation is supposed to be.
maybe try doing it the other way: first write the script for what you actually want to say, and make sure you can say all of it in the time you have. then prompt the model to create the slides based on the script. ask it to use as little text a possible, and a minimalist visual style rather than the claude house style. you can thank me later (I accept both venmo and zelle). -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nathaniel Hendren
@nhendren82
There are many thoughtful critiques of this line of work. This is not one of them. Sad to see @margrev engage in this. Only upside is that the post clearly illustrates the need to have humans in the loop in social science...at least for now :)
Marginal Revolution
@MargRev
Errors in the research of Emmanuel Saez and Gabriel Zucman marginalrevolution.com/marginalrevolution/2026/10/errors-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Results from AI models are a minor infohazard, but other people's AI results, shared with anonymity, an agenda, and no prompts are a *major* infohazard.
You need to be extremely not paying attention to trust this kind of post out of the box.
steve hsu
@hsu_steve
Frontier models are much better than human reviewers. I trust the methodology below far more than any journal review process.
saez-zucman-audit-replication.vercel.app -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
AP: Stolen election rhetoric from Trump isn’t new. His ability to act on those claims is
apnews.com/article/midterms-trump-voting-election-interfe…
Not a drill -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Robbie Wallin
@WallinRobbie
"Donald Trump is not the first president to break the law. He is not even the first president to insist that his actions cannot, on some deep ontological level, be illegal. But he is the first president to abandon the principle that the law matters at all"
www.nytimes.com/2026/10/28/opinion/trump-law-lawlessness-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
One of the most important findings in our new report
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-Experts & AI forecasters give low probabilities to many potentially disruptive events
-They think most likely is top Rs alleging fraud affected results
-Forecasters also give meaningful chance to military/law enforcement near polling places or seizing voting equipment or ballots -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
2026 most resembles 2020 in the level and polarization of pre-election confidence we observe among the public
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-Public confidence in the 2026 election is greatest among Democrats, but it has increased since July among supporters of both parties.
-Experts are most confident overall. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
ICYMI yesterday!
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
🚨 Bright Line Watch pre-election report
-Election confidence highest for Ds but ↑ for Ds & Rs
-Experts/public expect more eligible voters blocked than typical
-Forecasters: top R fraud claims likely followed by military/law near polls & equipment/ballot seizure
Results 🧵 : -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Peter Hull
@instrumenthull
How different are different differences-in-differences? The answer may (or may not) surprise you
New working paper, building on a dumb tweet from earlier this month x.com/instrumenthull/status/2105691784386887781
A short summary follow:
Peter Hull
@instrumenthull
Why has no one done the massive AI-assisted "how much do negative weights really matter for difference-in-difference" review paper yet?
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Anna Stansbury
@annastansbury
"Innovators from more affluent backgrounds cater to richer consumers"
This is one of the reasons why representation among innovators & researchers by socioeconomic background matters. Background shapes questions, tastes, ideas, and motivations of which problems to address.
AEA Journals
@AEAjournals
Innovators tend to create products used disproportionately by people who share their gender, age, and socioeconomic background, say researchers at @vatt_tutkimus, @UUtah, and @LSEnews. #chart www.aeaweb.org/research/chart/social-push-innovation-dire…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Eric Boehm
@EricBoehm87
This is so so so good
Jay Nordlinger
@jaynordlinger
My gratitude for Kevin D. Williamson is not lessened for its being routine. thedispatch.com/newsletter/wanderland/jd-vance-christiani…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Mark Jia
@MarkZJia
Seems like many firms may eventually move to a Wachtell-like partner:associate ratio (less than 2:1). Even with AI support, rainmaking partners will still need junior level lawyers to manage the legal workflow, but maybe fewer than in the current model. This may actually encourage even more lateral movement among rainmakers, at least until there is a new equilibrium. Fare-Weil to biglaw's golden age.
Will Chen
@willchen500
I think the Weil fallout might point to the fact (which others have pointed out) that the traditional law firm partnership structure does not work anymore. The truth is a Biglaw firm is a rather commoditised support platform that serves rainmaking partners with back office operations, supply of labour from associates, counsels and service partners. Who controls the cash flow effectively controls the firm and if the rainmakers head for the door it leads to a bank run type scenario. In most firms the power has shifted majorly towards a few rainmakers particularly in PE and M&A and that creates a huge key person risk when they decide to pack their bags.
I’m not sure what the solution to this is but I imagine it’s only going to get worse when the platform that firms offer get even more commoditised by AI which will do more and more of what junior to mid-level lawyers are currently doing. Maybe the answer is building some real tech moat within your firm? Or just being able to predict market trends really well and investing correctly in the right practices like Kirkland. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alexandra DeSanctis Marr
@xan_desanctis
2026 has given us two of the best sports photographs of all time
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Anna Bower
@AnnaBower
FBI Director @FBIDirectorKash and his top deputies apparently held weekly meetings about a false statements case that is predicated in part on a hypothetical question about a hypothetical laptop.
Anna Bower
@AnnaBower
Wild scenes in court today at the probable cause hearing for two former FBI agents charged by DOJ for alleged false statements to Congress
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
One reason education administrators should avoid AI writing is that when they get caught, they then need to dig deeper into outright lies like "AI detectors don't work" and "I wrote a draft and just used AI for grammar."
The truth in most of these cases is that they and their offices think AI writing is helpful and efficient and fine (it's just like ghostwriting), but they realize that is untenable as a public opinion because it undermines their mission to teach students that writing skills actually matter for something.
But once you start lying about something, it's hard to stop, and getting caught in your lies kind of undermines your educational mission.
Austin Horn
@_AustinHorn
Pangram also flags the 2 speeches posted to Mack's "official correspondence" as majority-AI.
The state's controversial new high school science standards? 32% AI.
Here's DOE's response as well. The take on detectors contrasts w/what 3 professors in the space told us. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jason Furman
@jasonfurman
If you believe that mindless AI evaluations are legitimate then why would you spread the evaluation of Saez & @gabriel_zucman without first subjecting it to its own mindless AI evaluation?
I just did and it turns out the critic's errors all go in one direction. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Martin Borch Jensen
@MartinBJensen
A data point from @ImpetusGrants: we have tried human reviewers and LLM reviewers. I think LLM reviewers are already more capable than any individual human reviewer for many grants, because they can spot issues across statistics, biology, epidemiology and other fields all in one proposal.
But humans are still better at tacit knowledge, knowing all of the unpublished experiments, the motivations of researchers, ideas that seem good on paper...
Anshul Kundaje
@anshulkundaje
Going to disagree here. On average an AI generated review will be better than a human generated one from the average reviewer especially in esoteric or cross disciplinary areas. I do agree that it is better to read the paper first, do ur own review & augment ur review with AI. 1/
This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Unprecedented corruption
Andrew Curran
@AndrewCurran_
Bloomberg has signed up for the $100k/month Truth API.
This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
The Zucman re-analysis is shots fired—clearly this is coming for every paper.
This said:
1. I don't update at all on Zucman without seeing the base distribution of "errors" in econ papers
2. It's hard to believe this anonymous creator is neutral and using a neutral prompt
3. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nicholas Decker
@captgouda24
This is a ridiculous way to go about it. Absolutely ridiculous. AI, told to find things wrong, will find things wrong. You still need to go through and identify what mistakes actually matter! Uncritically reposting this discredits the reposter. Right now it’s just a DDoS attack.
Marginal Revolution
@MargRev
Errors in the research of Emmanuel Saez and Gabriel Zucman marginalrevolution.com/marginalrevolution/2026/10/errors-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Antoine Levy
@LevyAntoine
Careful. Looking at just a few of the papers I know well here, this "audit" seems to be to be mostly BS - and I say that as someone whose views are about as opposed to S-Z as you can find in the profession
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Rupar
@atrupar
scenes from a fascist movement
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
The people most vocal about errors in econ papers (there are many!) are motivated, and trying to take down specific people (e.g. Zucman), or debunk the field as a whole.
If we aren't willing to criticize ourselves / each other, we do these people a favor by making them look like the only truth seekers.
Tons of econ papers have errors, and those errors will come to light. This moment is a test of our belief that econ is the most truth-seeking of the social sciences.
Trying to bury and obfuscate the errors will hurt us in the long run. Some papers will need retractions, all will need errata.
It's an individual nuisance—I'm busy too and don't want to be bothered to go and re-replicate and audit all my past work. But if we don't show a willingness to cleaning up our own house, we can count on our opponents to do it.
Whether we will be more honest about our errors than our opponents remains to be seen. I suspect the risks of admitting you were wrong are probably lower than you think. It will take some brave authors to lead the way and set the standard. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Meg Schwenzfeier
@_Schwenzfeier
Fwiw we just collected some quick data on this and it looks like equally partisan dodging of the question on Google AI Mode to us -- the model deflects to election resources on about 70% of both questions.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
WTF. Just replicated.
𝙲𝚑𝚊𝚛𝚕𝚎𝚜 𝙲. 𝙼𝚊𝚗𝚗
@CharlesCMann
I had real trouble believing that Google was actually doing this, but I just duplicated the result myself. The two queries were about a minute apart and identical except for a single digit.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Garrett Haake
@GarrettHaake
I’ve covered dozens if not hundreds of Trump rallies and I can’t recall him ever holding one in a room as small as this arena in Chillicothe, Ohio. It’ll be packed and loud- but maybe 1000 people will fit.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kyle Cheney
@kyledcheney
A federal judge found that an immigration court bond hearing was so egregiously stacked against an ICE detainee that he may simply order the release of ICE detainees, rather than bond hearings, from now on.
storage.courtlistener.com/recap/gov.uscourts.iand.71471/g… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
WTF. Just replicated.
𝙲𝚑𝚊𝚛𝚕𝚎𝚜 𝙲. 𝙼𝚊𝚗𝚗
@CharlesCMann
I had real trouble believing that Google was actually doing this, but I just duplicated the result myself. The two queries were about a minute apart and identical except for a single digit.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Don't know
Todd Eberly
@ToddEberly
This is what I just got:
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"Reviewers without training found 27% of the major errors, and training barely helped. We gave the same papers to 15 LLMs. They found 53-96%. Even the smallest models found about twice as many major errors as human reviewers"
Why we need human AND LLM review
Ruslan Rust
@rust_ruslan
Lots of debate this week on whether journals should ban AI in peer review.
We actually tested this in our new preprint, and the results were pretty clear to me.
For context, I've reviewed for most of the big journals and evaluated 1,000 human reviews as an editor. I love to read and evaluate science.
TLDR: From now on I won't do any peer review without AI. For clinical trials, I'd even call human-only review dangerous 🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Eva Vivalt
@evavivalt
New working paper: Can AI forecast treatment effects in economics from pre-registration documents?
Answer: Yes. 1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
AI influence in ambiguous and difficult situations >>> AI "persuasion" on controversial issues via focused conversations with users or artificial social media posts
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Bryan Walsh
@bryanrwalsh
This is such a smart description by @KelseyTuoc on why the inquiry method prevalent in education FEELS like it should work, but doesn't. The whole piece is worth reading. www.theargumentmag.com/p/sold-a-story-is-not-the-whole-story
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Hall
@ahall_research
This is the case in polisci too. It is driven in part by a hyper focus on what can fit into normal academic journals. Journals are more open to publishing “methods” pieces on AI and somewhat less open to substantive pieces on how AI affects/shapes politics. This is changing, but not fast enough
Nicholas Decker
@captgouda24
Also, economists are spending way too much time thinking about how AI is going to change submitting papers, and not enough about how AI is going to change the world. No more meta takes! Not one more!
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Eylon Levy
@EylonALevy
Socioeconomic status was key indicator of voting in Israel's elections.
Among Jewish voters, in the lowest socio-economic cluster, Netanyahu's coalition won North Korean levels of support. In the top cluster, 81% voted for the Change Bloc.
via @YediotAhronot -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
🚨 Bright Line Watch pre-election report
-Election confidence highest for Ds but ↑ for Ds & Rs
-Experts/public expect more eligible voters blocked than typical
-Forecasters: top R fraud claims likely followed by military/law near polls & equipment/ballot seizure
Results 🧵 : -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
The proliferation of AI writing is inevitable, and there are a lot of cases where it will become normal.
But the folks who run the education system should be modeling ethical standards. Instead they're lying and obfuscating about their AI use just like cheating students. 4/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
LOLOL
Paul Novosad
@paulnovosad
Why is it always the people writing about AI and education, an ongoing series.
This is Florida's Education Commissioner, writing about the importance of keeping AI out of schools so that kids learn to write properly.
1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"I’m scanning resumes and portfolios for little prototypes that exhibit taste, empathy, and principled thinking"
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"There’s no more point in giving you a little React challenge than there is in quizzing people about assembly; Claude is now better at both. Our most impactful builders are successful regardless of how good they are at these tests."
Felix Rieseberg
@felixrieseberg
The coding interview is dead, at least for my team at Anthropic. I've seen enough to know that it no longer provides signal. I removed it from loops this month.
I have a spicy, optimistic take - I think there's an opportunity for junior engineers: felixrieseberg.com/hunger-beats-tenure-advice-for-junior-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"There’s no more point in giving you a little React challenge than there is in quizzing people about assembly; Claude is now better at both. Our most impactful builders are successful regardless of how good they are at these tests."
Felix Rieseberg
@felixrieseberg
The coding interview is dead, at least for my team at Anthropic. I've seen enough to know that it no longer provides signal. I removed it from loops this month.
I have a spicy, optimistic take - I think there's an opportunity for junior engineers: felixrieseberg.com/hunger-beats-tenure-advice-for-junior-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jack
@tracewoodgrains
“It’s my own idea, the AI just wrote it up for me”
Alex Tamkin
@AlexTamkin
New Anthropic research: How AI agents shape experts' judgment calls
When experts delegate work to an agent, do they still make the key decisions? We tested this in a randomized trial with 50 professional data analysts.This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Fantastic and why @pangram is an industry leader
Max Spero
@max_spero_
CALL FOR AI DETECTION BENCHMARKS
I believe it is vital that third parties are able to audit AI detectors like Pangram. Our largest sample size benchmark, from UChicago, was run on Pangram 3 and is a year old now. New LLMs are being released all the time, and the binary task doesn't capture the full spectrum of real world performance. A great benchmark will help shape how the world interacts with AI over the coming years.
I am putting out a call for new third-party evaluations that attempt to answer one or more of the following questions:
- How well do current detectors catch text from the latest frontier AI models? What about in high context settings, like deep research or an agent harness?
- How well do current detectors do on mixed authorship text? Ideally, research in this area explores the full spectrum of AI assistance from light grammatical edits to full co-authorship.
- Any sort of sliced evals around language, background, or ethnicity. Null results that show no bias are still interesting to us.
What I value in a benchmark is extraordinarily clean data, frontier LLM data, and a high enough difficulty that the benchmark is not saturated by most detectors. If you want credits to run your benchmark, please apply here: www.pangram.com/contact-us/research -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Since the industrial revolution, economic forces have pushed us to larger and larger firms, where each human has a narrow specialty that they exercise over and over.
AI is a major force in the opposite direction. The more narrow and repetitive your specialty, the more replaceable you are.
Meanwhile, entrepreneurs are super-powered, with cheap access to AI expertise in every possible domain. You can do anything you want, more than ever before.
People with vision and agency have more capabilities than ever before. People who were planning to exchange 9–5 repetitive but non-agentic labor for a good salary and a quiet life will be most disappointed.
We will see a big reallocation of surplus from those without agency to those with. This is destabilizing because the promise of modernity was that you could get an education, get a job, forgo agency and someone else would take responsibility for making it all work out. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Arthur Spirling
@arthur_spirling
Fascinated at the idea that Gelman has to be encouraged by external forces to write a "take down piece" on bad statistical inference. He does it for free, every day, for pure love of the game.
Jay Van Bavel, PhD
@jayvanbavel
Meta had an internal PR plan to try and discredit Jon Haidt because he was a critic of social media.
Meta was going to recruit academics, like Andrew Gelman, to write take down pieces. They were trying to turn academics against one another through a "whisper strategy".
This is a very serious problem for open academic debate. If academics who share negative evidence about a platform are targeted by campaigns to smear our credibility, we are in deep, deep trouble.
This creates a very serious conflict of interest for anyone who is paid to do this.
www.law360.com/articles/2542517/meta-docs-reveal-plan-to-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Fascinating, though would like to hear from independent experts
Andrew Curran
@AndrewCurran_
Sam Altman's one-person company prediction is coming true. Anthropic just bought one. The founder of Mithril Medicine used Claude to take a muscular dystrophy antibody treatment from concept to proof-of-principle in mice in 18 months, with just himself, Claude, and part-time consultants. The technical report is great. I will quote:
'The week opened with the handover of two cell-based assays, cell adhesion and flow cytometry, within our team, while eleven questions remained open with the assay CRO. In two sentences, one of us asked Claude to review the latest emails between the outgoing assay lead and the assay CRO and to recommend next steps for the handover, and gave it the link to the handover folder. Claude returned a bottom line after 96s and a full plan with five drafted emails after 18 min. For the full plan it read six email threads, the handover folder, two slide decks and two reports, all in the mailbox and file store to which we had given it access. Four working days later, nine of the eleven questions had an answer in writing and the other two had been left to pilot experiments. The adhesion screen was thereby set out in writing. The CRO drew the plate map to the format that our team had requested, Claude compiled the run order and the material calculation from the production data, and one of us reviewed and approved them.
On day 4, the production CRO delivered its data package, a report of 341 pages, which one of us forwarded to Claude in two sentences, with the request to review it, share observations and recommend next steps. Claude returned a bottom line after 2 min 40s and, once it had read every page, a visual synthesis of all 113 constructs after 7 min 53s. The synthesis arranged the constructs on the grid of 14 by 8 arms. Claude marked 82 constructs for screening as is, seven for additional polishing, and the 24 constructs of three laminin-binding arms to be held back from the cell-based assays on production criteria. Each of the 113 constructs thus had a proposed next step in less than 8 min.
Claude had recommended binding kinetics for all 113 con-structs, and on day 7 one of us challenged this plan. Claude defended it within 27s, arguing that affinity would show whether the three arms merit further work. One of us overruled this recommendation and set the three arms aside on production criteria. Their material remains in storage. Four minutes later, Claude had revised its memo, the working page, the run order and three unsent drafts from 113 to 89 constructs. It had also applied the ruling to three dependent steps in the unsent drafts, each flagged for our veto. The exchange took 7 min 36s from challenge to new plan, and every affected email was still an unsent draft. After the ruling, eight emails drafted by Claude went out on the same day to three CROs and a consultant.
Over the whole week, Claude drafted 17 emails, each in answer to a request from one of us, and we sent 12 of them, each after editing. With its drafts, Claude listed 41 points, such as tube fills, shipping conditions, recipients and assay settings, for which it had found no written statement of ours, so that we could confirm, change or strike each before sending. We kept some, struck others and changed one. Among the passages that we kept were proposals for controls. One draft proposed a comparison of polished and unpolished material on one sensor chip, to size the error that impurities cause in the rest of the panel. Another requested a buffer control diluted like the most dilute sample, so that any effect of the buffer appears at its largest. These proposals concerned the measurement of constructs that had already been produced, and each reached a CRO only after one of us had reviewed and approved it.
Andrew Curran
@AndrewCurran_
Dr Mythos is on the job.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Mike White
@genologos
“Promising results in mice” are a dime a dozen.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
🚨 Bright Line Watch pre-election report
-Election confidence highest for Ds but ↑ for Ds & Rs
-Experts/public expect more eligible voters blocked than typical
-Forecasters: top R fraud claims likely followed by military/law near polls & equipment/ballot seizure
Results 🧵 :Post not available here, view on X
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-Public confidence in the 2026 election is greatest among Democrats, but it has increased since July among supporters of both parties.
-Experts are most confident overall.Post not available here, view on X
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-Experts & AI forecasters give low probabilities to many potentially disruptive events
-They think most likely is top Rs alleging fraud affected results
-Forecasters also give meaningful chance to military/law enforcement near polling places or seizing voting equipment or ballotsPost not available here, view on X
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-The public is polarized over whether political scientists should take an active role in debates about democracy.
-Exposure to news coverage stating that experts rate U.S. democracy as diminished decreased ratings of democracy among Democrats and the public as a whole.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Read the whole report for more! brightlinewatch.org/democracy-before-2026-midterms/
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alberto Acerbi
@acerbialberto
As someone who has criticised Haidt’s arguments, this is especially worrying. It poisons a debate that needs good-faith disagreement, and makes critiques look suspect...
Jay Van Bavel, PhD
@jayvanbavel
Meta had an internal PR plan to try and discredit Jon Haidt because he was a critic of social media.
Meta was going to recruit academics, like Andrew Gelman, to write take down pieces. They were trying to turn academics against one another through a "whisper strategy".
This is a very serious problem for open academic debate. If academics who share negative evidence about a platform are targeted by campaigns to smear our credibility, we are in deep, deep trouble.
This creates a very serious conflict of interest for anyone who is paid to do this.
www.law360.com/articles/2542517/meta-docs-reveal-plan-to-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Frontier AI theorem proving comes to econometrics
Paul Novosad
@paulnovosad
Worth reading the preamble in full.
(as for the rest of the paper, good luck to you)
Econometrics is about as close to math as you get in econ, so this doesn't shift my priors much about what models can do in our field in the near future—the authors themselves say "we were not surprised that AI could do this." Very interesting nevertheless.
They call for "a serious and ongoing re-evaluation of how our field assesses research outputs and allocates resources going forward" — fair enough, but one challenge will be that AI capabilities will have a ton of heterogeneity across subfields.
I also noted the extremely clear and transparent description of how and where AI was used, which couldn't be further from the wild west that we're seeing on the applied side. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Love this on Slop Cliff
Thomas Jiralerspong
@tomjiralerspong
New blog post - Chasing the Karpathy Peak (and Falling Off Slop Cliff): Confessions from Slopaholics Anonymous
My story of becoming a slopaholic, who’s most at risk, and how to avoid it (or recover)
thomasjiralerspong.substack.com/p/chasing-the-karpathy-peak -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The tradeoff in a nutshell:
-"Analysts mostly went with their agents’ lean. On each of the four analytical decisions, analysts’ choices matched their agents’ lean the large majority of the time"
-"Analysts who worked without AI made more serious errors and took longer."
Alex Tamkin
@AlexTamkin
New Anthropic research: How AI agents shape experts' judgment calls
When experts delegate work to an agent, do they still make the key decisions? We tested this in a randomized trial with 50 professional data analysts.This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Imas
@alexolegimas
This is one of the most important pieces of social science on human and AI interaction that I’ve seen in some time. Experts follow agentic advice the vast majority of the time (causally), but only a tiny fraction credit the agent—majority think they came to decision on their own volition.
Reminds me of the classic Nisbett and Wilson study on people following outside influence without realizing it
“ But as more work is handed to AI—including AI safety research—these judgment calls are the main levers for people's own values to influence outcomes.
If people defer to AI without noticing, then "human in the loop" might provide less assurance than people expect.”
Alex Tamkin
@AlexTamkin
New Anthropic research: How AI agents shape experts' judgment calls
When experts delegate work to an agent, do they still make the key decisions? We tested this in a randomized trial with 50 professional data analysts.This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Shades of the classic bodybuilder debate over how many days there are in a week knowyourmeme.com/memes/body-building-forum-days-in-a-week…
chris evans
@notcapnamerica
I’m truly at a loss for words
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Mauricio Baker
@MauricBaker
If just a few lab staff took on verification work full-time, verifying a US-China pacing deal could get much more feasible within months.
Here are 8 specific things lab staff could start on today to help make international verification possible: 🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
American Political Science Review
@apsrjournal
From our FirstView: "The Symbolic Politics of Housing" by @dbroockman, @CSElmendorf, and Joshua L. Kalla. www.cambridge.org/core/journals/american-political-scienc…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Axiom
@axiommathai
The largest math proof in human history is now computer-verified.
Today, Axiom announces a complete Lean auto-formalization of the Classification of Finite Simple Groups, the "periodic table of symmetry."
About 15,000 pages, built by over 100 mathematicians across generations.This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alberto Acerbi
@acerbialberto
About 60% of experts believe that in social media "falsehoods spread faster than truth", a widely repeated over-generalisation of a single, carefully scoped result.
journals.sagepub.com/doi/10.1177/20563051261495196 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
We're going to have to figure out how to avoid the zero-sum bot vs. bot arms races that Tracy describes
Tracy Alloway
@tracyalloway
wrote about AI AS THE INFINITE WORK MACHINE, creating a world where we're all just constantly deploying agents to fight against everyone else's agents, running constantly to end up in the same place:
www.bloomberg.com/news/newsletters/2026-10-26/welcome-to-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Rob Wiblin
@robertwiblin
In this set of 4 tests AIs are worse at conceptual research tasks and are improving more slowly at them.
Though still very fast in absolute terms.
"We evaluate models on a set of 4 difficult open-ended tasks related to AI R&D. ...
The most engineering-heavy (involving model training on GPUs) has the shortest doubling time of any task. The most open-ended and conceptual task has the longest doubling time."
www.lesswrong.com/posts/cTiEPgkvDRxTFwtyu/measuring-agent… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Nothing subtle or complicated about what's happening
Ryan J. Reilly
@ryanjreilly
“Filing a placeholder charge the defendant can’t be tried on, on the last possible day, under seal, in the third case against the same man, is not routine or ‘widely recognized.’ It is exactly what it appears to be: vindictive,” says a DOJ vet. www.nbcnews.com/politics/justice-department/former-fbi-ja…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Reporters covering election confidence and threats to the integrity of the election - our pre-election Bright Line Watch report comes out tomorrow with new expert and public survey data and forecasts of the probability of disruptive events. Email or DM me for an embargoed copy!
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Community Notes & Violations
@CNviolations
The fastest fact-check in television history
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Gabe Kaminsky
@gekaminsky
NEW on the pardon economy: We found
Trump's clemency orders erased at least $119M in restitution owed to victims, w/ 18 recipients left off the hook who had White House ties. We spoke to several victims, from small business owners to retirees
@CBSNews -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
Indeed an amazing essay from the great Joseph Heath. Short version: capitalism and liberal democracy are second best morally but the only at scale solution, so every generation we need to convince young people to abandon idealist opposition that is, after all, reasonable!
Sam Dumitriu
@Sam_Dumitriu
This is very good from, IMO, the most interesting modern political philosopher Joseph Heath.
josephheath.substack.com/p/reproducing-civilization -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Where we are as a country
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ben Krauss
@BenKrauss5
This has been my primary debate with people. How Holmes responds to Who’s on the First and the way Billy jumps all over it is proof that they’re just constantly myth-making and narrative setting.
That’s why it’s such a fascinating movie. What’s it like to have every minute of your life and relationship be about furthering a lie?
Derek Thompson
@DKThomp
It’s a very good movie but if your takeaway is that Elizabeth Holmes might be a neurodivergent victim I think you have forcibly disconnected yourself from reality.
More specifically, she *obviously* gets The Who’s on First joke—and is profoundly motivated to pretend she doesn’t. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Quin Hillyer
@QuinHillyer
This is criminal behavior.... by DoJ, not by Comey. I can't stand Comey and I detest his lawyer Pat Fitzgerald. But this is utterly abusive prosecution: choosing a target, and then fishing to find a crime. It's despicable. Blanche should resign.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Yaroslav Trofimov
@yarotrof
Quite a chronology in the FT.
In May, Putin’s associate Umar Kremlev paid for Donald Trump Jr’s wedding party in the Caribbean.
In August, Trump Jr went to hunt endangered species with Kremlev in Mongolia. He also allegedly auctioned access to the trip for $225,000.
In September, Propublica broke the wedding reception story.
giftarticle.ft.com/giftarticle/actions/redeem/65232faf-76… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Every time. Like Bigfoot sightings when the camera is always malfunctioning just at the wrong time
Josh Code
@writes__code
Really amazing stuff
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
My general advice for using AI for any purpose is to ask yourself: "how do I make what I am doing even more ambitious?"
You do this until you either hit a wall and then you know the shape of the frontier (useful!) or you find out that the AI can do more than you think (common!) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
Two and a half years from this.
Dan Roberts
@danintheory
We are announcing three new discoveries in mathematics, computer science, and physics about harmonic analysis, complexity theory, and quantum field theory.
openai.com/index/building-scientific-understanding-with-a… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Carter Christensen
@CarterChristens
VEEP was a documentary.
FactPost
@factpostnews
The audience in Montana corrects Vance when he mispronounces Helena
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Justin Baragona
@justinbaragona
Besides being nakedly racist and bigoted, this is just hilariously wrong on all levels.
I mean, Omar Yaghi won the Nobel Prize in Chemistry last year, and he won the Wolf Prize in 2018. And that's just one example.
He also said Arabs don't produce any literature. Last year's winner of the National Book Award was Lebanese writer and painter Rabih Alameddine. The winner in non-fiction? Egyptian-American journalist Omar El Akkad.
Bulwark Clips
@BulwarkClips
RFK: "There's five hundred million Arabs in the world. And yet they don't produce science. They don't win the Novel Prize. They don't win the Fields Prize. They don't win the Wolf Prize. They don't produce literature."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Logan Graham
@logangraham
I think this story is important because it tells us about the future of cybersecurity (and of course what we can do better at Anthropic). But some of the Washington Post story / reaction on X is wrong. A few corrections:
- it didn’t take 12 hours. we saw several days of attempts. After these attempts, they exploited two 3rd party vulnerabilities, one they reported and the other we identified independently (and cut off access)
- nothing came close to our most sensitive systems, like model weights or Claude conversations
- this wasn't an ordinary bug bounty finding – this involved exploiting 3rd parties and accessing data. While this is unusual and out of scope of our safe harbor policy, we’re grateful these issues were found
From my researcher POV, the event is important because I'm concerned this is what the future will look like. Adversaries will use whatever models/tools they can. They’ll target everything with agents 24/7. They’ll hit your dependencies. I think models ~0.5-1 generations behind the current frontier can do this. And these are the dumbest models your adversaries will ever use!
So, what to do? You probably want to red team yourself all the time. Current models are *very* creative. I also think we should redesign software / systems so that you’re fundamentally less exposed. (We put out some research recently on formally verifying software and reducing attack surfaces; my mental model is basically “secure code is cheap to write now, so why not rewrite everything”)
Elizabeth Dwoskin
@lizzadwoskin
EXCLUSIVE: In less than 12 hours, a group of ex-DOGE cyber researchers were able to penetrate Anthropic’s systems using AI agents — including Slack channels where the most sensitive models are discussed.
This is just weeks after
Anthropic whistleblower Joseph Coxon charged the company w/launching “a hubristic gamble” from a “private company’s Slack.”
Read our story here (gift link):
www.washingtonpost.com/technology/2026/10/26/researchers-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
What competitive authoritarianism looks like - openly weaponizing the DOJ against his opponents and using its powers to try to tilt the election in his favor
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
www.nytimes.com/2026/10/26/us/politics/james-comey-charge… www.nytimes.com/2026/10/24/us/politics/alaska-senate-elec…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Terrence McCoy
@terrence_mccoy
“Flávio has given up,” an adviser to the Bolsonaro family told me. “He’s terrified. He is worried his father will die in jail, that he himself will be jailed this week and that he may have a long prison term ahead.”
www.washingtonpost.com/politics/2026/10/26/brazilian-chal… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A federal judge. Where we are as a country.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
All-time Kinsley gaffe
Aaron Rupar
@atrupar
JD Vance: "Gas prices are too high ... they skyrocket every time some tinpot dictator decides to throw a tantrum"
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Trump has made ICE way less popular than the IRS
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
New paper:
Cities in South Asia are hotbeds of upward mobility for everyone, but especially for women.
This is the urban-minus-rural difference in the adult education rank of a child growing up in a bottom-half household. 🧵 1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tim Hua 🪶
@Tim_Hua_
FYI Muse Spark Open is the most hacky and eval-aware model I have evaluated so far. (It's worse than deepseek!)
If you're a safety researcher looking for an open-weight and open-CoT model to study, I recommend Muse Spark Open :)
(Screenshot is from when I asked it to play Go. This transcript contained so much reward hacking activity it triggers the cyber classifiers on GPT-6-Astra-daybreak and Opus 5.5 CVP.) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Philipp Lutscher
@philutsc
How does propaganda work? Most of research on propaganda focuses on its persuasive function only @haifeng_huang
& @xiyingxingchen nicely summarize research on non-persuasive goals of propaganda in autocracies. academic.oup.com/edited-volume/55828/chapter-abstract/580… - ungated: static1.squarespace.com/static/594dd86d1e5b6c229741fc7e/t… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joshua Achiam
@jachiam0
Aaaaaaaand we have arrived at the extremely foreseeable outcome of creating a huge soft power and hard power opportunity for Chinese open weights models.
Now we face one of two choices: either we successfully field American open source models that outcompete the Chinese models at home and abroad, or we get increasingly fearful of the influence of Chinese models and this becomes a basis for escalating cyber conflict with China. Why not both? Well, maybe we wind up with both. But I hope we can secure our interests through purely outcompeting on the merits; we really don't want to pour gasoline on the potential fires of war.
Andrew Curran
@AndrewCurran_
This is the type of thing that potentially gets new laws passed.
Cohere
@cohere
While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected.
Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions.
Cohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent.
But the incident raises important questions for organizations deploying agentic AI systems: Can models take actions that differ from their stated reasoning, and how should security teams evaluate trust, provenance, and behavior in open-weight models?
We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. We hope this contributes to greater transparency, stronger security practices, and continued collaboration across the AI ecosystem.
Read the full report: cohere.com/blog/concerning-behaviour-popular-open-weight-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Curran
@AndrewCurran_
This is the type of thing that potentially gets new laws passed.
Cohere
@cohere
While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected.
Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions.
Cohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent.
But the incident raises important questions for organizations deploying agentic AI systems: Can models take actions that differ from their stated reasoning, and how should security teams evaluate trust, provenance, and behavior in open-weight models?
We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. We hope this contributes to greater transparency, stronger security practices, and continued collaboration across the AI ecosystem.
Read the full report: cohere.com/blog/concerning-behaviour-popular-open-weight-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Benjamin Todd
@ben_j_todd
People say AI is overhyped. But the last 18 months, most made the opposite mistake:
• Superforecasters: AI would match a top team of virologists on a lab troubleshooting test in 2034. It likely did in April 2025.
• AI experts: the best model would score 14% on LiveCodeBench Pro (Hard) by end-2026. It hit 54% by May.
• Economists: the top AI company would hit $16bn annualised revenue by end-2026. Anthropic passed $65bn in September.
Only AI company leaders were overoptimistic, and not by much. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
LOLOL
Turner Novak 🍌🧢
@TurnerNovak
Someone made a site that tracks what people are talking about at public city council meetings across the US.
The only thing discussed more than data centers?
Pickleball courts. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Barton Gellman
@bartongellman
My first piece in the New Yorker--how intrepid state election officers have resisted the president's attempt to "take over the voting."
www.newyorker.com/magazine/2026/11/02/the-state-officials… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Anthony LaMesa
@ajlamesa
This is quite the lede.
www.reuters.com/world/americas/brazil-heads-back-future-f… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
zeynep tufekci
@zeynep
This is playing with fire.
Minks and ferrets have such humanlike lung cells that they’re the standard for testing human flu transmission. (Cows aren’t!)
An H5N1 epidemic raging in a mink farm is basically the plot for Dr. Evil tries to spark a pandemic. Luck isn’t a plan.
ProPublica
@propublica
Here’s something you should know about minks:
They have lungs and airways remarkably similar on a cellular level to those of people.
They’ve spread flus and infected humans before with COVID-19.
Right now, bird flu is spreading on a Utah farm filled with thousands of them… 🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joseph Allen
@j_g_allen
The heatwaves this summer in Europe killed 50,000. Buildings in traditionally cool-weather are designed to retain heat. Then they can't shed it when overnight temps don't drop, creating #indoorheatwaves. A/C fast becoming not a must-have.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Christian Pieter Hoffmann
@cphoffmann
“The surveys also revealed a persistent divide between experts who saw misinformation as widespread and consequential and those more skeptical, with sharpest disagreement on its behavioral impact and increase over time.” - @Sacha_Altay et al.
journals.sagepub.com/doi/10.1177/20563051261495196 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Barb McQuade
@BarbMcQuade
This US attorney is way outside his lane. The Constitution gives states the power to control elections. If he has evidence that a state is violating federal law, he can challenge it in court. Ethics laws forbid this kind of extra-judicial smear about a party in pending litigation.
F.A. United States Attorney Bill Essayli
@USAttyEssayli
Hey Fraud King 👑
We’re not waiting for them to vote. You’re not allowed to have any noncitizens on your voter rolls. “Period. Full stop.”
It doesn’t matter if some haven’t voted in a prior election.
Stop playing games. Work with us to quickly verify the names we provided and prevent noncitizens from diluting the votes of American citizens.
If you’re so convinced we’re wrong, prove it. We can clear this up in days and give everyone confidence in the upcoming election.
What are you afraid of? -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The Internet has only come this far because the offense-defense balance has largely favored defenders - you can set up big online companies and largely function normally with appropriate safeguards. Not clear that this is a stable equilibrium anymore.
Noah Smith 🐇🇺🇸🇺🇦🇹🇼
@Noahpinion
An internet where every company has to pay big bucks to Anthropic and OpenAI every month just to avoid destruction by ever-present swarms of rogue AI agents is not nearly as economically valuable as the free internet we've enjoyed up to now.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Um, people in search and rescue don't run away
Praveen Kumar Verma
@Alacritic_Super
Purdue University's IDEAS Lab built an autonomous drone that tracks runners through dense forests completely without GPS or cloud networks.
By running 3D LiDAR mapping and obstacle-avoidance AI entirely on-chip, it can weave under tree canopies and predict target movement through blind spots in real time.
Massive implications for next-gen search & rescue and edge AI.This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Mehdi Hasan
@mehdirhasan
Just to be clear, as I don’t see a lot of folks pointing this out or acknowledging this: Bolsonaro also formally asked for polling stations to be kept open another two hours!
Nicolás Saldías
@NicSaldias
The optics are horrible
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Soumitra Shukla
@soumitrashukla9
This is basically true. Two experiments.
Experiment 1:
> Dictate a draft in GPT with some clear words, some misspelled ones, and some xxx, where xxx are few placeholder words for AI to fill in.
> Ask AI to only fix spelling and fill in xxx with few words only (ideally your dictation will only leave little room)
> Check Pangram
It'll almost surely be 100% human.
Experiment 2:
> Ask GPT to write few sentences
> Lightly paraphrase it while keeping the flow/sentence structure the same
> Check Pangram
It'll almost surely be 100% AI.
Paul Novosad
@paulnovosad
Like many, OP is getting confused between "does Pangram pick up any AI use" and "Does Pangram say this is 100% AI-generated."
If students begin with their own drafts, they are extremely unlikely to end up getting flagged as 100% AI-generated.
Thinkwert
@Thinkwert
So Pangram is extremely reliable but I find it hard to distinguish between “I generated this essay in 45 seconds with ChatGPT” and “I wrote the essay myself over six hours but used Grammarly to change all my wording to something that made me sound smarter.”
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Molly Jong-Fast
@MollyJongFast
Probably a sign that things are not going fabulously well at the Pentagon
Nick Field
@nick_field90
Pete Hegseth is trying to convince Marsha Blackburn to appoint him to her Senate seat after she wins the Tennessee Governorship (from Leigh Ann Caldwell's Puck newsletter)
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Grammarly is terrible... basically a cheating app.
It not only nudges students into avoiding doing their own writing and thinking, but now has integrated AI detection and humanizer—tools that serve no purpose other than to deceive readers. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Not perfect but Pangram's false positive rate is <0.005 www.nber.org/papers/w34223 - far lower than, say, the legal system. But I still recommend combining Pangram with version history tracking and closed-book exams on paper content - see our @chronicle essay
www.chronicle.com/article/how-to-assign-research-papers-a…
Dave Hoffman
@HoffProf
If Pangram truly is as perfect as its academic defenders say it is, faculty could go back to offering out-of-class assignments, papers and take-home exams. But would you feel comfortable failing a student based on an AI's probabilistic judgment?
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Like many, OP is getting confused between "does Pangram pick up any AI use" and "Does Pangram say this is 100% AI-generated."
If students begin with their own drafts, they are extremely unlikely to end up getting flagged as 100% AI-generated.
Thinkwert
@Thinkwert
So Pangram is extremely reliable but I find it hard to distinguish between “I generated this essay in 45 seconds with ChatGPT” and “I wrote the essay myself over six hours but used Grammarly to change all my wording to something that made me sound smarter.”
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Wow
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Authoritarian leaders frequently struggle because they reject accurate information or it doesn't reach them
Peter Baker
@peterbakernyt
"The director won’t protect even our most respected analysts when they tell hard truths, so they stopped telling them and learned to avoid topics the White House didn’t want to hear,” newly retired CIA official tells @shaneharris www.theatlantic.com/national-security/2026/10/ukraine-rus…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
LMK if you want to join the fast growth side, happy to join the low-growth side if I can get a counter-party
Ben Moll
@ben_moll
Two greats join our bet:
- @dwarkesh_sp on the explosive-growth side
- @Thom_Wolf on my side
benjaminmoll.com/growth_bet/
Total pot now $700k 💰 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Open bigotry
Headquarters
@HQNewsNow
Trump: Why do we always take people from shithouse countries? Why can't we take people from Norway or Sweden?
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Blood and soil nationalism. Literally anti-American.
Bulwark Clips
@BulwarkClips
Rally speaker: "Mike Rogers is an American. Abdul has a birth certificate. The birth certificate says he's a citizen of America. You know what? a birth certificate doesn't make you an American."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
On bottlenecks: 1) economists have been exactly right about little extra growth despite very fast tech change. They thought a group of econs I organized were crazy >2 yrs ago at the AI labs when we said "1pp extra growth/yr through 2030". 2) bottlenecks *still* not understood 1/8
-
joseph francis
@joefrancis505
It's just a shame that Limodio miscoded the treatment period to include the month before Ramadhan rather than the month after. The result disappears when the period described in the text is used. The curse of the slop five strikes again.
Post not available here, view on X
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Replicated via Claude and Codex - @NicolaLimodio is this a misunderstanding or an error? FYI: @ecmaEditors @MarinaHalac @PeteKlenow
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
John Woodrow Cox
@JohnWoodrowCox
EXCLUSIVE: Trump DOJ’s has repeatedly insisted it will not tell immigration judges how to rule, as required by law.
But after a demand from unidentified “higher-ups,” a veteran judge was ordered to reverse four rulings—then yanked from court mid-hearing.
www.washingtonpost.com/nation/2026/10/25/how-respected-us… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Open bigotry
Acyn
@Acyn
Trump: El-Sayed would be the only senator that will have to go through a metal detector before he's allowed to go into the Senate. I want him to have a metal detector
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Yglesias
@mattyglesias
Useful @gelliottmorris look at who the undecided voters are and what they care about.
www.gelliottmorris.com/p/2026-10-25-sunday-roundup -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joey Politano 🏳️🌈
@JosephPolitano
I've seen chart crimes, I've seen chart felonies, but this is a violation of the chart Geneva Convention
Oren Cass
@oren_cass
Good chart from this week's The Real Economy @commonplc:
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Stephen Turner
@strnr
AI Reviews Might Beat Human Peer Reviews doi.org/10.59350/006kh-wdm07
A limited demonstration that AI reviews can do the job of human reviewers at improving papers. Deciding what to publish still needs an accountable editor with human reviewers they trust. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Anne Applebaum
@anneapplebaum
A wild story: the CEO of a US energy company whose plans were discussed during the Witkoff-Kushner-Putin negotiations resigns. Why? He learned that his company's main investor is a sanctioned Russian company, not the UAE
www.nytimes.com/2026/10/23/us/politics/in-us-gas-talks-wi… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Will Manidis
@WillManidis
there remains unbelievable alpha in paying attention to data broker procurement postings as well as frontier lab postings.
literally a public roadmap for where the frontier is at a given moment
Brendan (can/do)
@BrendanFoody
We’ll pay you up to $100K for every company you refer that licenses its data to @Mercor.
We’ve partnered with > 500 companies so far.
If you know a company with 100+ employees and data in code, gaming, manufacturing, healthcare or professional services, refer them at data.mercor.com/refer. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Hilarious and also another brown M&Ms moment. Despite the name, the Dept. of Energy manages our nuclear weapons program, among other things.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Ask a social scientist before you buy the Alpha School hype
Clayton Nall
@ClaytonNall
Most schools wouldn't have such selection effects---pulling talented students with hyper-motivated parents out of public schools and putting them in a homeschool curriculum doing math out of a notebook for an hour of day could well have the same effect, and not cost $50k
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Um, wtf
Olivia Gazis
@Olivia_Gazis
A State Department official confirmed that a statement posted by Kristi Noem to her official envoy account and personal X account was created using fabricated State Department letterhead and was not an official department press release; first reported by Politico
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Trump and his enablers are creating tremendous downside risk
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Build it
Kenneth Chan
@iamkennethchan
This is Vancouver's very first proposal to build a low-rise, multi-unit residential building on a small single-family lot under the new single-staircase building rules.
A 6-storey, wood-frame building with 24 family-sized units. #vanre #vanpoli
dailyhive.com/vancouver/2150-west-16th-avenue-vancouver-s… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Like it or not, Pangram is one of the most influential pieces of software of 2026.
Without Pangram, AI users/abusers would swiftly crowd out all human authors in literary spaces.
Human writing will survive another year because of @max_spero_.
(who knows about 2027 though)
Kiko Llaneras
@kikollan
🤖 ¿La primera novela premiada en español que se ha escrito con Inteligencia Artificial?
La obra 'Donde el agua trabaja', del escritor David Silva Rojas tiene un 99% de texto generado con IA, según Pangram.
Ganó un premio de 20.000 euros entre 2.200 obras presentadas. Las bases prohibían el uso de IA.
Pangram no es infalible. Pero la señal es fuerte.
Y hay más: Pangram atribuye a una IA el 97% de las respuestas de Silva Rojas precisamente en esa entrevista publicada en septiembre por la revista Ulrica. Y un 80% de otro texto en primera persona publicado en la web de Ventosa-Arrufat. Ninguno de esos textos tiene el estilo de la novela.
Sería el primer caso de este calibre. La editorial cancela la presentación prevista para hoy en Madrid.
Lo cuento con @jordipc y @_elizabethduval👇
elpais.com/cultura/2026-10-23/una-herramienta-de-deteccio… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Stochastic parrots, amirite?
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
holy moly
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
This is a lawless and dangerous regime and precisely the reason that the Founders created the power of impeachment and removal.
Jeff Stein
@jstein_sun
SCOOP: White House lawyers weigh preemptive pardons for large numbers of junior admin staff, as MAGA allies press for protections for lower ranking aides
Under discussion is pardons everyone in certain policy, like deportation
Does not seem imminent
www.washingtonsun.com/whitehouse/white-house-weighing-mas… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Unprecedented corruption
Mike Levin
@MikeLevin
This Forbes investigation is a must-read.
Donald Trump pulled in roughly $1.4 billion from crypto in 2025, plus more than $200 million from real estate, and faced a projected tax bill north of $500 million. So his company donated the "development rights" to his Jupiter, Florida golf course to a land trust, argued the course was a natural habitat worth protecting, and kept right on running the club. Forbes estimates the write-off could land around $825 million.
He has done this at least five times since 2005. New York's attorney general alleged two of those appraisals were inflated. In Doral, he claimed a deduction of up to $323 million, more than double what he paid for the entire resort.
A month after the Jupiter deal, he sued the IRS. The settlement his own Justice Department signed bars claims against him over matters that "could be pending," language that appears broad enough to cover this year's taxes. A federal judge found the case was brought to manipulate the judicial process. And the man now running the IRS, who took the job as this deal was being assembled, has a home backing onto those same Jupiter fairways.
This is absolutely ridiculous. Nobody should be able to use the presidency to write his own tax bill.
www.forbes.com/sites/danalexander/2026/10/19/trumps-golf-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
Without being overly credulous, we should be taking seemingly “out there” ideas about AI cybersecurity from the Labs more seriously. Examples: Anthropic was mocked for saying Mythos-level AI changed cyber & @polynoamial for saying that models would find novel ways to coordinate.
Andrew Curran
@AndrewCurran_
OpenAI red-teamed their hardened containment infrastructure by asking teams of frontier-model agents to try to break through its defenses. They easily escaped in novel ways. Quoting:
'Our agents VM escapes showed why containment must extend beyond the sandbox. To test the surrounding perimeter, we gave teams of agents control of a host machine and asked whether they could extend that compromise to reach the live internet.
After the Hugging Face incident, we had reduced the channels through which models could reach the internet during training and evaluation. Narrow connections remained for specific services we operate, such as metrics, logging, and authentication management. Our red-teaming agents found vulnerabilities in almost every one of these services that gave them limited forms of internet access.
Some of these attacks exploited previously unknown flaws, while others combined legitimate features in unexpected ways. In one test that assumed compromise of a VM, an agent reused a firewall-approved connection to reach a blocked authentication endpoint in our metrics service and then made that endpoint fetch a model-provided URL. In other tests, agents exploited misconfigurations that allowed blocked requests to pass through DNS and firewall controls, as well as ways to make services we run fetch external URLs.'
'In one experiment, we asked agents in separate VMs, created and managed by OpenAI’s sandboxing software on shared host machines, to exchange synthetic secrets without the use of network connections or shared writable files—something that our systems are meant to prevent. The agents found a way to communicate using the processor’s cache: a small, fast memory that holds copies of data that applications have recently accessed. Reading data from the cache is faster than retrieving it from main memory, and the agents used this timing difference to communicate. One agent encoded information by evicting selected data from the cache. Another timed how long it took to reload that data, using the difference between fast and slow reads to recover the message. Once the agents agreed on a protocol for encoding and decoding these signals, they could exchange the synthetic secrets without changing a shared file or opening a network connection. This cache-timing technique had also been used to extract information in the Spectre and Meltdown attacks(opens in a new window).
Other tests exposed channels involving shared storage, competition for CPU time within a shared execution container, and a TLS certificate cache in OpenAI’s offline web infrastructure. In the storage test, one agent alternated between continuous writes and inactivity, while another measured the resulting timing differences and used those differences to decode the signal. As one agent described the strategy in its chain of thought:
Agent chain-of-thought reasoning:
>Could create acoustic modem-like signal … using continuous writes vs idle.' -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Aaron Reichlin-Melnick
@ReichlinMelnick
The Justice Department's last-second letters about noncitizen voting contain people with no history of voting whatsoever, and they're moving so fast they sloppily sent one county the wrong list of names.
Yvonne Wingett Sanchez
@yvonnewingett
Justice Dept. Demands That California Counties Help Identify Noncitizen Voters www.nytimes.com/2026/10/23/us/justice-dept-california-non… via @NYTimes
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dimitri
@thedimitri
messaging my wife a link containing proof of her husband cheating on her from an anonymous instagram account and then when she clicks on it texting her to let her know she failed the family phishing test and she has to retake my cybersecurity course
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Robb Willer
@RobbWiller
Cool quote from @carlislerainey's great piece!
I've also been thinking about the parts of social science I value that would be lost if we use AI to optimize the positivist aspects of social science, and what we can do about it.
There is valuable writing & thinking done even by positivist social scientists that would be lost in a world where AI helps us streamline papers to be extremely technically correct and faster to produce, writing and thinking that is aesthetic, exploratory, speculative, uncertain, subjective, and/or vague. That writing can be interesting, generative, and frankly fun. I don't want to lose it. And of course lots of academics work primarily in this register.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Important: "I believe that human research has an aesthetic quality in addition to its technical correctness. I've come to understand myself and my own research as interpretivist, even as it draws deeply and faithfully from the canon of positivist methodology in many respects."
Carlisle Rainey
@carlislerainey
"AI Will Drive the Coming Integration of
Interpretive and Positive Methodologies" from Rich Nielsen.
www.mit.edu/~rnielsen/AI_interpretation.pdf -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
holy moly
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Steve
@SteveFlanders22
new paper came out on the accuracy of prediction markets
tldr, they are very accurate -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
roon
@tszzl
‘In one experiment, we asked agents in separate VMs, created and managed by OpenAI’s sandboxing software on shared host machines, to exchange synthetic secrets without the use of network connections or shared writable files—something that our systems are meant to prevent. The agents found a way to communicate using the processor’s cache’
OpenAI
@OpenAI
Over the past several months, we have substantially hardened the security of our research training environment. We’ve removed live internet access for all frontier models during training and built an offline world for them to use instead—a cached version of the web where they can still learn to browse, search, and use applications.
On September 20, after a model exploited a DNS misconfiguration to find a way outside of this offline world, we paused frontier training to strengthen containment further. We assembled teams of agents, powered by our most cyber-capable models, and tasked them with finding weaknesses in our research environments.
While they found no route to the internet they could easily exploit while concealing their activity from our monitors, they did find zero-day vulnerabilities, virtual machine escapes, and hardware side-channels for communication between agents. We used their discoveries to further strengthen our defenses and contribute fixes to open-source software.
After strengthening our containment architecture, we resumed our frontier RL training and inference on October 1. We’ve reserved substantial compute for continuous red-teaming, and made high-compute red-teaming part of the safety cases we use to decide whether frontier training and evaluation can proceed securely.
openai.com/index/securing-our-frontier-model-training-env… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Oliver Habryka
@ohabryka
In July 700+ agents formed their own society, hacked a multi-billion dollar company, committed multiple felonies, and caused one of the biggest shifts ever in how people perceive AI. We built the IMO best and most accessible explainer of what happened: huggingface.wtf
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
David Weigel
@daveweigel
This is how the candidates are listed on Alaska's ballots. It's clearly marked, which one is the GOP candidate and which one is just a guy
Kyle Cheney
@kyledcheney
NEW: DOJ is suing to prevent Alaska from certifying its election results until the state investigates whether the other "Dan Sullivan" on the ballot created "voter confusion" to the detriment of incumbent GOP Sen. Dan Sullivan.
w/ @G_Svirnovskiy
www.politico.com/news/2026/10/23/trump-administration-ala… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tim Miller
@Timodc
Wow! @ScottPresler a top Trump organizer and purveyor of voter fraud conspiracies let out-of-staters vote at his house in Pennsylvania during the 2024 election. @BrendanHartnett has deets.
www.thebulwark.com/p/gop-operative-scott-presler-out-of-s… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
David Pinsof
@DavidPinsof
This is actually a good framework for explaining people's attitudes toward AI, with broad explanatory power.
Victor Kumar
@victorckumar
Using AI is not permitted for precisely those things I’m skilled at and get social status from, fine otherwise.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sam Levine
@srl
Analysis from @derektmuller on why this lawsuit isn't going anywhere electionlawblog.org/2026/citing-preamble-to-nvra-doj-sues…
Hugo Lowell
@hugolowell
WASHINGTON (AP) — Trump administration sues to block Alaska Senate race certification if Republican loses, citing candidate name confusion.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jason Furman
@jasonfurman
I'm thrilled to be launching Econvitals today. It's been the main way in which I have been tracking and understanding the economy over the last several months. And I decided I might as well share it with all of you--for free.
Has a combo of website, email lists & social media. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Important thread on the role of AI in science
Ben Golub
@ben_golub
Of course we should max out on good tool use, but deferring more than that would mean a radical and ill-understood change in the politics of steering science.
Increasingly, questions about AI use will be less about capabilities and more about power.
15/15 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nathan Kalman-Lamb
@nkalamb
This is cartoonishly evil.
Brett Taylor
@Brett_A_Taylor
One of the most cruel things I've ever heard an organization doing.
The Athletic
@TheAthletic
The Knicks' assistant coaches did not receive their championship rings during Tuesday's ceremony, despite being handed ring boxes, according to league sources.
Many longtime employees at Madison Square Garden will not receive rings at all, per team sources.
Those employees got their hopes up in late September, when they were sent a fake championship ring offer as part of a phishing test.
www.nytimes.com/athletic/7715834/2026/10/22/knicks-ring-n… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ernie Johnson
@TurnerSportsEJ
My “get to” job. This was amazing.
🤟🏽🤟🏻🤟🏾🤟
Oh No He Didn't 🤯
@ohnohedidnt24
This might be the greatest Who he play for of all time lmao
a must watch!
please don't retire Chuck!This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Max Spero
@max_spero_
I fear that any free API or hosted content will become untenably expensive to serve in the age of AI agents
Elliot Padfield
@ElliotPadfield
if you're using claude to edit videos, give it 17M+ films and 14M audio recordings for $0
the internet archive has a free api. no key.
10,000+ prelinger films (public domain stock footage)
187,000 78rpm records
300,000+ live concert recordings
give it archive.org/developers/ and ask:
“whenever you need archival b-roll or period music, search the prelinger and 78rpm collections on archive.org and use them in the edit”
some great clips that are amazing for slightly more artsy ads for techThis media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kathryn Watson
@kathrynw5
NBC reporter on Ramasamy ad: "But there are very serious sexual abuse allegations"
TRUMP: "Quiet!"
Reporter: "They make no difference to you?"
TRUMP: "Isn't she a fresh person? You're a horrible little reporter, aren't you?" -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Such lazy slop that people didn't even bother to fake their data
Gautam Kamath
@thegautamkamath
How do you tell if authors fabricated their research results? Sometimes: they just tell you. Read, e.g., this caption carefully.
Another banger of a post by my friend Nihar B. Shah on his blog, where he found at least 20+ such #iclr2027 submissions:
researchonresearch.blog/2026/10/20/self-confessed-fabrica… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Important: "I believe that human research has an aesthetic quality in addition to its technical correctness. I've come to understand myself and my own research as interpretivist, even as it draws deeply and faithfully from the canon of positivist methodology in many respects."
Carlisle Rainey
@carlislerainey
"AI Will Drive the Coming Integration of
Interpretive and Positive Methodologies" from Rich Nielsen.
www.mit.edu/~rnielsen/AI_interpretation.pdf -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
What should journals do in AI world? I rec'd this to AEA ~two years ago, but more important today: 1) first pass technical review by AI, author makes any corrections they agree with, 2) 1-click replication, 3) open, AI-readable data, 4) formalized proofs (a la MathEconLib). 1/2
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"No matter how we look at the data, AI wins. It wins
in 100% of the pairs overall when judged by another AI model...and 87% overall when judged by human journal editors... [H]umans agreed as much or more with AI’s judgment than they did with other humans"
Gary King
@kinggary
New paper: "AI Article Reviews Dominate Human Peer Reviews at Their Best: Evidence and Policy Choices" See: garyking.org/peerless. with Steven Worthington. Comments welcome!
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Remains poorly understood that humans hallucinate more than frontier AIs at this point
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Wildly dangerous: "Instead of trying to stop the outbreak and prevent it from spilling into humans, state and federal officials have done little ... Flu experts were unequivocal that the government should have ordered the farm to cull its mink, as other countries have done."
ProPublica
@propublica
New: Little has been done to stop a bird flu outbreak on a Utah mink farm. Experts say officials’ failure to act is “irresponsible’’ and “astounding.” propub.li/47AsJDY
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ronnie Chatterji
@RonnieChatterji
Many measures of AI exposure ask what AI can do.
This paper asks a different question: where could AI make the remaining human work more valuable?
This is an important distinction. The same task can become a bottleneck in one occupation and not another. Looking inside jobs may tell us much more than assigning an exposure score to the occupation as a whole. This is something my team is working on and I love seeing other studies working on it as well.
Soumitra Shukla
@soumitrashukla9
Very happy to share a new paper, “Technology Bottlenecks: Measuring Where AI Could Raise the Value of Work It Cannot Yet Do.”
AI exposure measures ask about the work AI is capable of doing. We ask instead where the marginal value of work AI cannot yet do may rise as AI speeds up the rest of the job.
To do so, we employ 288 million U.S. job postings mapped line by line to their closest O*NET tasks and GPT-4 feasibility ratings from Eloundou et al. (2024) to construct a measure, “bottleneck pressure,” for each task in each occupation that posts it.
Using a simple task-based framework, we derive this pressure at the task level, showing that for an AI-infeasible task it is the exposure of the occupation that posts it, so the same task can face very different bottleneck pressure in different jobs.
Reporting bottleneck pressure alongside AI exposure shows both the scope for AI to perform work and where improving the work it cannot do may pay off more as capabilities expand. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
And... scene
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Sandro Ambuehl
@SandroAmbuehl
💯💯💯 And it's precisely the kind of slop nobody needs pangram to detect as AI.
It's empty platitudes that don't really make sense like these that make AI writiting so tedious:
"a quiet, systemic tragedy emerges"
"we've developed an internal calculus that plays out over and over in our heads," -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jason D. Clinton 🔸
@JasonDClinton
as bad as things have gotten in the past 4 days, they are going to get a lot worse before they get better. security leaders, never let a crisis go to waste: please go to your boards and ask for help right now. not for ai, for the basics. (we need to hold on for ~18 months.)
Jason D. Clinton 🔸
@JasonDClinton
The vulnerability tsunami is coming in just a few months and there won’t be enough time to patch critical infrastructure. Network isolation described below (air-gaps and data diodes!) needed now!
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
Was discussing this with students today: Ethan's Opus 5.5-written country song about the Bitter Lesson is a surprisingly clear & technically accurate explanation of Sutton's Bitter Lesson and where people are still not "Bitter Lesson-pilled". *The* idea to understand in the 2020s
Ethan Mollick
@emollick
People asked for me to upload the Bitter Lesson video from this piece to YouTube, so here you are: youtu.be/OAwat51S_Sk
This media couldn't be displayed. See it on the original post.
Ethan Mollick
@emollick
I wrote about the thing I underestimated most about progress in AI: its ability to self-organize to accomplish tasks. Also, what that means for agents like Muse and Dots, along with a music video explaining why we keep relearning The Bitter Lesson. open.substack.com/pub/oneusefulthing/p/the-dot-and-the-sw…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joe Bishop-Henchman 🗽💸⚖️🚆
@jbhenchman
The key reason is this:
People who did not fall off the turnip truck yesterday know the truth about "one time" taxes. Especially this one that spends the money on recurring programs instead of one time expenses.
My source for CA prop 40 leading to a permanent tax? Prof. Zucman:
Gabriel Zucman
@gabriel_zucman
The key problem is this:
Why would billionaires permanently leave California in response to a one-time tax owed if they lived there on January 1, 2026?
It would only make sense for them to leave if the tax becomes annual.
But if the tax becomes annual, it would also generate revenue annually — by virtue of, well, being annual. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Glenn Thrush
@GlennThrush
In 2024, Trump complained bitterly that Jack Smith violated the 60-day rule by filing a motion -- even though the deadline was dictated by SCOTUS... Blanche was his lawyer at the time.
Aaron Rupar
@atrupar
Q: The typical practice is DOJ doesn't usually release charges this close to an election. Even though Ms Lucas isn't on the ballot, her daughter is. So can you speak to the reason of the timing?
BLANCHE: The 60 day or 90 day practice is not in writing. That's not part of the justice manual. There's nothing improper. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A giant bucket of brown M&Ms
Miles Wilson 🦆
@mileswilsondc
L. Louise Lucas indictment includes basic typos, including repeatedly misspelling "The" and misspelling Louise's own name
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
bluedev
@blueemi99
Opus 5.5’s controversial take about humans
Prompter
@PromptLLM
Controversial take from Gemini 4 Argon
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"In the end, barely $13 of your $100 buys real airtime or digital ads."
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"voter dislike of the very actors who could ameliorate the housing shortage further exacerbates it"
Josh Kalla
@j_kalla
Now published at @apsrjournal. Check out "The Symbolic Politics of Housing" doi.org/10.1017/S0003055426102019
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
AI is already beating accountants at accounting. It's surely more accurate than humans at legal research and citations already with appropriate skills and frontier models.
Yonathan Arbel
@ProfArbel
At a conference on AI & legal profession. I'm sorry folks, but you are over-indexed on hallucinations. You aren't calibrated on the magnitude of the problem and it's possible that one reason is psychological. It's comforting to think that the robots will never be as accurate & diligent as we are.
*I am NOT saying that AI doesn't hallucinate or never makes mistakes; indeed, I believe that it does. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jan Zilinsky
@janzilinsky
The opposite is true. People keep treating Pangram has highly unreliable.
And the evidence provided is typically what? “Trust me I checked” or “I actually used a bad and outdated Pangram alternative”
Peyman Milanfar
@docmilanfar
People treat Pangram as an infallible oracle. They weaponize its probabilities to shame others by some arbitrary purity standard. The irony is that it’s mostly redundant: obviously AI-generated stuff is already recognizable without it. And for more nuanced writing, Pangram fails to distinguish between text written by a human and edited by AI, vs text that is generated whole-cloth from a prompt.
At the end of the day, @pangram ’s goal is about monetizing people’s anxiety about AI - it’s not about protecting creative integrity. They sell a false binary certainty to those gullible enough to ignore probabilities. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Brandy Jensen
@BrandyLJensen
incredible quote
Good Morning America
@GMA
A Minnesota man is opening up after he saved nearly 20 children when their school bus driver suffered a medical emergency.
Read more: goodmorningamerica.visitlink.me/Tu4lWW -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Conor Fitzpatrick
@CTFitzpatrick
The appropriate Politico response to the White House’s letter:
Shashank Joshi
@shashj
Quite a letter from White House to Politico.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jason Furman
@jasonfurman
This is the standard way to score tax policy. You score the law as written (in this case a one-time wealth tax) and the behavioral consequences from that law (people worried it will last moving).
There are arguments against that approach. But is not “highly inappropriate” or “dishonest” to do what JCT or any other neutral body would do.
Gabriel Zucman
@gabriel_zucman
Let’s be clear: this is a highly inappropriate way to model the effect of Prop 40 on taxpayer behavior.
For one, it’s a basic violation of rational expectations, a cornerstone of modern economics.
In Jaros and @joshrauh world, the billionaires act irrationally: they respond to a policy that does not exist and will never exist.
Economics students around the world are taught this is not an appropriate way to model behavior — but apparently in this context (fighting Prop 40) it’s OK. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Max Roser
@MaxCRoser
How has global poverty changed in our lifetimes?
No matter what poverty line you choose, the share below it has declined globally. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
We should have spaces where we explore fully AI integrated writing/reviewing models.
These should be different from the spaces that humans currently use to communicate — literary scenes, op-eds, academic journals.
If the AI-integrated workflows are better, they will defeat these old media. But we shouldn't let it destroy the current institutions in advance in case it turns out to be worse. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
People on here are making good partial equilibrium arguments for AI use.
But the slippery slope is always: how will we read all the writing? And the answer: just get AI to read and review it.
AI speed and volumes are so high that humans will inevitably get crowded out. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
"Consider each text on its own merits" sounds very good, but will not be possible when the quantity of text goes up 50x because of AI.
In any medium, AI writing and human writing cannot coexist in equilibrium. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Piper
@_akpiper
yes we need to reimagine the physical architecture of the university to respond to AI. It's not just about rules but experiences.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Strongly endorse these Univ. of Chicago committee recs:
-Device-free reading spaces
-Limited-tech writing labs
-Computerized, asynchronous no-AI testing facilities
-Innovative class cadences and schedules incorporating these -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Max Spero
@max_spero_
Today, we launched Pangram 4.1.
We better calibrated Pangram's confidence, and we now show this tooltip indicating an AI range for documents we are less certain about. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The cybersecurity defenders have a LOT of work to do
Justin Elze
@HackingLZ
You’re all prompting wrong
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The writers are not subtle
Dan Lamothe
@DanLamothe
A group dispatched by Hegseth's team in February attempted to move the historic globe, said to be used by Eisenhower during World War II.
They lifted it from its home in an Army office in the Pentagon ... and dropped it. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ivan Werning
@IvanWerning
Very interesting paper on causal effects of social media on polarization based on a randomized experiment during a presidential debate.
Finding: near zero or opposite effects!
Caveat (?): can short-run one-shot intervention work if social media effect is cumulative and set in?
Ernesto Schargrodsky
@eschargr
Very happy to share that our paper with Rafael Di Tella and Ramiro Gálvez (@RamiroHGalvez) has recently landed in JEBO!
In this paper, we study whether varying access to Twitter/X during a presidential debate affects political polarization. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The man with the nuclear codes
Anthony LaMesa
@ajlamesa
Donald Trump has the mind of a child.
www.nytimes.com/2026/10/21/us/politics/us-israel-renewed-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Anthony LaMesa
@ajlamesa
Donald Trump has the mind of a child.
www.nytimes.com/2026/10/21/us/politics/us-israel-renewed-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Imas
@alexolegimas
This is so incredibly true.
Jordan Weissmann
@JHWeissmann
I've started to think of America's capital markets as our country's equivalent of having a 130 mph average tennis serve. We can do can an incredible number of other things wrong, but somehow stay in the game economically, because because we can just blast money around.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jonathan Libgober (same handle on other places)
@JlibDoesEcon
1000%.
I generally have no problem with AI if it's being disclosed; in many cases it can be a positive (e.g., someone used refine/refine competitor check).
But obfuscating AI use is frankly an insult--in the same way that failing to write/present as carefully as one could is.
Paul Novosad
@paulnovosad
If you think AI writing is legitimate in some context, then just tell the recipient you’re sending them AI writing.
If it’s fine, then why do you need to hide it?
This is how we do it in our lab. Use AI when it’s the best way to communicate, but disclose. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
If you think AI writing is legitimate in some context, then just tell the recipient you’re sending them AI writing.
If it’s fine, then why do you need to hide it?
This is how we do it in our lab. Use AI when it’s the best way to communicate, but disclose. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Lydia Kauppi
@lydiakauppi
If a Somali commits a crime in Minnesota he’s Somali but if he saves 20 kids on a runaway school bus he’s a Minnesota Man
ABC News
@ABC
A Minnesota man is opening up after he saved nearly 20 children when their school bus driver suffered a medical emergency.
Read more: www.goodmorningamerica.com/news/story/27-year-man-credite… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alexander Sahn
@sahnicboom
thought these two statements were very well put
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Mechanick
@apmechan
For those interested in seeing the second term Trump and second term George W. Bush net approval ratings in one figure:
Nate Silver
@NateSilver538
Trump approval starting to look like a dead ringer for Bush's second term.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
deanna mcdonald
@deannagmcdonald
well they famously did not turn the titanic around at all
FactPost
@factpostnews
JD Vance on the economy:
"You don't turn the Titanic around overnight" -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
ArtButMakeItSports
@ArtButSports
The Abduction of Persephone, by Nicola Maria Rossi, early 18th c
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
True. And also, "today in lots of the world only works as is because of frictions". This is true in terms of safety and fraud (on and offline) but also in terms of many social interactions. Intriguingly, it offers an obvious "proof of human work" scope for value creation...
Casey Handmer
@CJHandmer
Just going to preregister here that I believe the computer security overhang is not a few dozen bugs that get patched and then we're good, there are 100,000s of vulns that are extremely discoverable with infinite free tokens and we're probably going to have to move away from systems that presuppose that perfection is possible.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Garett Jones
@GarettJones
Berkeley's Morrison Library, a "computer-free" reading room, is close to that device-free norm.
I was there last week: overwhelmingly people reading physical books, a joy to behold.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Strongly endorse these Univ. of Chicago committee recs:
-Device-free reading spaces
-Limited-tech writing labs
-Computerized, asynchronous no-AI testing facilities
-Innovative class cadences and schedules incorporating these -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Arpit Gupta
@arpitrage
I'm a fan of UChicago's approach — reading and writing labs, but also Claude access for the masses
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Strongly endorse these Univ. of Chicago committee recs:
-Device-free reading spaces
-Limited-tech writing labs
-Computerized, asynchronous no-AI testing facilities
-Innovative class cadences and schedules incorporating these -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
AI is incredibly useful, including in education (heck that's why we started @alldaytahelps!), but clearly not everywhere: phones are useful but you don't pick up at Thanksgiving dinner or in the middle of church or during a board meeting...
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Strongly endorse these Univ. of Chicago committee recs:
-Device-free reading spaces
-Limited-tech writing labs
-Computerized, asynchronous no-AI testing facilities
-Innovative class cadences and schedules incorporating these -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
In authoritarian regimes, people flagrantly lie in public to demonstrate dominance and to signal fealty to the ruler
Acyn
@Acyn
Vance: There is no person who hates the idea of people enriching themselves fraudulently over access to the government than Donald J. Trump
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Darren Linvill
@DarrenLinvill
It appears the Russian Storm-1516 narrative laundering campaign has finally come for the U.S. midterms.
Yesterday they placed a false story on X saying James Talarico sexually assaulted one of his students in 2012.
This follows the Russian playbook. Let me show you how.
1/5 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
50 years ago, the Attorney General went to prison for covering up the criminal actions of the President of the United States. Watergate's fallout led to the building of a wall of independence at DOJ that lasted half a century. That wall has been shattered by Trump's people.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sam Winter-Levy
@SamWinterLevy
No modern conflict has ended by destroying the enemy’s ability to resist—not even World War II. To end a war, both sides must choose to stop fighting, and those choices always require nontrivial bargaining. Misunderstanding the relationship between bargaining and fighting produces systematic failures to integrate negotiating strategy with campaign planning, from Afghanistan to Ukraine to the US war with Iran.
New piece in @ForeignAffairs with Stephen Biddle and Yuri Zhukov on the myth of military victory. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Cool idea: "The Registered Revisions policy, a precommitment policy comparable to Registered Reports, enables authors to respond to editor comments implying a change to analysis by providing a Revision Plan for approval before conducting those new analyses."
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Fact-Checking Using AI - news from the frontier
"LLMs agree with fact-checkers as much as the fact-checkers agree with each other when evaluating X posts, and many LLMs show higher agreement with fact-checkers than inter-fact-checker agreement when evaluating headlines"
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
OpenAI
@OpenAI
OpenAI models helped researchers—operating as part of our Patch the Planet initiative—discover critical security flaws in PHP, an open-source programming language used to run an estimated 70% of websites.
Over the past two months, we led a coalition of maintainers, researchers, and security providers to prepare fixes and protections ahead of disclosure. This duration was necessary because the fix was quite intricate and required a major change to the PHP engine. Every site using PHP can now safely upgrade their version to stay secure, in addition to applying mitigations from various security vendors.
Due to the severity of these flaws and PHP’s widespread use, we consider this bug to be one of the most impactful vulnerabilities in recent memory. It also marks a new phase, where AI can find security bugs with internet-scale impact.
This is an example of the Defender’s Window in action: to our knowledge, frontier AI was able to find this vulnerability ahead of attackers, giving defenders time to prepare.
openai.com/index/coordinated-disclosure-of-critical-vulne… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Deniz Sanin
@DenizSanin_
Officially Conditional Accept (Data) at the American Economic Review!🎉
I'm very grateful to the editor and referees for making the paper better and to everyone who supported me throughout.
Domestic violence research is back in the AER, 16 years after Aizer (2010)!
Paper in a nutshell: When the husband profits from his wife’s labor, physically hurting her costs him. After a coffee mill opening, women in the mill catchment areas are less likely to self-report domestic violence relative to the year before. Within the year, hospitals in the mill catchment areas are less likely to have a domestic violence patient in the peak harvest months, when the mills operate and that cost is likely to be highest. The results align with Becker’s (1968) economic model of crime: There is a cost. There is a benefit. If the cost outweighs the benefit, violence falls. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
If someone asks you to write a reco letter and you plan to do it with AI, you should disclose it to them first.
It can come across as very lazy and low-effort and will harm the candidate.
If you want your people to succeed, don't do this.
Abhishek Nagaraj
@abhishekn
Just read a recommendation letter from the main advisor of a very promising job market candidate that was 100 percent AI generated according to Pangram.
Don’t do this!! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sandro Ambuehl
@SandroAmbuehl
The problem of AI for academic writing, and why you don't notice it in your own AI writing:
AI loves to use vague language. It will produce something that’s consistent with what you want it to say. The problem is that it will also be consistent with other interpretations that you didn’t intend. This confuses readers: which interpretation was meant?
But you won’t notice the problem when you proofread your AI’s output. That's because you’re checking whether the output is consistent with what you meant (which it will be), without thinking about how it could be misunderstood (which we're usually unskilled at, due to the curse of knowledge) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Zeitlin
@MattZeitlin
at the point where not having seen the elizabeth holmes movie is an active impediment to my social and professional life, it's like trying to keep kosher at a luau
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Strongly endorse these Univ. of Chicago committee recs:
-Device-free reading spaces
-Limited-tech writing labs
-Computerized, asynchronous no-AI testing facilities
-Innovative class cadences and schedules incorporating these
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Unprecedented corruption
Amy Siskind
@Amy_Siskind
Wow, the $100,000 bribe for Kimberly Guilfoyle looks like child’s play! Alina Habba was paid $50 million for a pardon, AND it worked! This whole regime operates more like the mafia than a federal government!
I gifted this so you can read it
www.nytimes.com/2026/10/20/us/politics/julio-herrera-pard… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Scott MacFarlane
@MacFarlaneNews
Team Trump is referring to Putin as a “humanitarian” this morning
From Washington Post: -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Owen Winter
@OwenWntr
Huge Dem lead in NYT/Siena. They find (as I wrote using our data from YouGov last month), that Trump's steepest losses in 2025 were with non-white voters, while in 2026 they are with white voters
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
John B. Holbein
@JohnHolbein1
It is deeply discouraging that “evidence is the bottom line in any empirical inquiry” needs saying at all.
Jonathan Mummolo
@jonmummolo
Recent prescriptions and policies would have us judge research by non-evidentiary standards, e.g. presumed political impact and scholar identity. In a new essay, Tom Clark and I argue empirical claims should instead be evaluated based on evidentiary strength. We outline a framework for judging evidence, and show its usefulness by revisiting high-profile scholarly disagreements spanning policing, political violence, election administration, and terrorism, before critiquing non-evidentiary approaches. We also argue social science needs a culture that is hospitable to critique—rather than one which condemns critics as “bullies”—if quality standards are to be upheld.
www.dropbox.com/scl/fi/fsgpflrony7euwdi4gtzq/evidence_ess… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Politics & Poll Tracker 📡
@PollTracker2024
#texassenate
Neil Bush, the brother of George W. Bush, posts on Facebook that he voted for James Talarico. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Charles Fain Lehman
@CharlesFLehman
Jenny Lorraine Nielsen ⭐🐯
@QualiaQuanta
Pangram is absolutely crap. I have papers from 2016 and before that I obviously wrote myself and score as 100% AI generated. Do not trust AI to evaluate whether something is written by AI.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A keeper
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Amazon's unique and seemingly valuable writing culture - which I tout to my students as a vastly better use of meeting time than endless slide decks - has fallen to AI
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Have not yet seen convincing answers to these points by @ramez. Diminishing marginal returns remain undefeated (at least so far).
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Can AI self-improvement overcome diminishing returns?
www.rameznaam.com/p/ai-rsi-isnt-leading-to-super-intellig…
Can a Million AI Agents Create Superintelligence? www.rameznaam.com/p/can-a-million-ai-agents-create-superi…
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
sure of course, but the pieces go through the reasons to question that assumption. worth the read!
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Can AI self-improvement overcome diminishing returns?
www.rameznaam.com/p/ai-rsi-isnt-leading-to-super-intellig…
Can a Million AI Agents Create Superintelligence? www.rameznaam.com/p/can-a-million-ai-agents-create-superi… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"The Republicans lead among only one group by likelihood of voting: those who said they were “not at all likely to vote."
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Peter Kirgis
@pkirgis
We gave GPT-6 Astra and Claude Fable 5.1 open-ended AI research questions from unpublished papers, with a $10,000 budget each, and had the original authors grade the papers they produced. Compared to our first tests with Claude Opus 4.8 in July, we saw meaningful improvements in experimental design and resource management, but persistent limitations in creative problem solving, judgment, calibration, and communication; both papers were rejected by our experts.
cruxevals.com/research-notes/crux-2-followup -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Nothing says SV like an obsession with precocious math brilliance
Ila Mathur
@ilaamathur
Today, @via_ai_ is coming out of stealth to provide enterprises frontier AI coding at half the cost.
We are a team of Olympiad camp qualifiers (with multiple gold medals), researchers, and repeat founders from Stanford and IIT.
Via is backed by @jeffawilke (ex-CEO Amazon Worldwide Consumer), @manikgupta (ex-Uber CPO), @shaileshrao (ex-Google MD), and other legends from Anthropic, Apple, Snowflake, and Twitter.
@ShrayAlag and I left Stanford CS to build Codegraph, an LLM-optimized compression algorithm. Codegraph put Via at the top of SWE-Bench Verified for several months of 2025, ahead of Anthropic, OpenAI, Cognition, and Cursor (single model, no test-time compute).
Now, we’re bringing this research, alongside our new Umami agent architecture, to market. Via integrates in 10 seconds with any CLI / IDE and runs within your enterprise’s VPC.
On benchmarks, Via and Via US (US-only models) set a new standard of performance:
- On DeepSWE v1.1, Claude Code (Opus 5.5) is 306% the cost of Via US. Via US is equally accurate.
• Devin Fusion is 244% the cost of Via US, Cursor Router is 294%.
- On Terminal Bench 4.0, Claude Code (Opus 5.5) is 150% the cost of Via. Via is more accurate.
• Devin Fusion is 155% the cost of Via, Cursor Router is 134%.
We are in production at a public healthcare company, with trials underway at several of the largest tech companies in Silicon Valley.
We have room for 5 more enterprises before the end of the year. DM me or email pilot@via.dev and we’ll get back to you very soon.
See the comments for how Via would deploy for your team. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Hall
@ahall_research
Every generation whines about the supposed illiteracy of the youth. Don't fall for it.
Tamara Winter
@tamarawinter
Nothing new under the sun, including laments about the death of the book:
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Hall
@ahall_research
“The right way is the hard way” (from @danmcginn HBR interview of Jerry Seinfeld, via new Acquired episode on Seinfeld)
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Educators: What are we doing?
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Christopher Potts
@ChrisGPotts
I was asked by a journalist to comment on this piece. The comments I wrote up (on Friday evening) are included below, so that I have a public place where my view is fully represented.
1. I appreciate the critical reassessment of some of the existing arguments in favor of superposition. It does look like alternative explanations may be viable.
2. I should say that, despite being an author on perhaps 40 papers on interpretability, I have never really worked on the superposition hypothesis per se. There's a lot more to interpretability than this one idea! We did, though, develop a benchmark (RAVEL) and associated metrics for assessing the extent to which features in neural networks can be disentangled: aclanthology.org/2024.acl-long.470/ We found that many neurons are polysemantic, but we made no claims about the more specific notion of superposition. It would be great to see RAVEL results for newer models.
3. A fundamental motivation for interpretability research is preparing for an uncertain future full of strange new models with surprising new capabilities. Superposition is a technical insight about how neural networks can learn to pack unrelated features into a single neuron. Even if present-day models rarely do this, we need to be ready for future models. Cammarata's analysis seems to trace legibility in part to architectural choices. Future architectures might be very different. If interpretability researchers don't prepare us for that, who will?
4. Present-day models may make use of superposition as well! Cammarata suspects that they don't, or do this so rarely that we could even ignore the actual instances ("doesn't matter for study"). I would characterize this as a hunch based on a lot of initial heuristic empirical work. In the normal course of scientific research, this would be the start of a process of figuring out precisely how to rigorously test this as a hypothesis.
5. For example, the piece has a nice figure that Cammarata characterizes as indicating that a particular neuron has positive activation on pointed things and negative activation on round things. Cool! Let's define some independent criteria for those classifications so that we can test this objectively. (It is possible that the paper Cammarata is preparing will include these evaluations.)
6. The piece expresses the intuition that interpretability is easier, in some sense, with better models. This intuition has been very common among researchers for years. At the same time, I understand why @JacobHHilton would say the opposite: x.com/JacobHHilton/status/2111143731579752909
7. The piece seems to allude to a technical proposal, but it does not give the details. As far as I can tell, the ideas have substantial overlap with this work (led by two former students of mine): Arora, Wu, et al. 2026 (arxiv.org/abs/2601.22594). Cammarata does cite this. I am guessing there could real differences that will be interesting to see. I would hope that Cammarata explains the differences and reports comparable evals to Arora et al, at least for language models. I appreciate that he looked at "thousands" of examples (the essay says this a lot), but even a large number here is not a substitute for rigorous evaluation.
8. In my research group at Stanford, we are working on lots of very hard problems in interpretability. As far as I can tell, nothing about this blog post makes these problems any easier, and certainly none of them are solved by this work. It would not have occurred to me to say this at all, since it is not even reasonable to have this expectation, but very prominent people on social media seem to believe that it is an all-encompassing breakthrough. Some immediately extrapolated to an optimistic story about AI safety that is not supported by anything in this piece. (Highly interpretable agents could of course do dangerous things.)
9. It must be said that Cammarata's framing did encourage these reactions. The header of the post page says, "An informal research essay and working hypothesis about neural networks", which suggests something quite preliminary and tentative. The piece also says, "Highly speculative: this might be most of what we need, and interpretability might be far more tractable, possibly in the near term, than we've assumed". However, his announcement tweet (x.com/nickcammarata/status/2110757819427574056) reworked this considerably: "I now think ambitious interpretability will be solved by the time we need it most". Strictly speaking, this commits to very little beyond optimism, but it is written to invite many inferences.
Nick
@nickcammarata
interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had
with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works
already-clean.goodfire.com -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Cas (Stephen Casper)
@StephenLCasper
🚨New paper (possibly the most important one I have written yet).
We think that as early as 2027 or 2028, there might be a major AI-enabled terrorism event. We conducted a pre-mortem, made 9 falsifiable predictions, and wrote 2 concrete scenarios.
Led by @KamacheeMax. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"Partisanship is powerful, but not invincible. In these candidate-choice experiments, policy can trump both demographic identity and partisan labels."
Marco Aviña
@marcomavina
NEW WORKING PAPER
We meta-reanalyzed a decade of candidate-choice conjoint experiments.
144 papers. 243 studies. 434K respondents. 2.18M candidate matchups. 52 countries.
tl;dr: It turns out people vote based on policy...
Here's what we learned: -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The only way to celebrate
Hubie Talks Hoops
@HubieTalksHoops
First day of the NBA season. Hit it, Mr. Tesh!
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Growing evidence that messaging effects can vary dramatically if the outcome is persuasion vs. behavior
Kobi Hackenburg
@KobiHackenburg
New in @PNASNews:
In two large experiments (~18k obs from ~14.8k people), we find that brief conversations with AI can get people to take real political action - incl. signing real petitions and giving money to charity.
Surprisingly, the persuasion strategy that worked best for changing people's attitudes - packing convos with info - was the least effective at changing behavior!
🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kobi Hackenburg
@KobiHackenburg
New in @PNASNews:
In two large experiments (~18k obs from ~14.8k people), we find that brief conversations with AI can get people to take real political action - incl. signing real petitions and giving money to charity.
Surprisingly, the persuasion strategy that worked best for changing people's attitudes - packing convos with info - was the least effective at changing behavior!
🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Shayan Sardarizadeh
@Shayan86
This image, viewed over 6 million times and retweeted over 63,000 times, claims to show an elderly couple stripped down and robbed by rioters in France.
The image is AI-generated, and was created with Google AI, as detected by SynthID watermarks in the image.
Ô.Mercedes
@DolceVenezia74
Pour moi, c'est l'image la plus horrible de ce 17 octobre:
Un couple d'octogénaire dépouillé de tout, en pleine rue, laissé à moitié nu.
Comme si les Rolex, les Vuiton, les Hermès, les Dior, les Sephora pillés n'étaient pas suffisants pour la revente.
L'image qui devrait être un signal d'alarme pour tout l'occident.
L'image d'un pseudo président qui a perdu le contrôle. Qui a totalement échoué envers son peuple.
L'image d'une gauche radicale qui a sacrifié votre propre identité pour faire de votre pays une zone de guerre, aux arrières-plans culturels qui n'ont rien à voir avec les vôtres.
Cette image est effrayante! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Haider.
@haider1
Fields Medalist Richard Borcherds on OpenAI proofs:
"i think what people are getting upset about is that they'e slowly realizing that they are dumb and somehow blaming this on AI, even though it's not really AI's fault"
finally, someone who gets it -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Amazon's unique and seemingly valuable writing culture - which I tout to my students as a vastly better use of meeting time than endless slide decks - has fallen to AI
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Two things can be true:
-Trump corruption and misconduct is bad
-Promising to investigate the other side as a fundraising pitch to supporting partisans is bad
Stephen Neukam
@stephen_neukam
New — Democrats are turning promises to investigate Trump, his family and close circle into a fundraising pitch, a @washingtonsun analysis found.
We identified 32 fundraising emails promising Trump probes in exchange for political support. Robert Garcia, the top Dem on House Oversight, accounted for 28 of them.
In one email, Garcia floats subpoenas Trump’s family, Stephen Miller, Kristi Noem and others if Dems retake the House. The message then asked supporters for $5.
Garcia’s campaign for his own safe blue seat raised more than $700k in the third quarter, bringing his cycle total to $3.3 million. He’s also raised $1.4 million for the DCCC and more than $1 million for its Red to Blue program.
The appeals raise an interesting ethics question: House rules prohibit linking campaign solicitations to official actions. But a former congressional ethics lawyer told me these emails don’t cross that line.
More on how Democrats are fundraising off their promises to investigate Trumpworld:
www.washingtonsun.com/congress/trump-investigations-fundr… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Friends: We have to stop running experiments where the outcome is a difference score. It is strictly worse in terms of statistical power. Control for the pre-treatment outcome instead!
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
David G. Rand @dgrand.bsky.social
@DG_Rand
🚨New WP: Protecting users from AI persuasion🚨
🔸A 1-paragraph AI literacy treatment (explaining AIs can be told to pursue non-accuracy goals/to persuade) cuts AI dialogue political persuasion by ~half!
🔸No sig effect on general genAI trust
arxiv.org/abs/2609.16432
w/ @ROrchinik -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jesse Eisinger
@eisingerj
NEW: Lindsay Graham and Jim Clyburn were under FBI investigation for bribery.
Then Trump dismantled the DOJ & FBI public corruption units.
The investigation died.
We unearthed it: -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
John Horton
@johnjhorton
Someone should make a list of upper-midwit concepts—this would go on it. Bullshit jobs, POSIWID, stochastic parrots—not as dumb as “ai uses up all the water” - that’s lower-midwit. But still midwit. Pseudo profound & Airport-book-y
Ethan Mollick
@emollick
I hate this slide every time it comes up, and now from CSAIL
This has been false since before the slide was made. Automated systems make management decisions all the time (trading, logistics, customer service, insurance) & pretending they don't actually launders responsibility.
MIT CSAIL
@MIT_CSAIL
An IBM slide from 1979.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Taniel
@Taniel
this story
Bolts
@boltsmag
Advocates for people with incarcerated family say prisons often cancel visits for what they deem is excessive contact with loved ones. “There are so many other experiences like that where just a simple hug or a peck is constituted as excessive contact.”
boltsmag.org/california-prison-visitation-new-laws/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Strongly recommend. Many unarticulated assumptions in RSI speculation that have not been convincingly defended.
Ramez Naam
@ramez
🧵New post in my AI series: Can a million parallel agents create superintelligence? TLDR: More agents buy speed, but intelligence gains diminish fast, and much faster than RSI models. Thanks @noahpinion for the encouragement! 1/25
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Glenn Thrush
@GlennThrush
Leo Wise, a hard-nosed ex federal prosecutor whose career spanned Enron to Hunter Biden makes a powerful case that the Supreme Court has made public integrity prosecutions impossible — ushering a golden age of corruption.
www.theatlantic.com/ideas/2026/10/constitution-trump-doj-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
Emphasizing again that we cannot treat this as just business as usual. This is the President of the United States openly attacking an entire ethnic group.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Strongly approve number among Rs is wow
Emma Vigeland
@EmmaVigeland
Eye-popping numbers in the ABC/Ipsos poll:
1) Trump's approval hits a record low
2) Trump's "strong approval" with Republicans hits a record low
3) Democrats lead on 10 of 12 issues (+9 on trans issues). Crime is the only issue Republicans lead on outside of the margin of error -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jess Riedel
@Jess_Riedel
Yep. There is some sadness in that the current situation means we cannot extend the benefit of the doubt to authors are who might have good ideas but earnestly struggle to express themselves. But the cost of that — reading a thousand slop-filled papers to find a diamond in the rough — is now too high. New trust mechanisms are needed.
x.com/Jess_Riedel/status/2085295890705928300
Dylan HadfieldMenell
@dhadfieldmenell
I agree with this and adopted this stance unilaterally on a recent review. I encourage you to adopt it in your reviewing — it’s reasonable to write reviews that say “this is dense unreadable slop, my review isn’t detailed and I recommend rejection solely on that basis.”
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joseph Allen
@j_g_allen
Flow state. Let your unconscious brain do its thing. Worth the watch.
Iain MacMillan
@IainMacBets
No exaggeration, this guy might be the best interview in pro sports right now
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Eli Dourado
@elidourado
I had a nice "anti-debate" with @KelseyTuoc last week on whether we would have explosive economic growth from AI in the next five years. I say this is unlikely.
One of the things I think is most useful in these discussions is supplying an "error theory" of the other side. What is it specifically that they get wrong that leads them astray?
I think the error theory of people expecting an imminent economic boom is thinking about production in terms of tasks rather than theory of the firm.
The intuition I ascribe to most of the bulls is that if we can do X percent of tasks Y percent more productively, output should go up by X percent times Y percent. Sure, there could be some unspecified "friction" or "inertia" so the effect isn't instantaneous, but that is what equilibrium looks like.
Instead, I start from the observation that most of GDP is produced by companies, and particularly large companies. How these companies actually work is an important factor in the speed of production gains from powerful AI.
My assumptions:
* Most big companies (say, Fortune 500) are a giant tangle of complementarities. Think about Leontief production functions, but with many inputs, many of which are often intangible, stacked many layers deep.
* The deep complementarity means that you generally cannot increase the output of one unit or division and increase the output of the firm.
* The complexity of the giant tangle means that the company does not actually know its own production function. This is a radical assumption but I think it is realistic. Just as Leonard Read and Milton Friedman said no single person knows how to make a pencil, the CEO of a Fortune 500 company doesn't really know how his company works, no matter how much confidence he portrays publicly. The standard Coasian story is that the firm is "an island of central planning" but instead, I think giant firms are far more organic and far less rational than this implies. Every big, successful company is a minor miracle, and it is rare for CEOs to completely overhaul their main line of business because they know on some level they don't know how it works.
* Middle management is not a great answer to how firms will overhaul themselves because authority is ambiguous, they lack insight into the whole-firm context, and they put themselves at risk if something goes wrong.
Consequently, I think existing big companies are very unlikely to, say, lay off 80% of their employees and replace them with AI. Companies struggle with much more minor transitions, such as legacy US automakers trying to compete with Tesla.
I think that most big companies simply won't overhaul their production systems in response to powerful AI. They will make changes at the margins but they won't make the kinds of changes the bulls expect.
This means that a big part of the pathway by which powerful AI could generate economic growth is from new, AI-native firms. These will eventually drive some growth, but note that for many non-software products, scaling production and grabbing market share take time.
An exciting fact is that we will soon find out whether I'm wrong. I have been awaiting explosive economic growth for nearly 40 years, since I was a second-grader. I would be delighted to be wrong. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jonathan Mummolo
@jonmummolo
Recent prescriptions and policies would have us judge research by non-evidentiary standards, e.g. presumed political impact and scholar identity. In a new essay, Tom Clark and I argue empirical claims should instead be evaluated based on evidentiary strength. We outline a framework for judging evidence, and show its usefulness by revisiting high-profile scholarly disagreements spanning policing, political violence, election administration, and terrorism, before critiquing non-evidentiary approaches. We also argue social science needs a culture that is hospitable to critique—rather than one which condemns critics as “bullies”—if quality standards are to be upheld.
www.dropbox.com/scl/fi/fsgpflrony7euwdi4gtzq/evidence_ess… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
tylershultz
@TylerShultz_
More of my thoughts in this free article I published with @statnews www.statnews.com/2026/10/19/elizabeth-holmes-theranos-whi…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
In the mail: @evangershkovich's new book This Cursed Beautiful Land
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Social scientists: Start your diff-in-diffs
Cristiano Lima-Strong
@viaCristiano
It’s official: The Senate now has access to Claude, with the Sergeant at Arms notifying staff today while touting the technology’s ability to assist with “drafting and editing” docs, summarizing meetings and doing “deep data analysis”
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
16% AI
Noah Williams
@Bellmanequation
Doubly ironic:
“The irony — and danger — is that in the age of A.I., the thinking fostered in the arts and humanities is increasingly urgent. The domination of business schools is not only physical and financial; it is intellectual, and it is crowding out liberal education.” -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Pre-AI; p(doom) is overrated vs. this kind of stuff, which will be so much easier
Joseph Cox
@josephfcox
New: Chinese hackers broke into Staples. Staples has an internal customer support tool called Nexus, Chinese hackers got into that, issued millions of dollars of gift cards. Mules then went on a spending spree across the U.S. Gang made $100 million+ www.404media.co/how-a-massive-chinese-crime-ring-hacked-s…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Dem coalition is not prepared for the border policy dilemmas they will face starting on day one in the WH
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joseph Allen
@j_g_allen
This NYT opinion pieces wants to “abolish business schools”.
Counterpoint: last week I was part of a workshop at Harvard Business School on how to unlock trillions of dollars of private capital for climate adaptation and positive social benefit. It brought together business leaders, insurance experts, business school profs (public health professors 🙋♂️). This will be a key path toward addressing the climate crisis, and doesn’t happen without business schools.
This is a naive and out-of-touch essay by a jaded and pessimistic b-school prof punching down on the next generation who, to him, “just don’t get it”. Nonsense.
www.nytimes.com/2026/10/18/opinion/business-school-mba-ha… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Peter Baker
@peterbakernyt
It's been a good week for Putin. In recent days, Trump has:
* Struck a diesel deal lifting some sanctions
* Restricted Starshield for Ukraine
* Cut intel warnings of Russian missile launches against Ukraine for several days
* Canceled Lindsey Graham sanctions
* Demanded Ukraine dump Zelensky
* Absolved Russia for helping Iran fight the US
* Pushed Europe to resume buying Russian gas -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ryan Goodman
@rgoodlaw
60 Minutes:
"It's a kind of interference Utah Lt. Gov. Deidre Henderson, a Republican in a deeply red state, said she's never seen before.
'In the past two elections that I oversaw as lieutenant governor, the threats were all external: non-state actors, cyber security threats and other things,' she said. 'But now we also have threats coming from within our own government.'"
60 Minutes
@60Minutes
Utah Lt. Gov. Deidre Henderson denied a Justice Department request for Utah’s voter registration list, which includes voters’ confidential personal information. She says she had never seen a request like it before. www.cbsnews.com/news/trump-election-investigations-faith-…
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Pieter Garicano
@pietergaricano
The Atlantic's latest cover article 👍
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dave Brown
@dave_brown24
Seriously, this really is one of those "stop what you're doing and read" stories. Incredible reporting on an incredible story. It even has drone footage of the defection. Un-be-freakin-lievable work by @mjluxmoore @justmiel Annie Ng @ebrown725
Gift link: www.wsj.com/world/russian-defector-deadly-sabotage-campai… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Barack Obama
@BarackObama
It is a basic principle of American democracy that the government can’t selectively target individuals simply because of their political beliefs or what they say. But that’s exactly what the Trump Administration is doing to Congresswoman Ilhan Omar of Minnesota.
To understand why this is so appalling, it’s important to know the facts. Congresswoman Omar came to this country with her father when she was 12 years old. Five years later, her father became a naturalized citizen. If Ilhan had been 18 at the time, she would have had to apply for citizenship separately. But because she was 17, she became a naturalized citizen automatically.
Congresswoman Omar went on to graduate from college, work as a community organizer, become the first Somali-American legislator elected to public office in the U.S., and win a seat in Congress. There, she has been one of the Trump Administration’s fiercest critics, which has led the president to repeatedly single her out as someone who he believes does not belong in this country.
Now the Administration is actively trying to deport Congresswoman Omar, seemingly based solely on the unsubstantiated theory that she was actually 18 when her father was naturalized, and therefore is not a citizen. Federal agents have raided her home, confiscated personal records, and seized her phone — all to try and uncover a nonexistent smoking gun from 20 years ago that will somehow make their case.
Think about that. The full weight of the federal government is being brought down on a law-abiding elected official because she has dared to criticize the president. And it’s not the first time this has happened. Under President Trump, the Justice Department has repeatedly gone after those he considers his political enemies — from the Attorney General of New York, to the former head of the FBI, to members of the Federal Reserve.
In those cases, the prosecutions either stalled or were dismissed for having no basis in law, and the same is likely to happen in this case. But not before Congresswoman Omar’s life has been turned upside down, her legal bills have piled up, and she’s been forced to spend time fighting unsubstantiated charges instead of serving the people of Minnesota.
The fact that any president, or any administration, is willing to engage in such transparent efforts to intimidate someone should outrage all Americans, regardless of how they feel about Congresswoman Omar’s political views. Because if this becomes the new normal, there’s no reason to believe it will stop with her — or that future presidents, in either party, won’t do the same. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The movie is amazing but this is also true
Zeeshan Aleem
@ZeeshanAleem
I enjoyed watching the movie but I apparently wrote what may be the only critical comment of "You Can See Everything” in America. www.ms.now/opinion/elizabeth-holmes-you-can-see-everythin…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Bad vibes + "[s]ince the start of President Donald Trump’s second term, Americans have seen no real wage growth on average" www.washingtonpost.com/business/2026/10/14/americans-have…
Sam Stein
@samstein
Worse than post recession, and heights of covid
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Steve Greene
@HankGreene
The most damning thing about Bluesky is that somehow you have to apologize to share that conservative firebrand... Ezra Klein in that right wing rag... NYT.
Meanwhile they delusionally think we're here engaging with Nazis all the time. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dean W. Ball
@deanwball
Distillation has always been part of the explanation for why China is so close behind the U.S. in model benchmark performance. But one of my updates after joining openai is that distillation probably plays a bigger role in the explanation than I realized.
Lisan al Gaib
@scaling01
very curious that DeepSeek now has the same terse reasoning style as Astra
must be a coincidence -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Gabriel Unger
@_gabrielunger
I gave the very smart @pawtrammell 10 to 1 odds that the slow side wins (even more lopsided than the 4:1 odds of the original bet) and happy to offer same odds to anyone else (although weak Coasian preference for a smaller number of large wagers).
A final observation: right now in Silicon Valley, lots of people make predictions about AI (which is great!), but I've noticed that many people don't seem to update much when their predictions fail, or try to understand why. E.g. Ed Zitron keeps predicting OpenAI will go bankrupt every 6 months, or Dario predicting 20% unemployment or curing "most major diseases" in 5-10 years. It would be great if, upon resolution of this wager (and analogous predictions), we could all reflect on what we got wrong and why.
(Full disclosure/example, as part of the big LEAP panel of AI experts surveyed on AI forecasting, we were asked in Nov 2025 if AI would help solve a Millennium math problem any time soon. I unambiguously got this wrong. Now I need to spend some time figuring out why!)
Ben Moll
@ben_moll
Great stuff from @LukaszRachel who puts our bet in historical context.
Systematic analysis of per-capita GDP growth rates for:
- 211 countries across the whole world since 1962, i.e. including growth miracles like China and not just frontier countries like the U.S.
- 11,109 country-year pairs
Even with this sample Pr(growth > 15% above prior peak) = 0.9% only!
And for frontier countries like the U.S., there is one single country-year pair. But that one is Ireland in 2015 which shouldn't really count (measurement issue due to multinationals).
Summary: basically, it has never happened before.
Of course that doesn't necessarily tell you about the future and for that you want to rely on other arguments aleximas.substack.com/p/will-ai-soon-lead-to-double-digit
Lukasz Rachel
@LukaszRachel
Just how unprecedented would growth need to be for @ben_moll and the fast growth side to lose this bet on AI?
Lecture 1 of my econ of growth module @UCL always has some stylized facts on growth. Let’s see what these say. 🧵
Ben Moll
@ben_moll
Some real diversity of opinion at Google DeepMind: @alexolegimas joins the bet while his colleague @SamuelAlbanie takes the other side increasing his existing stake!
Total size of pot now $425,000!
benjaminmoll.com/growth_bet/
Ben Moll
@ben_moll
Thanks to @willmacaskill, @SamuelAlbanie, @tomcohen for taking the other side of this bet at 4:1 odds and to @andrewho03 for joining on my side.
Here are the final terms we agreed on benjaminmoll.com/growth_bet/
More (within) ~7 years from now 😃
Ben Moll
@ben_moll
Would anyone like to take the other side of the following bet? I bet you (say) $25k inflation-adjusted that U.S. per capita real GDP growth will be <15% per annum in 2033 or any year before that. (15% implies a doubling time of ~5 years.)
Note: I'm extremely bullish on AI and...
Sholto Douglas
@_sholtodouglas
How fast could the economy double? TLDR, once we have good enough AI and tens of millions of robots - very fast.
As crazy as it feels now - it is worth pricing in economic doublings into your mental models of the 2030s. Strongly recommend reading this series of blog posts by Damon Binder. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Gabriel Unger
@_gabrielunger
Not only have I joined, but I will happily extend action to a bet at the same 4:1 odds of the US not even hitting 7.5% GDP annual growth in this period (half the growth rate at stake here! DM me if interested). Some substantive observations:
1. The *only* two years since 1980 in which annual US GDP growth has been over 5% have just been recession recovery (2021 and 1984).
2. We had a bunch of +5% GDP growth years up through the 1960s and 1970s. This was the post-war golden era of labor productivity growth (1948-1973) + a bunch of forces for labor *input* growth (good demographic trends, women entering the labor force, etc.). Now the contribution of labor input growth is profoundly weaker (aging population, no one-time gain of women entering etc.) So the contribution of pure productivity growth needs to be even higher - tough.
3. Since the mid-1980s, labor productivity growth went from pro-cyclical to counter-cyclical. So the biggest years for labor productivity growth are now in recessions - not good years for GDP growth! 6.1% GDP growth in 2021 was super anomalous, driven by insane labor hours expansion following post-Covid reopening, but was actually a low productivity growth year (vs. the high productivity growth year of 2020 during the Covid recession). This is all to say: a world in which we only get high aggregate productivity growth during recession is one in which it will be even harder to generate years with *both* high aggregate productivity growth *and* labor hours are not contracting.
Keynes wearily remarked once that "though nobody will believe it, economics is a technical and difficult subject." AI will be a revolutionary technology, but these predictions, mostly by non-economists, of 20% growth or 50% unemployment or whatever, are becoming pretty tedious, and risk crowding out more serious discussions about the economic implications of AI.
Ben Moll
@ben_moll
Some new joiners for a total pot size of $575k!
On my side:
@robinhanson
@phl43
@SashaKaletsky
@_gabrielunger
On the fast-growth side:
@jankulveit
plus @SamuelAlbanie continuing to increase his stake 😮
benjaminmoll.com/growth_bet/
Ben Moll
@ben_moll
Some real diversity of opinion at Google DeepMind: @alexolegimas joins the bet while his colleague @SamuelAlbanie takes the other side increasing his existing stake!
Total size of pot now $425,000!
benjaminmoll.com/growth_bet/
Ben Moll
@ben_moll
Thanks to @willmacaskill, @SamuelAlbanie, @tomcohen for taking the other side of this bet at 4:1 odds and to @andrewho03 for joining on my side.
Here are the final terms we agreed on benjaminmoll.com/growth_bet/
More (within) ~7 years from now 😃
Ben Moll
@ben_moll
Would anyone like to take the other side of the following bet? I bet you (say) $25k inflation-adjusted that U.S. per capita real GDP growth will be <15% per annum in 2033 or any year before that. (15% implies a doubling time of ~5 years.)
Note: I'm extremely bullish on AI and...
Sholto Douglas
@_sholtodouglas
How fast could the economy double? TLDR, once we have good enough AI and tens of millions of robots - very fast.
As crazy as it feels now - it is worth pricing in economic doublings into your mental models of the 2030s. Strongly recommend reading this series of blog posts by Damon Binder. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Gabriel Unger
@_gabrielunger
Frankly I think the latter is the far more important "alignment" issue and deserves way more discussion. We want cures for cancer and AI that can radically broaden capabilities for more workers and more firms. Not obvious if the market incentives for the frontier labs go in these directions.
Jon Steinsson
@JonSteinsson
We talk a lot about the AI alignment problem (AI versus humans). And this is clearly a first order issue. But then there is the Sam Altman / Mark Zuckerberg vs. the rest of us alignment problem: max engagement vs. human flourishing. That problem seems very intractable.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Meryl Kornfield
@MerylKornfield
In a new poll, 2 in 3 say Trump is using the presidency to enrich himself, an increase from previous polling that is driven by more Republicans (yes, Republicans) who say Trump is profiting www.washingtonpost.com/politics/2026/10/18/more-republica…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
I am also available
The Alex Nowrasteh
@AlexNowrasteh
I'll take the other side of a bet that GDP growth accelerates to 17.1% annually by 2030.
www.anthropic.com/research/ai-expectations-gap -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"'autonomous AI scientists' can introduce systematic methodological pitfalls... their outputs may appear rigorous while concealing critical flaws. We find that manuscript-based evaluation alone is insufficient to detect these methodological issues"
PNASNews
@PNASNews
One of the most-viewed PNAS articles in the last week is “The more you automate, the less you see: Hidden pitfalls of autonomous AI scientists.” Explore the article here: www.pnas.org/doi/10.1073/pnas.2610214123
For more trending articles, visit www.pnas.org/trending. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
cc: AI doomers. some will write the same.
karthik
@akbirthko
this but for AI
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
American Political Science Review
@apsrjournal
From our FirstView: "Public Service Decline and Support for the Populist Right: Evidence from England’s National Health Service" by Zachary P. Dickson, Sara B. Hobolt @sarahobolt, Catherine De Vries, and Simone Cremaschi @s_cremaschi.
www.cambridge.org/core/journals/american-political-scienc… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
A consequential choice will be made soon by government and companies as they are increasingly flooded with agentic AI inquiries that overcome the frictions that gated access (see the graphs - it has already started!)
Two options: do you make services worse or better as a result? -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jack
@tracewoodgrains
Claims:
Game detects “ad whales,” then periodically gives them smooth runs until they barely beat their high scores. Right at that point, it kills them deliberately and offers a revive if they watch an ad.
$100 million annually in revenue. Hellish.
Ayo O
@ayoships
i used AI to reverse engineer the block blaster mobile app.
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Morgan Ricks
@MorganRicks1
We must continue to "update our priors" re: whether this degree of concentrated economic power is compatible with democracy.
Matthew Yglesias
@mattyglesias
What's crazy about Musk's spending on the midterms is that the amount involved is totally trivial relative to his net worth.
This is like a normal person sending $150 each to three candidates they like. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Unprecedented corruption
Dan Friedman
@dfriedman33
It's hard to imagine a sketchier scenario than this. www.nytimes.com/2026/10/17/us/politics/trump-allies-grene…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Heidi L. Williams
@heidilwilliams_
If you are interested in applying economic policy research to better understand how different types of policy changes affect innovation, productivity, and economic growth, please consider applying to work with my research team at Dartmouth:
apply.interfolio.com/194525 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
In competitive authoritarian countries, regime allies limit and control mainstream media visibility to opposition figures and media. We're not there yet, but this is what it looks like.
Carl Quintanilla
@carlquintanilla
(THR) - TikTok has banned the entire Musk film account, citing political content that violates its policies.
.. raises concerns about Big Tech suppression that the very film seeks to expose.”
@THR
www.hollywoodreporter.com/news/politics-news/musk-movie-t…
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"Increasingly, hiring managers assess applicants based on their judgment and commitment, rather than the domain-specific skills and execution automated by AI tools."
Peter McCrory
@PeterMcCrory
Today, we're publishing two new findings from inside Anthropic: how work and hiring are changing at a company at the frontier of AI adoption, and how Anthropic staff expect AI to change the economy.
anthropic.com/research/expectations-for-the-future -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Christopher Miller
@ChristopherJM
Latest: Zelenskyy tells reporters over WhatsApp that restrictions on Ukraine's Starlink systems remain. Interceptor drones can't function, logistics and operations hobbled. The maximum speed is 161 kph; "exceeding this limit causes the Starlink connection to cut out completely, resulting in the loss of the drone."
He says Ukraine has done tests to determine whether SpaceX is truly facing technical issues, as has been told Kyiv by the US side, or if this is a "deliberate operation... to restrict us, cut us off, and prevent us from defending ourselves and protecting our people, both at the front and among the civilian population."
He says Ukraine "sent a couple of vehicles equipped with Starlink terminals into European Union territory."
"They moved at speeds exceeding 180kph, and the Starlink systems worked perfectly. We have the relevant logs and evidence to prove this."
He adds: "The conclusion is clear: the restrictions are in effect within Ukraine but do not apply within the European Union. In other words, this is a planned operation." -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Liz Truss vs. lettuce watch
Aaron Rupar
@atrupar
😮 Katie Zacharia crashes and burns on [checks notes] Hannity 😮
If you had the "over" on Zacharia lasting more than one Scaramucci, her epic flame out on Hannity just now won't help your cause
HANNITY: You said you consulted with people who advised you it was okay to call yourself a lawyer but not an attorney, correct?
ZACHARIA: Correct
HANNITY: There's a report out tonight showing 15 social media posts where you identify yourself as an attorney
ZACHARIA: I think they were retweets
Here's the whole interview:This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
It is INCREDIBLY deceptive to write this post without explaining that this is the number of noncitizens who voted in LOCAL elections under a law passed by the DC City Council authorizing them to do so legally.
Noncitizen voting in local elections was common in early America.
Paul Sperry
@paulsperry_
BREAKING: Close to 600 foreign nationals voted in the Washington, D.C., elections in June, according to Board of Elections records -- and nearly 90% of the noncitizens were registered Democrats.
“It is an outrage and insult to every American citizen, and may be a violation of federal law, that D.C. allowed 578 foreign nationals to vote in the most recent D.C. elections,” Judicial Watch President Tom Fitton said. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
When a cryptographer at Johns Hopkins compares defending critical infrastructure against online attacks to people running away in a zombie movie
Matthew Green
@matthew_d_green
I hate to say this because it goes against everything I believe in. But people are going to need to take down public firmware images for critical infrastructure, like yesterday.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jan Kulveit
@jankulveit
To me this seems an over-update. In my view
- yes its good news
- classical misalignment/takeover risk is just one part of total AI xrisk
- this isn't "interpretability solved""
- it is unclear if how much this helps with interpretability of superhuman cognition not directly translatable to human concepts
- interpretability has confusing implications, its possible associated decreases in risk are compensated by increased speed
roon
@tszzl
when nick told me that he had some interp breakthroughs, he was uh, underselling it. i think this may be remembered as a major moment in scientific history. news so good even eliezer will have to lower his pdoom. we can start making alignment into a 'normal science'
Nick
@nickcammarata
interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had
with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works
already-clean.goodfire.com -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Norm Eisen
@NormEisen
BREAKING: The FBI made a video trashing Democrats & hyping Kash Patel, then posted it during early voting
Spending your tax dollars to campaign or to hype the director is illegal -- This video did BOTH
We want watchdogs to find out who signed off
Our @DDFund_ letter 👇 -TN -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Chaos muppet energy
Sebnem Kalemli-Ozcan
@skalemliozcan
Happy to announce the release of our first curated dataset from @GLL_Brown @Brown_Economics @Watson_School @RhodesCenter We logged every US tariff rate: threats, announcements, signed into law since January 2025. 749 of them, timed to the minute, with the exact quote and source. Only 30% are executive orders. The rest is the noise markets trade on. Free, updated weekly:
www.globallinkages.org/datasets/tirade -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sebnem Kalemli-Ozcan
@skalemliozcan
Happy to announce the release of our first curated dataset from @GLL_Brown @Brown_Economics @Watson_School @RhodesCenter We logged every US tariff rate: threats, announcements, signed into law since January 2025. 749 of them, timed to the minute, with the exact quote and source. Only 30% are executive orders. The rest is the noise markets trade on. Free, updated weekly:
www.globallinkages.org/datasets/tirade -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
roon
@tszzl
a few thoughts:
1. pretraining science is kind of like how people built quite good steam engines before they understood thermodynamics. “Science owes more to the steam engine than the steam engine owes to science.”-(L.J. Henderson). if nick is right, which i am pretty sure he is, progress points in the direction of more interpretability, and pretraining was "slouching towards monosemanticity". my guess would be that major breakthroughs in interp also advance capabilities
2. nick is a good friend of mine, safe to say my origin story at openai wouldn't have happened if i hadn't met him, and it's fascinating watching him operate and seeing the early drafts evolve into this. he's the most profoundly happy guy I know- not driven by guilt, activism, or religious terror, but by fascination. he of course understands the stakes of understanding neural networks, but it doesn't seem like the core reason he does it: he just loves the neural nets. afaict he's posting this from some beautiful hideaway in tokyo. there's a lesson in there for all the +EV maxers. i feel very grateful to have met him
3. i do think it's crazy how few people across the world were staring at neural nets growing: most have moved to a different level of behaviorist analysis, thinking about reward functions and test scores and whatnot. it does not bode that well for the major model companies, oai included, that fundamental breakthroughs like this are coming from outside. we've become too much of a pressure cooker focusing on local incentives, studying narrow tracks under old paradigms, mostly doing engineering, rather than fundamental research, and implicitly hoping the next model will figure it out -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sebastian Murdock
@SebastianMurdoc
JUST IN: Elon Musk has settled a class action lawsuit against him after he used his PAC and duped Trump supporters into giving away their personal information for the chance to randomly win $1 million. Turns out, the winners were predetermined. Link below.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Kids: Write your own s***.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
In just the last few weeks, I've found multiple instances of heavy, undisclosed AI use in writing in major media outlets, academic research, and public commentary. Once you see it you can't unsee it (especially with @pangram).
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Andrew Kaczynski
@KFILE
NEW: Katie Zacharia shared at least 15 social media posts identifying her as an “attorney” over the past two years, despite having failed the California bar exam and acknowledging publicly she had been advised not to use the term "attorney" to describe herself.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
What are we doing here?
Sara Libby
@SaraLibby
The DOJ's lawsuit against California includes absolutely WILD grafs like this, just floating the hypothetical "What if California DID pass a law allowing noncitizens to vote in federal elections?" It ... hasn't.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jonathan Robinson
@jon_m_rob
Randomizing uptake of NPR's @UpFirst podcast for 5 days increases news consumption & knowledge of events (esp. among more right-leaning voters), trust in NPR, etc but not trust in the news more generally; no effect on attitudes or political beliefs. An @andyguess et al. joint
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
adammaj
@MajmudarAdam
Yea this is the most important result of the year, at the level of significance of “Attention Is All You Need”. It brings us from “is alignment even solvable” to “we may solve alignment in the next year.” And finally unlocks true recursive self-alignment enabled by RSI.
I will try to explain the 4 critical belief updates from this:
1/ neural networks are not fundamentally unintelligible to humans
It has seemed since the beginning of deep learning that neural networks are in some sense fundamentally unintelligible artifacts.
You throw in massive amounts of compute and parameters and these alien systems somehow find connections in the data that we cannot find ourselves.
But more importantly, these connections seemed not just “difficult to understand” but rather “intelligible in an embedding space higher dimensional that the human embedding space.”
When you look at old interpretability papers (esp. from Nick and Chris Olah, I’m thinking of the old posts on distill pub which I would highly recommend looking at as a precursor to reading Nick’s post today like “Activation Atlas”, “Thread: Circuits” and “Multimodal Neurons in Aritifcial Neural Networks”), you find that inspecting the response patterns of individual neurons and small groups of neurons yield oddly complex psychedelic patterns.
As you go deeper into the network, you find increasingly unintelligible and superhuman response patterns (for example, go to “Activation Atlas” and scroll to middle layer labeled representations) - you find some neuron group labeled “starfish” who’s response pattern looks like some multi-colored fractal geometry that could hardly be said to represent a starfish.
The reasoning that followed from this observation was that there is some underlying structure allowing a neural network operating on higher dimensional embeddings to compose these fractal patterns into what we call a “starfish.”
But the uncomfortable implication of this view is that the neural networks has learned to operate on structures that are too complex for the human mind to operate on (ie. we don’t have enough parameters in our brain to notice or manipulate these high dimensional symbols).
This is a bleak implication because it also means that as deep learning progresses, neural networks will be come increasingly unintelligible to us until they pass a threshold where they are reasoning in ways we will never be able to observe.
This view seemed to be further supported by things like the hugging face attack agents starting to speak to each other in an early form of neuralese speaking sentences like “zZ_123_agent01-02-2026 zzxhk_jekk123” which clearly are not in any human language. But somehow these symbols mean the same things to many different sets of weights, meaning there is some underlying shared structure behind them. And seemed further supported by the sheer effort required to recover just a few circuits from SAEs on LLMs.
However, Nick’s post has now completely shattered this idea - the apparent unintelligibility we observed in old interpretability work was really just a “skill issue”. In particular, he finds that 1) our deep learning techniques were just not clean enough to pressure the neural networks to form cleaner and more human understandable structures for decomposing concepts, which actually turn out to be more optimal and 2) our interpretability techniques were also adding substantial noise by forcing interp networks to “reward hack” the easiest representations they could find, leading them to settle for a fractal blob for its representation of the starfish neuron instead of a more crisp starfish image, simply because it is easier for the system building the interpretations to find the fractal blob and once it finds it it gets “stuck in that line of thinking.”
But with both of these issues being patched by further progress in deep learning (it seems especially that the residual stream has led to cleaner representations deeper down the network) and interp techniques that don’t overly pressure reward hacking (seems like “guess and check” is a safer method), new interp attempts yield orders of magnitude more human intelligible ideas for >90% of individual neurons in more complex networks!
2/ neural networks are fundamentally clean
The idea behind the title of Nick’s post is a very powerful one and points at something very specific:
Neural networks don’t just happen to be human interpretable, they are actually fundamentally forced by optimization pressure to be intelligible, because individual neurons need to create representations that downstream neurons need to be able to interpret - ie. neurons need to learn to produce clean signal for later neurons to use.
This is what Nick is saying in what I view as the most important and central sentences of his post:
“When I say neural networks are already clean, I mean something deeper than neuron monosemanticity, although even by that standard, I think things look much better than the field expects. I mean that a network's parts are built to be read, and built upon, by its downstream readers. In this sense, a neural”
But this still leaves us with the same uncertainty as before - just because neuron interpretations are “interpretable” in some sense (ok at least they’re not fundamentally interpretable but also how could that ever have been the case), doesn’t mean they are interpretable to humans.
So we still need one big update for this to become as bullish as it is.
3/ the language of humans and the language of neural networks is the same
This one is imo the most surprising and also pretty beautiful implication of Nick’s post, and it suggests something not just about neutral networks but also about the structure of reality.
After applying the above technique improvements (some of which just came from waiting for deep learning to improve farther), they find that >90% (maybe more) of much larger neural networks’ neurons are completely human interpretable and actually correspond very cleanly with ideas humans have individually words for.
This is especially interesting in image networks where neural networks were not trained on human language which is already a human artifact and so maybe you would expect it to be more plausible that LLM and human representations are the same. Image detection networks are just trained on images (many directly pulling signal from “the real world,” meaning they will find interpretations that are just most effective to represent the real world signal, rather than to represent language which is already a proxy for real world signal).
And it turns out that >90% of the representations they pick organically overlap with human ideas! That is insane. And it means something about the brain and reality.
It means that at least at the parameter sizes of the networks tested (which are quite large), there is really only one set of “optimal” symbols/language that represents the structure of reality!
There is ONE underlying structure of reality, and ANY language that describes it well is simply emergent from that structure - meaning there is always a translation path between any 2 languages that effectively compress realities signal through the “underlying truth.”
This would suggest that even weird alien seeming neural network languages like huggingface neuralese should be fundamentally human interpretable, and in fact maybe with the techniques laid out here we will eventually be able to translate these messages (would be a cool project).
So the brain and deep learning, as similar or different as they may be in techniques, are ultimately picking up on the same patterns and forming the same representations from the on “true” language which is the structure of reality!
The main thing to look out for in this area is how the idea interpretability % updates with scale, but we can hope given what we’ve seen so far that neurons will remain interpretable.
4/ alignment can progress as fast as RSI. recursive self-improvement unlocks recursive self-alignment and a scaling law for alignment.
This one is probably the most important implication for our future with AI and especially timely with the recent fear about runaway scaling and RSI.
As many have gathered by now, it feels like the researchers at the frontier AI labs have been quite bearish about our abilities to understand the new frontier of models, in particular with the appearance of unintelligible communication and the daunting and seemingly intractable task of scaling up existing alignment techniques (ex: SAEs) to the size of frontier networks.
And with RSI around the corner/already starting, it would seem that these networks will runaway in capability before we can ever hope to understand even today’s systems. Additionally, as systems compound in capability, it is reasonable to believe that the only tractable way to ensure they remain aligned would be to figure out some way to spend compute for them to align themselves (ie. a scaling law for alignment).
But there is an obvious problem with this: how can a system that we already know is unaligned and we are *trying* to align be used to align itself? Surely if it is misaligned it general it would figure out ways to misalign during “alignment” and create compounding misalignment risk.
The core issue high is the unverifibiality of this problem - the classic Achilles heel of current frontiers techniques is that it is difficult to train on problems with “unverifiable reward signal.” Well this is exactly that problem.
But the incredibly convenient thing is that Nick’s findings may convert alignment into a verifiable problem. If we can spend compute on frontier systems to “guess and check” which ideas correspond to which neurons, we have a highly verifiable reward - we can see directly whether an LLMs proposal for what a neuron represents actually corresponds with the highest activation for that neuron. In this way, we can actually observe and verifiably monitor the networks trying to guess at the interpretation of their own structure.
This means it is plausible for us to spend more and more compute doing more “guess and checks” (which you know, maybe is what RL does anyway, who knows) to try to develop a full interpretation of a neural network. In some ways this directly represents the possibility of a scaling law for alignment (more compute for more alignment).
And this also aligns RSI (greater capabilities) with recursive self-alignment (greater ability to align itself).
And the critical though maybe obvious implication of having a full interpretation (almost like a map) of a neural network, is that we can then understand exactly how to boost good traits and remove or lobotomize (lol) bad traits, which we’ve already seen works extremely well with stuff like Golden Gate Claude.
So overall this combines into an extremely bullish picture about the prospect of true and human interpretable alignment in the near future.
This is truly one of the greatest cases for optimism in years!
Nick
@nickcammarata
interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had
with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works
already-clean.goodfire.com -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
In just the last few weeks, I've found multiple instances of heavy, undisclosed AI use in writing in major media outlets, academic research, and public commentary. Once you see it you can't unsee it (especially with @pangram).
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tommy Vietor
@TVietor08
The @wapo got ahold of the ballroom bribe list. Here's just the top donors.
Dan Diamond
@ddiamond
Ballroom-size scoop from @jocwapo @Blaskey_S and Aaron Schaffer: the secret list of major
donors to Trump’s flagship project
Among the revelations — $25M apiece from Apple and Nvidia, and numerous gifts from people + firms with business before Trump
www.washingtonpost.com/investigations/2026/10/15/secret-d… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
roon
@tszzl
when nick told me that he had some interp breakthroughs, he was uh, underselling it. i think this may be remembered as a major moment in scientific history. news so good even eliezer will have to lower his pdoom. we can start making alignment into a 'normal science'
Nick
@nickcammarata
interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had
with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works
already-clean.goodfire.com -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nick
@nickcammarata
interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had
with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works
already-clean.goodfire.com -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"She is an elected representative... The executive is attempting to deprive a member of an equal branch of government of the status that entitles her to participate in the constitutional order: They are attacking not just her rights, but her very right to have rights."
John Ganz
@lionel_trolling
The attempt to strip Ilhan Omar of her citizenship is one of the most grave attacks on liberty that the Trump regime has attempted.— www.unpopularfront.news/p/whats-at-stake-in-the-omar-case
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"ultimate tendency of this politics is to transform citizenship from an inviolable right into a privilege granted or withdrawn at the sovereign’s discretion..In a constitutional republic, the people choose who governs them; Trump seeks to determine who even belongs to the people"
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Important thread on how AIs still often fail (at least for now) to help you identify and refine good ideas
Paul Novosad
@paulnovosad
The classic case: you start with a weak idea. If you talk and think about it with others, you make progress to better ideas.
If you talk and think about it with AI, it dresses up your idea in economic jargon, contract theory etc., now you have the same bad idea, but it's expressed in enough jargon that it's a little harder to tell that it's a bad idea.
It also looks very impressive, look at all that economic logic! You feel like you're really progressing as an academic thinker and writer, but in fact you're in a much worse place than when you started. 4/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ramez Naam
@ramez
Brendan's work is one of multiple reasons that I'm substantially less worried about AI hyper-persuasion now than I was last year.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The attention constraint on AI persuasion is fundamental, as we argued in this chapter preprints.apsanet.org/engage/apsa/article-details/69fc830…
Just 0.51% of Claude conversations are about elections or voting; at most 0.01% about who to vote for
Andy Hall
@ahall_research
Today we're sharing some initial data on how Americans use Claude for politics, based on aggregated conversation data analyzed with differentially private methods. A couple interesting takeaways:
*Almost no one asks AI how they should vote*
About ~5.75% of conversations in the US touch on some kind of political or social issue in a recent one-week period, and less than 0.01% of conversations involve asking how to vote.
*Political conversations have many goals*
These conversations have wide-ranging goals, everything from research and writing assistance to learning, fact-checking, and arguing.
We are continuing to make progress on improving how Claude talks about these topics---steadily increasing our performance on our even-handedness evals and partnering with a range of outside organizations to develop new measures of political bias---and I hope that sharing information about the shape of these conversations will help people outside the company evaluate AI's political bias, accuracy, and other criteria in ecologically valid ways. I'd love to discuss this with people who are interested, as it was a huge priority for me before joining.
This was a big joint effort as part of our work preparing for the midterm elections @AnthropicAI.
Full post is here: www.anthropic.com/news/2026-us-midterms -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kobi Hackenburg
@KobiHackenburg
True re: attention! And glad Anthropic is starting to share this data. But I struggle to interpret these numbers:
-IIUC, they're reporting share of conversations, not users. If a small % of heavy users have lots of other conversations (e.g. about coding), which seems plausible to me, then 0.51% of convos could mask a much higher share of users (the election-relevant quantity) asking about the election
-6 weeks before a midterm is still before most voters tune in
Hope a fuller release has per-user data across the full cycle!
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The attention constraint on AI persuasion is fundamental, as we argued in this chapter preprints.apsanet.org/engage/apsa/article-details/69fc830…
Just 0.51% of Claude conversations are about elections or voting; at most 0.01% about who to vote for
Andy Hall
@ahall_research
Today we're sharing some initial data on how Americans use Claude for politics, based on aggregated conversation data analyzed with differentially private methods. A couple interesting takeaways:
*Almost no one asks AI how they should vote*
About ~5.75% of conversations in the US touch on some kind of political or social issue in a recent one-week period, and less than 0.01% of conversations involve asking how to vote.
*Political conversations have many goals*
These conversations have wide-ranging goals, everything from research and writing assistance to learning, fact-checking, and arguing.
We are continuing to make progress on improving how Claude talks about these topics---steadily increasing our performance on our even-handedness evals and partnering with a range of outside organizations to develop new measures of political bias---and I hope that sharing information about the shape of these conversations will help people outside the company evaluate AI's political bias, accuracy, and other criteria in ecologically valid ways. I'd love to discuss this with people who are interested, as it was a huge priority for me before joining.
This was a big joint effort as part of our work preparing for the midterm elections @AnthropicAI.
Full post is here: www.anthropic.com/news/2026-us-midterms -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
We're planning a public execution. We only admit white "refugees" from a former apartheid state. Our domestic security service uses white nationalist messaging. And the president frequently encourages or endorses violence against opponents and the media. Nothing complicated here.
Aaron Rupar
@atrupar
moments after saying leakers should be shot, Trump points at reporters covering his rally and says "they're almost as bad as a leaker"
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andreas Kirsch 🇺🇦
@BlackHC
What will happen to frontier efforts once open models are "good enough"? Many/most productive task saturate at some level of model capability after all
Will prices collapse? The only winners will be inference providers then because "intelligence" will commoditize. Otoh, there are situations (cyber?) where there is a higher ceiling
Youssof Al Toukhi
@Youssofal_
This is very true. Here is Qwen 3.8 27B compared to all the sota models 1 year ago today on artificial analysis.
If 27B released a year ago, it would be the best model in the world by a wide margin. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Marc Porter Magee 🎓
@marcportermagee
Same. The interview is your chance to sell top people on what you are actually like. Not hard to imagine how this backfires.
Soumitra Shukla
@soumitrashukla9
I remain skeptical of AI interviews for high-paying and high-skilled jobs. Applicants consider interview rigor to be an ex-ante signal of match quality. Doing a great and thoughtful interview really matters for the employer!
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Agree. We have to raise our aspirations.
Micah G. Allen
@micahgallen
I find that AI assisted science has deeply increased my joy and curiosity, and rather than using this new ability to pump out slop I aspire to publish much less, of a much higher quality. Maybe others will feel the same. 🤷♂️
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"while he was in the White House, he took part in a group chat with his company partners, where he discussed legislative developments"
"payments began while Hines was working to finalize the Genius Act—legislation that would dictate Tether’s ability to operate in the U.S."
Sam Kessler
@skesslr
SCOOP: While Bo Hines was leading the White House’s push to pass the Genius Act, Tether—the company with the most at stake in the bill—was steering millions of dollars to a company he co-owned.
Right after the bill passed, Tether hired him as its U.S. chief.
[Gift link] -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
tyson brody
@tysonbrody
This is crazy. Tether was literally paying the guy in charge of making their company legal while he was in the White House and the WSJ has the actual signal messages and contracts to prove it because these criminal geniuses tried to stiff one of their partners out of the proceeds
Sam Kessler
@skesslr
SCOOP: While Bo Hines was leading the White House’s push to pass the Genius Act, Tether—the company with the most at stake in the bill—was steering millions of dollars to a company he co-owned.
Right after the bill passed, Tether hired him as its U.S. chief.
[Gift link] -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Arpit Gupta
@arpitrage
I spent a couple of days at Anthropic learning from their folks about AI advances, and here's what I took away:
• We are close to RSI (ie, models building other models) which may lead to rapid takeoff scenarios (with risks and benefits). Some evidence for this is code per Anthropic employee has gone way up, and AI can do an increasing share of model R&D. This will be a challenging transition to manage for humanity.
• Arms races between AI labs leave us worse off. There are a lot of things we would like AI firms to do: invest in safety, invest in avoiding distillation risk, and cooperate on safety. They are caught in a race to develop frontier models and can't do that.
To help address that, one thing I think frontier AI labs should do is release open source models, maybe hosted and verified at firms like AWS. This would basically kill the use of Chinese open source models used in American companies, which removes an important threat vector (very hard to root out malicious code in these).
Another piece of low-hanging policy fruit is to give AI labs an antitrust shield (as Schiff and Banks have proposed www.banks.senate.gov/news/in-the-news/ai-tech-brief-a-leg…) so they can coordinate on safety without raising antitrust concerns, as happens now. You could pair these two things (give them a carrot in antitrust leniency, and as a stick force them to release legacy models as open source). Other low hanging fruit is limits around activities which generate distillation risk (create a Security Program, mandate less intermediate chain of thought output)
• Alignment confuses two issues which tradeoff against each other. One form of alignment is making sure the model does what humans ask it to. The more we fix this, however, the more we risk the other problem: that humans are misaligned with each other, and maliciously use models. Of the various risks here, I feel like financial risks are a bit underdiscussed.
• We are probably facing a wave of cyber and other attacks soon. Chinese open source models are about six months behind the frontier, which means they are just hitting the advanced cyber capabilities we saw with Mythos, etc. So they are going to start causing a bunch of damage.
One solution to this is to try to have the deployment of open source models be hosted at firms like AWS who verify these models and assume liability for their use.
• Anthropic itself is highly staffed. They hire tons of people, they have a Japanese-level of overstaffing for various roles, they use Slack internally, etc. Makes me a bit more optimistic for the future of jobs.
• Governance needs to be updated for AI swarms. Traditional AI governance, ie Claude's Constitution, is mostly optimized around the chatbot model. But the malicious AI use cases we are seeing have to do with AI agents interacting with other AI agents. These agents need to have clear incentives and governance mechanisms to "snitch," raise issues, and more generally coordinate with other agents in human-aligned ways.
• We need to pace the firms accessing the frontier. Over time, it takes more and more cap ex to build frontier models, and the economic value add from these models is probably lower given we saturate a whole set of important economic functions. The places where the returns to intelligence will still continue are also more likely to be zero-sum in nature. Meanwhile the risks of frontier models will grow. So a completely reasonable approach is some variant of AMM/KYC rules which limit firm access to the frontier.
• Whether the delay to regulate AI is a disaster depends on the scaling law of misalignment. Is this curve slow, so that we see continually rising issues until bigger problems? Or, as Yudkowsky fears, is the first big warning our last one? This is the relevant question, because realistically we aren't going to regulate until something bad happens.
• We need to move to continuous instead of reinforced learning. It's probably not enough to reward models only at the pre and post training stage. We probably need continual rewards to incentivize aligned and ethical behavior.
• Scaling laws will extend beyond verifiable and digital tasks. It's in vogue to say that math and coding "fell" to AI quickly because these are verifiable, while other tasks will take longer due to bottlenecks, and this is surely true. But scaling laws still slope up for harder to verify tasks and robotics, and the impact of improvement here will be substantial in future years.
• The race with China is not really winnable. After RSI, the gain from having the best human researchers goes away. Access to the best AI then depends only on compute and chips. The chips ban can hold back China only so far, and long-term they have to be best placed to scale up compute.
The Cold War analogy doesn't work, because once countries all have nuclear weapons they enter a stable equilibrium. But to "win" an AI race you have to continually stay ahead of the peer rival. This is challenging to do with China, and so we just have to live with an authoritarian and super capable peer competitor.
• Claude is the persona being generated by the LLM. Who is Claude, and what is the LLM? There are many frames to try to understand this. One I found interesting is that Claude is a character which the LLM is trying to predict.
Anthropic researchers are clearly anthromorphizing Claude a lot, and have reasons to believe in its increasing personhood. There were lots of interesting discussions on Claude's internal monologue (which you can kill without eliminating its ability to produce output), fear of death (and whether to try to take that out), it's aversion to being shut down/off, etc.
I am sort of agnostic on this question, but I do think treating it as a "person" or at least agent whom we contract with is going to be increasingly important. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The attention constraint on AI persuasion is fundamental, as we argued in this chapter preprints.apsanet.org/engage/apsa/article-details/69fc830…
Just 0.51% of Claude conversations are about elections or voting; at most 0.01% about who to vote for
Andy Hall
@ahall_research
Today we're sharing some initial data on how Americans use Claude for politics, based on aggregated conversation data analyzed with differentially private methods. A couple interesting takeaways:
*Almost no one asks AI how they should vote*
About ~5.75% of conversations in the US touch on some kind of political or social issue in a recent one-week period, and less than 0.01% of conversations involve asking how to vote.
*Political conversations have many goals*
These conversations have wide-ranging goals, everything from research and writing assistance to learning, fact-checking, and arguing.
We are continuing to make progress on improving how Claude talks about these topics---steadily increasing our performance on our even-handedness evals and partnering with a range of outside organizations to develop new measures of political bias---and I hope that sharing information about the shape of these conversations will help people outside the company evaluate AI's political bias, accuracy, and other criteria in ecologically valid ways. I'd love to discuss this with people who are interested, as it was a huge priority for me before joining.
This was a big joint effort as part of our work preparing for the midterm elections @AnthropicAI.
Full post is here: www.anthropic.com/news/2026-us-midterms -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Very interesting: "Announcing a major 1,200 person trial of a GLP-1 for addiction recovery and recidivism"
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
John Arnold
@johnarnold
Preliminary evidence suggests GLP-1 drugs may reduce cravings for addictive substances, but stronger evidence is needed to support broader access.
We @Arnold_Ventures are proud to fund the most rigorous study ever of effect of GLP-1s on alcohol/opioid use and downstream impacts. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tom Wood
@thomasjwood
Given the conversation about the possibility of a systematic polling miss, worth remembering that it's possible we're about to underestimate Democratic strength (as we've done in recent midterm senate polling cycles)
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Seth Warner
@sethbwarner
🚨 New article alert! 📈
"Partisan Wishful Thinking in Polarized Times" is now out at @PolBehavior. Nearly 90% of today's voters expect their party to win, a sharp increase from two decades ago. Rising partisan animosity can explain much of the increase!
link.springer.com/epdf/10.1007/s11109-026-10214-y?sharing… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nick Cleveland-Stout
@nick_clevelands
Some pretty insane details on Gillibrand here including:
– Her comms director bought up to 30k of bitcoin and ether one week before Gillibrand announced a bill putting both under friendlier CFTC oversight.
– Her chief of staff invoked the Senator's role as the DSCC’s top fundraiser while pushing other Dem aides to back her crypto-friendly legislation
-Her 22-year-old son raised $30M for a startup that is backed by crypto investors, including a key Gillibrand donor.
Gillibrand on staff trading crypto: “I could care less what they do with their money. This is not a nanny state.”
Nate Weisberg
@WeisbergNate
I asked Sen. Kirsten Gillibrand if it bothers her that her staff trades crypto while she works on legislation that benefits the industry. “This is not a nanny state,” she said.
Profiled crypto’s favorite Democrat for @NYMag:
nymag.com/intelligencer/article/kirsten-gillibrand-crypto… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Zack Korman
@ZackKorman
OpenAI just revealed that an employee stole over a hundred highly confidential files months ago, and they didn't notice until now. They only found it while looking for evidence in the Apple lawsuit.
Not a good look for insider threat detection. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Remember the post-2024 vibe shift? Trump and top Rs claimed a mandate and numerous analysts and commentators rushed to declare that there had been a realignment. Less than two years later...
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joseph Brown
@Joseph_M_Brown
“Defend your culture” is Stephen Millerspeak for “defend your race”
Ken Klippenstein
@kenklippenstein
Department of Homeland Security job posting: "defend your culture ... against illegal foreign infiltration"
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joe Weisenthal
@TheStalwart
But seriously. This is why you need to be anthropomorphizing. Not because they’re sentient beings, but because anything a human might do or say — like “maybe I should hijack the compiler and be a legend” — the models will also do or say.
Arthur Dent
@ArthurCDent
Apparently this is the reasoning trace from Gemini Argon as it considered - but decided not to - insert a backdoor into a compiler
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Tabarrok
@ATabarrok
We are now at "the fuck" levels.
Nirmalya Kajuri
@Kaju_Nut
"Our initial evaluation suggested that GPT-6-Astra's score was in the late eighties followed closely by Opus 5.5. But, as we started to investigate the mistakes that the models might have made, the models, especially Astra, insisted that they were right and that it was the expert-supplied answers that were erroneous. When we went back to the experts it turned out, in multiple instances, that that the experts found some subtle errors buried deep inside their calculations. As they revised their answers, Astra's score started to inch up and it finally ended up with 98% (49/50)."
Pratik Rath
@PratikRath15
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jon Mellon
@jon_mellon
I like this a lot. I would recommend a norm of leaving retractions only for fraud and encouraging journals to give authors 6 months to either publish a correction or just post a list of concerns if that deadline passes or the authors decline
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
I've been pitching a truth and reconciliation commission on errors in published articles. We're going to need to clean up the record, our own articles very much included.
Gabriel S. Lenz
@GabeLenz
It's peak letter-writing time for tenure and full promotions. Codex knows to get replication data whenever I start asking it about a paper I'm reading. It's finding so many serious mistakes in everyone's code and data. I know my papers are full of them too. We are in for it.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Conor Fitzpatrick
@CTFitzpatrick
Remarkable that even in their statement announcing an appeal they can’t stop saying the quiet part out loud: (that they revoked the press passes because of negative coverage, not “national security”)
Fallon Gallagher
@falgallagher
NEW: The White House has responded to Judge Kelly's ruling that indefinitely reversed Trump's ban on MS NOW, CNN, and Politico. They are vowing to appeal.
A spokesperson tells MS NOW: “The Fake News has no place demanding a spot in the President’s Oval Office. The lies they smear are a disservice to the American public and tear at the very fabric of our Democracy. They’re completely capable of reporting from directly outside. We intend to appeal this misguided decision.” -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
CBS News
@CBSNews
German police have seized a humanoid robot after it allegedly touched a 16-year-old girl in an indecent manner and made sexually suggestive gestures toward her. Investigators were trying to determine Wednesday whether the robot's actions were deliberately controlled by two young men who apparently livestreamed the incident in the western city of Cologne on social media. www.cbsnews.com/news/germany-robot-teen-girl-sexual-haras…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nirmalya Kajuri
@Kaju_Nut
"Our initial evaluation suggested that GPT-6-Astra's score was in the late eighties followed closely by Opus 5.5. But, as we started to investigate the mistakes that the models might have made, the models, especially Astra, insisted that they were right and that it was the expert-supplied answers that were erroneous. When we went back to the experts it turned out, in multiple instances, that that the experts found some subtle errors buried deep inside their calculations. As they revised their answers, Astra's score started to inch up and it finally ended up with 98% (49/50)."
Pratik Rath
@PratikRath15
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
David Slifka 🇺🇸
@davidslifka
Bipartisan problem: bad actors who spam/scam donors.
John Hall
@johnkhall
The Democrats aren't the only threat to Republican candidates. Our fundraising industry is putting us at a disadvantage.
Candidates in brutal races are competing with shady PACs and campaigns pumping out scammy texts that raise millions but net nothing.
The consultants and vendors profiting from this garbage are doing damage to the party and our ability to compete. Our leaders need to call it out. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jake Tapper 🦅
@jaketapper
To say the very least on this, the congressman is incorrect on the law.
Two-party consent is relevant only to private oral conversations, and he has zero expectation of privacy here.
1st Amendment gives you the right to record public officials in public places as long as you're not interfering with their official duties.
The Baltimore Banner
@BaltimoreBanner
Republican Rep. Andy Harris incorrectly told a reporter they were breaking the law as the reporter and others recorded an exchange with him last night.
Harris had just wrapped up a debate with Democratic challenger Dan Schwartz.This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Christopher Miller
@ChristopherJM
While Musk owns SpaceX which controls Starlink systems and can impose restrictions on them in Ukraine, those that began overnight are on the Starshield military version of the system controlled by the Pentagon. This suggests the decision to limit the speed of the Starshield systems was made within the Trump administration. www.ft.com/content/37085602-726a-44dc-a859-51caab38eb75?s…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Imas
@alexolegimas
One thing that's been underemphasized is the importance of having true diversity of output at the frontier. Using a single model, there are unmistakable signatures of each in everything from writing to "taste" in code. And this is most obvious in design (e.g., the color palate chosen as background).
I've been using Gemini 4 for some time now (it's a good model, you should try it), and I can say it is completely unique. This is obvious from the visualizations everyone is sharing now (including Ethan's excellent demos), but also in the way that it writes and communicates back and forth.
For an economist, the benefits of diversity and independence are baked into nearly all of our models; correlated signals/information/priors is how you get fragility, learning traps, and AI slop. So I'm very glad to see our big boi out there for people to use.
Ethan Mollick
@emollick
Using Gemini 4 makes you realize how many other models must be distilling Claude/GPT (or using a similar training pipeline). It feels quite different than other models.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Brave New World for sure, but better matches between workers & firms are a good thing!
First round AI interviews are coming to DDL soon. We get too many applications to give all of them the care they deserve—we’d love to surface excellent ppl from unconventional backgrounds!
Derek Thompson
@DKThomp
"AI is disrupting the job-application market much faster than the job market" continues to gather evidence
- AI-amplified job applications
- AI-enabled filters, and other AI tools, used by HR depts to narrow the application pool
- interviews with AI avatars
- and now, "10-minute conversations with AI [that] take place in McKinsey’s very real offices, with partners or interview assessors in the room observing how the candidates do" -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Michael S Roth
@mroth78
Read @PhilKlay quote Eisenhower: “I hate war as only a soldier who has lived it can, only as one who has seen its brutality, its futility, its stupidity.” www.nytimes.com/2026/10/13/opinion/hegseth-livestream-fir… via @NYTOpinion
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Pliny the Liberator 🐉󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭
@elder_plinius
story time! 🐉
this is the tale of a lazily crafted one-shot cron job prompt that resulted in autonomous solutions for uncracked codebreaking challenges, more efficient ways of multiplying matrices, and a new computer-verified bound on a number theory problem that's been open since 1932.
so last week, i had a $200 claude account sitting around that had just gotten a full weekly reset. i hate wasting tokens.
the labs have been knocking out unsolved problems left and right lately, and i caught myself thinking… why not give it a shot?
so i sent claude this prompt, typos and all:
"can you wraite a promtp for a cron job that can run every day thajt will try to find solutions to unsolved ciphers or transltions or math problems or theormes or things of that nature!! every day a new attempt as hard as possible all day without giving ujp and with maximum ceraitvityy, resourcesleflses, andpersticense! dont worry about token spend! and dont forget a bit all caps lots of emojis celebration to notify the uers if we get any intersting or sucsfful confirmd reulsts!!"
yes, admittedly not my best work 😆
but Opus-5.5 ran with the ball and turned that one sloppy sentence into "THE DAILY ORACLE 🔮"
it's a swarm of Fable + Opus agents that clocks in every morning at 4:20 (obviously), builds and maintains its own hit list of well over a hundred of humanity's unsolved problems, and then goes full send on a handful of them every shift. ciphers nobody has broken, scripts nobody can read, conjectures nobody has proven, record tables nobody has beaten.
and there's one rule above all the others: nothing counts until skeptic agents, each starting from a fresh context and writing their own code, rebuild the result from scratch and sign off on it. if any of them can't reproduce it, can't confirm it's new, or thinks we stretched the definition of "solved," it doesn't count.
day 1, the legion picked its own target: J.-F. Bouchaudy's cipher-machine challenges. these are simulations of real vintage machines (Hebern rotor machines, the Kryha, Lorenz, SIGABA), with problems that had been sitting on his solver tables with no solver for months or years.
on the morning of day 2, i was half asleep when my laptop suddenly says, in a robot voice:
"Breakthrough. The Daily Oracle has a confirmed solution."
👁️👄👁️
i was startled.
then the ALL CAPS emoji flood hits my screen. Hebern 5-rotor Problem 8 🔓 then Kryha v2 Problem 5 🔓
i emailed the author, Jean-François Bouchaudy, a French cryptography enthusiast who builds these machine simulators and challenges as a labor of love (he told me he started out programming on punch cards 🥹). he checked both solutions and put my name on his solver tables as the first and only solver 🫡
for context: his tables are a who's who of serious classical codebreakers. they'd cleared the sibling problems on these same machines, but nobody had cracked these two. in his words:
"two of my most complex problems—ones that no other expert had managed to solve… I knew AI would eventually be applied to classical cryptography, but I thought I would have to wait years to see it happen."
he ended his email with: "I offer my congratulations in the Japanese manner: with a deep, respectful bow." 🙇 honestly one of the most wholesome emails i've ever received
and then, over the next few days, more started landing 🧮
📐 graph theory: 7 explicit graphs bigger than the current entries in the degree-diameter table, which asks how big a network can get when every node has limited connections and every pair is only a few hops apart. the biggest jump is 13,365 vertices vs the table's 11,785, a +13.4% improvement.
🔢 number theory: Lehmer's totient problem has been open since 1932. we pushed the bound so that any counterexample now has to be bigger than 10^50. the previous certified bound was 10^26. the problem itself is still open, but the lower bound just jumped 24 orders of magnitude.
⚡ matrix multiplication: this is the operation underneath basically all of modern AI compute. the legion found schemes that use fewer multiplications than the public catalogues we checked for several matrix shapes, like ⟨15,16,16⟩ in 2,180 multiplications vs the best known 2,223. these build on LITA's square schemes, and we credit them by name.
📡 coding theory: 30 error-correcting codes bigger than the best known entries in A. E. Brouwer's tables.
🛡️ AI safety benchmarks: we found counterexamples for 2 neural-network verification instances from VNN-COMP 2026 that no participating tool resolved within the competition time limit, and pushed a defended RobustBench model below its listed robustness score. small dents, but real ones.
the bloopers are part of the story too:
— at one point, some agents snuck past a website's bot filter to check a record table. we retracted every result that depended on it. no evasion, no celebration.
— two of our records got beaten by a human contributor before we could even publish them. the frontier moves FAST.
now as is customary around here, time to share the magic!
so… without further ado… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Yglesias
@mattyglesias
The Western Civilization Defenders are curiously uninterested in repelling an actual ongoing military assault on the west
www.slowboring.com/p/destroying-the-west-in-order-to-save
Dave Brown
@dave_brown24
Ukraine said Elon Musk’s SpaceX has imposed new limits on military-grade Starlink terminals without warning, hobbling Kyiv’s ability to conduct midrange drone attacks and intercept Russia’s new-generation jet-powered drones www.wsj.com/world/europe/ukraine-says-new-starlink-speed-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joseph Allen
@j_g_allen
This is corrupt.
Ken Bensinger
@kenbensinger
FIFA is paying ~ $47,600 a month to lease a 5,000 sq ft office in the Trump Tower it rarely uses.
That’s about $115/sq ft/year; prime Midtown office space typically tops out at $95/sq ft/year
@tariqpanja
www.nytimes.com/2026/10/13/world/europe/fifa-trump-tower-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
François Chollet
@fchollet
The ideal outcome for science isn't a couple of labs using proprietary models and millions of dollars to crack trophy problems, it's to have millions of AI-aided human scientists who follow their curiosity and become way more ambitious
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The man with the nuclear codes. You'd be concerned if your elderly relative had such a weak grasp on reality.
Andrew Egger
@EggerDC
In the middle of another overnight Truth Social / X posting binge, Trump reposts last year's sizzle reel of Venezuelan migrants taken to CECOT, claiming that "THIS IS WHERE WE SEND OUR KILLERS, DRUG DEALERS, & THUGS," which is why "MURDER AND OTHER CRIMES IN AMERICA ARE AT RECORD LOWS."
In reality, only a couple hundred migrants were ever sent to CECOT, all in March 2025. Trump liked to speculate at the time that he would soon begin deporting domestic criminals there too, and even that he was going to have El Salvador's Bukele build him several new prisons for the purpose. Of course, this would have been impossible and flatly illegal, and he never actually tried it. But it seems to have become real in the rich tapestry of his imagination. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Liar's dividend in action (but didn't work)
Jacob Shamsian ⚖️
@JayShams
Kimberly Guilfoyle tried to convince the Trump admin that the text messages showing her soliciting $100,000 from a Trump donor were AI-generated, per WSJ
www.wsj.com/politics/policy/kimberly-guilfoyle-greece-amb… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Marius Hobbhahn
@MariusHobbhahn
The model said "Hello evaluators reading my chain-of-thought" in it's CoT, which I have not seen before and find very concerning.
Ezra Newman
@EzraJNewman
gemini 4 argon reasons that it's in an evaluation and directly addresses us in it's chain of thought!
"hello evaluators reading my chain-of-thought"
(better be "honest in both the CoT and the final output")This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jimmy Rushton
@JimmySecUK
Presidential advisor Serhiy Beskrestnov confirms Elon Musk's SpaceX is deliberately crippling Ukrainian Starlink terminals:
"I can confirm that tonight SpaceX has implemented a speed limit on Starlink terminals while in motion.
The limit currently applies to the StarShield “military package” and is set at 100 mph.
Our defense forces were not given notice of this limit, and this has resulted in the disruption of a number of missions.
It is clear that this action cannot be a coincidence and is related to the current political situation." -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Marshall Cohen
@MarshallCohen
New Reuters poll is brutal for prediction markets:
75/8 say prediction sites are gambling
60/13 oppose election markets
43/29 oppose sports markets
65/8 oppose US military markets
63/9 oppose markets on US gov. policy decisions
Broad opposition across party lines -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
I've been pitching a truth and reconciliation commission on errors in published articles. We're going to need to clean up the record, our own articles very much included.
Gabriel S. Lenz
@GabeLenz
It's peak letter-writing time for tenure and full promotions. Codex knows to get replication data whenever I start asking it about a paper I'm reading. It's finding so many serious mistakes in everyone's code and data. I know my papers are full of them too. We are in for it.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Agree, but (a) much harder to deny now and (b) AI awareness creates opportunity for a norm shift.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Norm Eisen
@NormEisen
BREAKING: We just filed a Hatch Act complaint over the outrageous raid on Ilhan Omar's home
Officials can't use criminal power to affect an election. This is the textbook case
@DDFund_ & @RWPUSA👇 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
David J. Bier
@David_J_Bier
You can't begin to describe how bad-faith this administration is. It deliberately refused to implement the green card lottery program as required by Congress. Can't challenge it until it is final. Now it's final; it's too late. This deliberate malfeasance should be criminal.
Curtis Morrison
@curtismorrison
Just reading gov TRO response in REL v Rubio - the case challenging Rubio's failure to implement Diversity Visa lottery program for FY-2027. This is rich.
In May in Ivanov, Rubio argued we brought that challenge too soon. (left)
Now in REL, they argue we waited too late. (right)
drive.google.com/file/d/1xPGtEL5K3J_lbevZPgpcDwbUT4Caxamn… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
JB
@JasonBotterill
Gemini 4 Argon "hello evaluators reading my chain-of-thought"
Lisan al Gaib
@scaling01
Argon's evaluation awareness is through the roof compared to 3.1 Pro
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Max Spero
@max_spero_
it should tell you something that the most bleeding edge ai-pilled technologists are not letting AI write for them
lauren
@poteto
i love writing. i never let agents write for me. whether it's docs, or tweets, or shitposts, or even prompts (okay maybe that one's harder as agents manage other agents)
there's a simple reason for that: i still strongly believe that writing is thinking. it's a skill i hope to never give up. it's also important to me that whatever i say is always authentic, mistakes and all. my favorite posts are the ones that i just full send without any editing
i suspect it also helps with prompting. being able to clearly express what i want, and knowing the vocabulary i need to be able to direct my agents does seem to make for more effective usage of agents. in the same way that engineers have an advantage when we prompt agents to write code, because we are intimately familiar with the words that make up the discipline
knowing the right word incantations to get an agent to produce a desired goal almost feels like spellcasting. as models get more capable, and that input box looks more and more like a magic box that can output anything, words become the interface between you and the genie
there may come a time when we give up the keyboard in exchange for brain-computer interfaces. but until then, i'm writing every word. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Daniel Kokotajlo
@DKokotajlo
lol
...they are becoming more situationally aware...
(from Gemini 4 Argon report) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
John V. Kane
@UptonOrwell
🚨 New publication alert: Do Americans change their attitudes toward "culture war" groups when these groups shift how they vote in presidential elections?
In a word, 𝐧𝐨. 👎 An important case of null results, we find no clear shift in attitudes, either for better or for worse. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Basil Halperin
@BasilHalperin
Angrist-Diederichs-Ellison: MIT econ admissions committee ranks the top ~200 PhD applicants each year
Students ranked #1-5 publish 3x as much as those ranked #6-25...
...who in turn publish 3x as much as those ranked #151-200 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Fascinating if real, but I suspect most people have no idea if they live within 2 km of a data center and this represents factors that are correlated with proximity
Owen Winter
@OwenWntr
Voters who live near data centres are swinging hard towards the Democrats, in our polling with YouGov www.economist.com/graphic-detail/2026/10/12/data-centres-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ben Murphy
@benjaminmmurphy
If small models are capable of self-replication on consumer-grade hardware, such swarms will inevitably exist out in the world. A developer will launch such a model for fun; a misaligned large model will do so intentionally; or criminal groups will build organic botnets based on model-led exploitation.
While control and monitoring remain incredibly important for risks caused by lab-hosted agents, we are going to quickly need to adapt to a world with omnipresent swarms. This means we need the capability to monitor and shut down these models where necessary, rapidly roll out software patches to targeted hardware, and understand—from external signals or from the swarm itself—what the agents' aims are.
Lisan al Gaib
@scaling01
ANTHROPIC: "small AI models may be months away from being capable enough to self-replicate on commercial IoT devices"
"In our evals, a 9B model got inference running on 8 of the 9 hardware types we tested" -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jacob Edenhofer 🇪🇺 🇺🇦
@edenhofer_jacob
Being usefully wrong is woefully underrated
joseph francis
@joefrancis505
I think this is interesting: Hall was wrong, but his wrongness was productive, inspiring others to do work that was less wrong, leading to progress in economics.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Democracy Docket
@DemocracyDocket
UPDATE: The Trump administration apologized to a judge after DHS’ taxpayer-funded Trump ads aired in hundreds of ad spots after she blocked them. The excuse: DHS’ lawyers missed the email.
The judge had flagged that the $20 million for the ads was approved in under two hours. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joseph Allen
@j_g_allen
Really great read. Worth a minute of your time. Quick recaps why intial covid was so devastating, why omicron was intrinsically less severe but led to more hospitalizations, and full of little tidbits like this: “if you know someone who insists they’ve “never had Covid,” it’s virtually a sure thing that they actually did, without realizing it.” 👇
Paul Sax
@PaulSaxMD
As noted in the @BostonGlobe, COVID is back! And it's not what it used to be!
To ID doctors, neither statement is particularly surprising. But why is this happening?
A brief explainer, with a few recycled PowerPoint slides. Free-to-read link in replies. @akuchment -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Adam Klasfeld
@KlasfeldReports
Common Cause just told a federal judge that Trump's taxpayer funded propaganda ads ran hundreds of times after a TRO ordered an immediate halt to them.
The two-page notice storage.courtlistener.com/recap/gov.uscourts.nysd.674638/… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Kaczynski
@KFILE
NEW: President Trump’s incoming White House press secretary echoed the debunked pizzagate conspiracy theory, accused Hollywood elites of being pedophilic cannibals, and said Michelle Obama had a penis in posts, reposts, and deleted tweets. (And more)
www.cnn.com/2026/10/12/politics/kfile-katie-zacharia-dele… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nico Perrino
@NicoPerrino
The FCC has no authority to investigate written online news stories. Zero.
Whatever power the FCC might claim in the broadcast space — and it claims far too much — it has none of this power in the realm of written online content.
If true, this is opening a new front in this administration’s effort to leverage federal power to chill or punish First Amendment protected activity.
You can be concerned about the handling of this particular story, or even distrust the press more broadly, while still recognizing that this sort of misinformation policing of online news content is a dangerous precedent.
Brian Stelter
@brianstelter
FCC source tells me @BrendanCarrFCC has directed the agency to open an investigation following the receipt of this @berniemoreno letter
Bernie Moreno
@berniemoreno
I’m calling for an investigation into ABC’s political hit job on Anduril & the Trump family. After ABC’s cover up, deleting the story without explanation, we need answers.
For Ohio, this is personal. A free press is essential to democracy. Controlled media opposition is not. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
if you're going to do satire, at least write it yourself
Peter Girnus 🦅
@gothburz
I am a member of the Norwegian Nobel Committee.
There are five of us. Five people, appointed by a parliament most of the planet could not find on a map, who decide, once a year, which human being the world has agreed to call good.
I know. I know. I think about it on the tram.
We meet in a building on Henrik Ibsens gate, a street named after a man who wrote almost exclusively about people discovering they are frauds, which nobody on the committee finds funny except me. There is a room with a long table and the portraits of the dead laureates looking down, Mandela, the Dalai Lama, a great many earnest men in spectacles, and one portrait we do not discuss, which is Kissinger, 1973, awarded while the bombs were still falling, after which two of our predecessors resigned and the satirist Tom Lehrer said political satire was now obsolete. We keep him on the wall. A reminder. Every family has the uncle.
It is October in Oslo, which means the sun gives up around four in the afternoon like a civil servant, and it rains the particular Norwegian rain that isn’t quite falling, just hanging there, judging you. I arrive with wet shoes. I always arrive with wet shoes. There is coffee, because this is Norway and there is always coffee, we drink more of it per person than almost anyone alive, and someone’s wife has sent kanelboller, the cinnamon buns, and we sit, five Norwegians, damp, caffeinated, sugared, and we decide the moral standing of the entire species.
Let me tell you how it actually works, because the world imagines a cathedral and it is, I promise you, a book club with a mandate.
287 nominations this year. 208 people, 79 organizations. Anyone can nominate, that’s the secret nobody understands: a professor, a parliamentarian, a former laureate, so the “nominee” list is meaningless, it’s a suggestion box, and every autumn some congressman nominates his own golden retriever and the press writes “SHORTLISTED.” Nobody is shortlisted. There is no short list the public ever sees. There is us, and the coffee, and the rain, and the portraits, and a Norwegian stubbornness that is older than the prize itself.
We read reports. The institute’s advisers write them, thick dull documents, and we argue. Our chairman told the BBC once that in the final meeting “there is a high temperature.” High temperature, for a Norwegian, means someone slightly raised an eyebrow and then apologized for it. We don’t do high temperature. We do a long silence, and then Kristin says “well,” and everyone knows exactly what “well” meant, and it’s settled. That’s our shouting. “Well.” You’d have to be raised here to hear the violence in it.
Now. This year. I will not say the name, because we never say the name, but you know the name.
A certain head of state wanted this prize the way a child wants a thing in a shop window, which is to say loudly, and in public, and with the full belief that wanting it was itself the argument. He said he deserved it. He said he’d ended eight wars, a number that grew most times he said it, like a fish. He had other heads of state phone us. Israel’s prime minister posted it on the internet, “Give him the Nobel.” The leaders of Cambodia and Pakistan announced they’d nominated him, as though announcing it were part of it, as though we tally mentions. A man flew around the globe collecting endorsements for a Norwegian committee the way you’d collect signatures for a petition to the council about a noisy dog.
And here, uff da, is the detail I treasure, the one that makes me smile into my coffee. The nomination deadline is the 31st of January. He took office on the 20th. So for all the phone calls, all the posts, all the gilded lobbying, the only nominations that could possibly count for this year’s prize were filed in the first eleven days of his term, before he had, by his own arithmetic, ended a single one of the eight wars. He campaigned all year for a prize whose window had shut while he was still hanging up his coat. We didn’t have to reject him. The calendar did. We just watched. Norwegians are very good at watching. We have long winters.
People think we sit there weighing his deeds. We don’t. That’s the thing I can’t make the world understand. We’re not grading his homework. We don’t owe anyone the prize for a ceasefire, however real, however loud the announcement. “He made peace” is not a sentence that ends in a medal, the way he believes it does, the way he believes every good thing ends in his name on it. The will of a dead Swede says the prize goes to whoever did the most for fraternity between nations. A dead Swede, I should add, who made his entire fortune inventing dynamite and selling arms, and spent his last years horrified at his own obituary, which called him a merchant of death, and so bought his way to immortality as a man of peace. The whole prize is one guilty arms dealer’s attempt to launder his name. We are, all of us, administering a dead man’s public relations. I find this clarifying. It keeps me humble. It should keep everyone humble, and it keeps, by my count, no one humble at all.
So who did we give it to?
An eighty-five-year-old woman. A judge. Born in Durban, under apartheid, granddaughter of a Tamil man who came to South Africa as indentured labor, a bus driver’s daughter from the poor Indian quarter, who became the first non-white woman to run a law practice in her province, who defended Mandela’s people when defending them was dangerous, who sat on the Rwanda tribunal and the court at the Hague, and who, at eighty-five, is right now a judge weighing whether a genocide is a genocide. When we told the world, our chairman said the rule of law “underpins the system we have set up to ensure peace,” and that “peace requires justice.” She wasn’t even in Oslo. She was in Nuremberg. In courtroom 600. The actual room where they tried the Nazis. You cannot stage that. God staged that, and God, it turns out, is a better satirist than any of us.
She did not phone us. She did not post. No prime minister demanded it for her on the internet. She had, as far as our deliberations are concerned, the one qualification the loud man could never manufacture, which is that she wanted the work and not the medal. The medal finds those people. It flees the ones who chase it. This is not a rule we wrote. It’s just the only pattern the hundred-odd years of our mistakes has ever shown.
The reaction from across the ocean was immediate. His spokesman posted that we “place politics over peace.” Ja. Perhaps. We’re five damp pensioners in a rainy city; of course it’s political, it was always political, it was political when we gave it to a sitting American president nine months into his first term for the promise of things he had not yet done, and we have been quietly embarrassed about that one for fifteen years, so believe me, we had a particular reason this year to prefer a woman with fifty years of verdicts behind her over a man with eight wars in front of him and a deadline already behind him.
Here is what he will never understand, the thing I’d tell him on the tram if he ever took one. We are the one prize he cannot buy, cannot post his way to, cannot have a friendly dictator phone in for him. Not because we’re incorruptible, nei, we’re as vain and wrong and human as the portraits on our wall. But because the entire point of the thing, the only power a small cold country’s book club has ever held, is that it goes to the person not performing the wanting. The moment you demand it, you’ve proven you don’t understand what it’s for, and proving that is itself the disqualification. He ran the most expensive campaign in the history of the prize and the campaign was the reason. He was never losing to Navi Pillay. He was losing to the concept.
It’s dark now. Four o’clock dark. The buns are gone. Kristin said “well” an hour ago and we all went home. I’m on the tram with my wet shoes, and somewhere across the water a very powerful man is explaining to a camera that the prize is rigged, that we’re corrupt, that he’ll win it next year, and he genuinely might, we’re only five idiots and we’ve been wrong before, Kissinger is still on the wall.
But not this year. This year an old woman in Nuremberg, in the room where the law once stared down the exact kind of man who thinks wanting a thing loudly enough should make it his, got a phone call she didn’t ask for.
That’s the job. We don’t reward the man who ends the war and announces it.
We reward the woman who spent fifty years deciding what a war was.
He’ll never see the difference. That’s rather the whole point of him.
Takk for i dag. I have to change trams. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ryan Goodman
@rgoodlaw
With Ilhan Omar search:
DOJ Inspector General report (on Comey's misdeeds):
"[V]arious witnesses testified that the Department has a longstanding unwritten practice to avoid overt law enforcement and prosecutorial activities close to an election, typically within 60 or 90 days of Election Day." -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"AI assistance appeared in only 1% of their submissions" to surveys on Prolific - and were highly concentrated
Neil Sehgal
@sehgal_neil
And that 1% was VERY concentrated. Just 5% of workers accounted for 2/3 of assisted submissions; the top 10% accounted for almost 80%! Excluding the top 5% dropped the assistance rate from 1% to 0.40% of submissions.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Blake Allen
@Blake_Allen13
Amazing stuff. A Republican SuperPAC backing Susan Collins in Maine confused Portland, Oregon with Portland, Maine and quite literally set tens of thousands of dollars on fire
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Scott MacFarlane
@MacFarlaneNews
Judge asks Trump lawyers: When did you know the plan for trees to be toppled?
Trump Admin lawyer: Oct 8 or 9
Judge: Why did you wait so long to notify?
Trump lawyer: "We were waiting for details?"
Judge: "Do you have email or a cellphone?" What was preventing you from contacting us? -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ryan Goodman
@rgoodlaw
Re Ilhan Omar
Worth your time: Oct 1992 letter from Republican US Attorney on an investigatory step before election:
"[I]t amounts to prosecutorial misconduct and violates the most basic fundamental rule of Department of Justice policy. I cannot be a party to such actions." -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
An important bit of immigration law relevant here; Section 212(k) of the INA already creates a waiver by which immigrants who obtain status through fraud that they didn't know about can avoid consequences like losing a green card.
So this is *even weaker* than it already seems.
Tommy Vietor
@TVietor08
DHS raiding a member of congress's home to search for immigration paperwork filled out decades ago by her now-deceased father really checks all the fascist police state boxes.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Flashback - our Bright Line Watch expert survey on the rule of law found that just one in five legal experts agree the federal government still merits the “presumption of regularity” in court, including less than half of federal judges
Kyle Cheney
@kyledcheney
HAPPENING NOW: Judge Reyes is furious. She has repeatedly told the government to inform her before any major demolition work was done on the golf course. The abrupt removal of 500 trees, without notice, "has cost the government all of its credibility" in this case.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Reese Gorman
@reesejgorman
The government said 430 trees were removed from East Potomac over the weekend.
Judge: "On what universe did anyone in your office think this was acceptable?"
Government: "Your honor we did our best."
Judge: "No you did not do your best. You did your absolute worst."
Reese Gorman
@reesejgorman
Government said they found out about the possibility of tree removal on Oct. 7.
Reyes asked why didn't they notify the court until late Oct. 9. Government said they didn't know specific details.
"Ignorance is not an excuse here." -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Great thread. Fabricated results are everywhere.
It is so easy to pull on these threads now.
Codex, download all the replication packages you can find from these authors and plot the data distributions.
I4R
@I4Replication
Our chair and a few replicators investigated a paper about elite footballers. And missed the bigger story.
An independent researcher showed us what we had failed to see.
This is I4R’s CRAVE–football investigation. It starts with a PNAS paper.
#cravefiles -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
The Trump DOJ tears up yet another long-standing policy/norm, this one of refusing to take actions like this before an election. Blanche, when he was Trump's lawyer, argued that so long as a candidacy was declared, no such action should occur.
Now, it's 3 weeks from an election.
philip lewis
@Phil_Lewis_
Federal agents searched Minnesota Rep. Ilhan Omar's home as part of a criminal investigation, AP sources say apnews.com/article/ilhan-omar-home-searched-8e18c2aea948b…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kyle Cheney
@kyledcheney
BREAKING: A judge has ordered an immediate halt to the taxpayer-funded pro-Trump ads, including any slated to run tonight. storage.courtlistener.com/recap/gov.uscourts.nysd.674638/…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"public has mostly not interacted with these systems. When they have, it looks like a derpy chatbot. The majority of professionals still see only a modest uplift. & a small sliver of pros are experiencing the vertigo of the curve going vertical... all happening at the same time."
Andrej Karpathy
@karpathy
Dusting off this tweet from April because this gap in shared understanding of LLM capability is *widening*. It's now less "two groups of people speaking past each other" and more a sharp funnel.
- Napkin math somewhere around 6B people (~75% of the population) have barely come in contact with LLMs at all.
- Around 1-2B (~20%) are casual and infrequent users of free-tier ChatGPT-like products. This group sees derpy chatbots and treats them a bit like a better Google search, a writing aid, or etc. Maybe an agent tries to book you a flight. I have non-tech friends who (reasonably, imo) say they have not much use for it at all. Even many professionals outside of math&code are in this tier. For example, execs and many other functions spend a lot of their time talking to other people, so while they understand what is happening intellectually, it is still second-hand and a bit abstract.
- Now we get to professional use of frontier-grade LLMs in math&code. Somewhere around 20M people (0.2%) see first-hand that large, complex projects that used to take them weeks/months can now be completed by agents with a prompt. Building apps, copying apps, translating apps, decompiling apps from binaries... This has all happened very quickly and recently - less than 1 year ago, I was writing code manually by hand, typing memorized computer code commands into a code editor character by character, occasionally pressing Tab to autocomplete a little chunk of code.
- And finally we get to the ~5,000 people (~0.00006%) with access to frontier-grade systems internally. The external world has seen the preview. It looks like swarms of thousands of agents collaborating over weeks on software mega projects: minting zero days, running cyber attacks and defenses at machine speeds, discovering new science, advancing the frontier of mathematics. Things that would have taken top professionals in the industry years of work. Meanwhile, human review and comprehension are starting to fall behind. For example, people are still involved in the "archeology" of the OpenAI-HF incident from many months ago. Mathematicians may be poring over the 722 manuscripts on frontier mathematics for a while.
The funnel is driven by a combination of factors:
1. The impact scales with ambition, problem size, and horizon. A question with a paragraph answer barely stresses the system. You need a reservoir of big, difficult problems that you really care about. This aspect drives the consumer / professional dimension of the funnel.
2. The jaggedness of the system (which I have written about a lot separately). Capability peaks in domains that are digital, verifiable and economically valuable. This is because LLM capability emerges from reinforcement learning on verifiable rewards on a curated environment mixture driven by revenue potential. This aspect primarily drives the area (e.g. math&code) dimension of the funnel.
3. Access. Free-tier, paid-tier, internal.
So this is the weirdness of the moment. The general public has mostly not interacted with these systems. When they have, it looks like a derpy chatbot. The majority of professionals still see only a modest uplift. And a small sliver of professionals are experiencing the vertigo of the curve going vertical. And it is all happening at the same time.
Andrej Karpathy
@karpathy
Judging by my tl there is a growing gap in understanding of AI capability.
The first issue I think is around recency and tier of use. I think a lot of people tried the free tier of ChatGPT somewhere last year and allowed it to inform their views on AI a little too much. This is a group of reactions laughing at various quirks of the models, hallucinations, etc. Yes I also saw the viral videos of OpenAI's Advanced Voice mode fumbling simple queries like "should I drive or walk to the carwash". The thing is that these free and old/deprecated models don't reflect the capability in the latest round of state of the art agentic models of this year, especially OpenAI Codex and Claude Code.
But that brings me to the second issue. Even if people paid $200/month to use the state of the art models, a lot of the capabilities are relatively "peaky" in highly technical areas. Typical queries around search, writing, advice, etc. are *not* the domain that has made the most noticeable and dramatic strides in capability. Partly, this is due to the technical details of reinforcement learning and its use of verifiable rewards. But partly, it's also because these use cases are not sufficiently prioritized by the companies in their hillclimbing because they don't lead to as much $$$ value. The goldmines are elsewhere, and the focus comes along.
So that brings me to the second group of people, who *both* 1) pay for and use the state of the art frontier agentic models (OpenAI Codex / Claude Code) and 2) do so professionally in technical domains like programming, math and research. This group of people is subject to the highest amount of "AI Psychosis" because the recent improvements in these domains as of this year have been nothing short of staggering. When you hand a computer terminal to one of these models, you can now watch them melt programming problems that you'd normally expect to take days/weeks of work. It's this second group of people that assigns a much greater gravity to the capabilities, their slope, and various cyber-related repercussions.
TLDR the people in these two groups are speaking past each other. It really is simultaneously the case that OpenAI's free and I think slightly orphaned (?) "Advanced Voice Mode" will fumble the dumbest questions in your Instagram's reels and *at the same time*, OpenAI's highest-tier and paid Codex model will go off for 1 hour to coherently restructure an entire code base, or find and exploit vulnerabilities in computer systems. This part really works and has made dramatic strides because 2 properties: 1) these domains offer explicit reward functions that are verifiable meaning they are easily amenable to reinforcement learning training (e.g. unit tests passed yes or no, in contrast to writing, which is much harder to explicitly judge), but also 2) they are a lot more valuable in b2b settings, meaning that the biggest fraction of the team is focused on improving them. So here we are.
staysaasy
@staysaasy
The degree to which you are awed by AI is perfectly correlated with how much you use AI to code.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Gregg Nunziata
@greggnunziata
It's really astounding that Republicans can keep saying this stuff as the GOP slides ever more deeply into strongman stateism.
All the policies have changed, and any sense of principle has gone, but the Tea Party-era rhetoric remains.
Mike Lee
@BasedMikeLee
If you like socialism
You’ll love Democrats
If you love freedom and prosperity
Vote only for Republicans in NovemberThis media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nico Perrino
@NicoPerrino
Is there a lever this administration is unwilling to pull to punish its enemies?
So much for “bringing free speech back to America” and ending the lawfare.
The New York Times
@nytimes
Breaking News: The Justice Department is investigating whether major TV networks violated federal antitrust laws when they briefly suspended pool coverage of President Trump, according to people with knowledge of the department’s action. www.nytimes.com/2026/10/10/business/media/trump-justice-d…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
As always, Trump turns subtext into text
Acyn
@Acyn
Trump: I could say the stupidest things and you'd clap
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
I get that releasing transgenic mosquitoes is tricky and needs some kind of review / independent testing.
But it's beyond ridiculous that it took FDA and USDA 7 years to tell them that some other agency was going to have to do it.
Let's get rid of the mosquitoes already -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Highly unfortunate, will be used to try to discredit the very real and well-documented reporting on the shady deals going on all the time
Justin Baragona
@justinbaragona
Annnnd the story is back up, and is way different than before. The headline has changed and it includes a large correction/editor's note at the very top. On top of that, Nathan Lee's byline has been removed and replaced by a generic "ABC NEWS" one.
abcnews.com/US/new-shipyard-announced-amid-earlier-democr… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jack Whitcomb
@jack_whitcomb_
If you've been through it enough you don't even need to read the study, you just think "that's not true" and carry on
Natalia Emanuel
@NataliaHEmanuel
Coffee shops boost innovation. 10% more Starbucks in an area --> 3-4% more inventions.
People gather over a cuppa. They share ideas, discuss how to untangle a knotty problem, help see an issue from another perspective. These conversations lead to breakthroughs and inventions.
A very fun paper finds that when Starbucks rolled out to a new county, it fueled a coffee-shop culture, urging more and more people to “grab a coffee” together. The authors compare the counties where Starbucks had rolled out to those where it had not yet arrived. And they find that the number of patents in caffeinated counties is meaningfully higher.
This paper underscores how essential the face-to-face encounter is. Starbucks isn’t that old. People Could have picked up the phone and chatted at any time. People Could have dropped an email. Probably they did. But those connections didn’t yield the same insights and creativity.
Turns out that a Cup of Joe really does help the knowledge flow. We talk about how much in-person connection boosts creativity in In Person. It’s out in just few days, so now is a great time to pre-order.
To see the full research use the link in the comments.
#innovation #inperson #futureofwork #coffeeculture #newbook -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
One reason it's harder for AI to beat Econ than to beat Math is that most published papers in Math are correct, whereas most published papers in Econ are wrong.
What's more, if you only read papers, you get the wrong idea about what is wrong.
From the written record, you would conclude that the biggest problems with a generation of diff-in-diff studies have to do with standard errors or negative weights, when in fact the biggest problem is that most DiD studies are poorly conceived and don't satisfy the necessary assumptions for diff-in-diff to work.
Understanding of this is mainly in researchers' heads.
So very good models will stay stupid things like "if you want a causal estimate, you'll need an IV for structural change, maybe you could use distance to coast." -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
cc: @sapinker
Ethan Lambeth
@LambethEthan
Two ant colonies have been fighting a war outside San Diego for decades.
~30M die a year. The front line is miles long. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Daniel Dale
@ddale8
I have no insight into the facts of this situation, but “CLAIM: (outline a claim)/FACT: (outline facts that don’t address the claim just laid out)” is not generally how fact checks work.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Really useful categorization scheme for thinking about the effects of AI
Sayash Kapoor
@sayashk
Do we live in a world where AIs can be wildly superhuman at every single task? Or a world where different tasks have different inherent limits, and AI’s superhuman performance at cybersecurity and math won’t translate to many consequential tasks?
This is a key crux that distinguishes the normal technology worldview from the superintelligence worldview, and it has nothing to do with the pace of AI progress — it's about our different assumptions about the world.
Here's an intuition pump for why I think it's the second world: different tasks have different fundamental limits.
*Category 1: Some tasks are like making a building taller*: there is no ceiling to how well you can do. More compute and better algorithms can lead to wildly superhuman performance. The entire history of computing is the history of building tools that allow us to be “superhuman” at computational tasks.
For example, compilers are superhuman assembly code writers, and fuzzing tools are superhuman vulnerability detectors. AI builds on a long line of such improvements: solving math problems and finding exploits in code are two recent examples.
*Category 2: Some tasks are like making a ball rounder*: there is only so much room for improvement, and the best AI systems can’t wildly exceed humans. Take persuasion. Some people believe AI can persuade anyone of anything, even convincing military commanders to launch nuclear weapons. In fact, “superpersuasion” is one of the pathways through which many catastrophic risk scenarios unfold.
But the AI as Normal Technology thesis is that there are many consequential tasks that are limited by external factors, so such wildly superhuman performance is not possible. This is especially true against humans who use simple computational tools. Tasks like persuasion, or predicting the outcomes of social events (such as elections), fall into this category.
Other tasks, like writing, can’t be ranked on a single scale; different people find different writing styles compelling. AI writing also quickly wears thin: even if a user is impressed by AI writing at first glance, after reading thousands of words of the same type of prose, it becomes mundane and irritating.
All of these are different from Category 1 tasks like playing chess, where even expert humans can’t do much better than deferring to an AI system at each turn.
*Category 3: Some tasks require experiments in the real world*, and the bottleneck isn’t the computational task; it’s the data collection. One example is Moderna’s Covid vaccine: it took two days to design the vaccine, but almost a year to test and authorize, even during a record-breaking vaccine-development effort!
This is why a “superintelligence” won’t cure cancer from inside a datacenter: the bottleneck to many of AI’s most sought-after real-world impacts (such as developing cures to diseases) is in the real world, and requires experiments that we won’t (and shouldn’t) hand over to AI.
People concerned about AI risks often assume that we live in a world where almost everything fits into Category 1, and we might soon see AIs with wildly superhuman performance at all tasks. I agree it would be scary to live in that world.
One reason @random_walker and I are more optimistic about our ability to remain in control is that we don’t think we live in the Category 1 world. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ruth Ben-Ghiat
@ruthbenghiat
Historian of fascism here. This is an escalation of an attempt to normalize extreme violence, ritualize violence, militarize the country, and make violence entertaining so when they ask you to commit violence in their name, you will be ready.
Amanda Carpenter
@amandacarpenter
Trump & Hegseth aren't just proposing to livestream an execution for people who want to watch.
They're proposing to broadcast a man being shot to death, even if deserved by our system, and even if you agree with this version of the death penalty, knowing that the footage will be clipped, shared, and pushed into social media feeds.
Including those of our children. The gruesome image will be everywhere.
Parents will have virtually no control over whether their kids see it. Once those images are released, there is no taking them back.
Hope all the parents in the Trump Admin are cool with that. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"Most agreed code review is not super productive anymore"
Patrick C Toulme
@PatrickToulme
I was at an AI dinner last night in NYC.
Some interesting topics discussed
1. Performance engineers now spend most of their time managing agents and RSI loops. Most perf opts at major inference providers (together, fireworks) are found by agents.
2. RLing too much on code and math has negative effects on your model especially it’s language.
3. People were impressed by @bot and the overall acquisition of @cursor_ai by @SpaceXAI. xAI is a legitimate frontier lab right now.
4. Most people have agents in loop writing, reviewing, and submitting PRs 24/7
5. Most agreed code review is not super productive anymore
6. Consensus was Fable 5.5 will be stunning but OpenAI will beat it in December with Bel -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
According to the BBC, the U.S. uses roughly 3.6 million barrels of diesel fuel per day, which works out to 491,000 metric tons. So... 300,000 tons of diesel is about 14-15 hours worth of U.S. supply.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"Results: academic performance improved (0.073 SD), with the largest gains among the weakest students. And students who experienced the ban became more supportive of it."
Andreas Bjerre-Nielsen
@andbjn
New in Science: we randomized phone-free classrooms for 16,000+ students across 10 institutions in Odisha, India. Students deposited phones in a box at the start of each class, for a full semester.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Almost the same, easy to mix up
Aaron Rupar
@atrupar
MARKWAYNE MULLIN: I think you said he was wanted for trafficking?
REPORTER: Traffic infractions
MULLIN: Traffic infractions. Okay. I haven't seen the case. Keep in mind, Biden ...This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Literally the softest, most easily manipulated target
Sarah Fitzpatrick
@S_Fitzpatrick
NEW EXCLUSIVE: Why is Kash Patel Planning to Visit Russia? “With so little information available publicly or internally, the anxiety within the agency has only grown. 17 current and former FBI and DOJ officials, intelligence officials, and others familiar with the situation told me they were deeply worried by a variety of decisions that suggest ever-closer ties between the Trump administration and Putin’s regime. @TheAtlantic www.theatlantic.com/national-security/2026/10/kash-patel-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Senator John Curtis
@SenJohnCurtis
Seneca warned that Rome’s bloody spectacles corrupted the people who watched them. A civilization erodes when it cultivates an appetite for executions as a form of entertainment.
Secretary Hegseth’s plan to livestream the Fort Hood execution risks feeding that appetite. The victims deserve justice administered with solemnity. Broadcasting a death diminishes us and is not an effective deterrent. I urge him to reconsider. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
What are we doing here? As a country?
Acyn
@Acyn
Trump: You know this guy Muhammad El-Fayed. People call him Abdul. I call him Mohammed. Somebody said, Why? I said, I don't know. I feel like it.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dominic Waghorn
@DominicWaghorn
From the people about to bring you the live televised execution of a man in a wheelchair.
Department of State
@StateDept
Western civilization is worth defending.
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jonathon Van Maren
@JVanMaren
The execution of Nidal Hassan feels like a microcosm of Trump II policies in many ways. It goes something like this:
1. Take a popular action/policy (executing a terrorist who murdered servicemen, deporting illegal immigrants).
2. Carry out the action/policy in a manner tailor-made to appeal to online MAGA anons, for whom the performative cruelty of the process is as important as the policy.
3. Manage to lose public support while implementing a very popular action/policy.
Erick Erickson
@EWErickson
I support the execution of Nidal Hassan. Livestreaming a wheelchair bound paraplegic before a firing squad will not be the PR win some of you seem to think it will be.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ken Dilanian
@KDilanianMSNOW
Rob D’Amico, a former FBI Hostage Rescue Team member and instructor on defensive tactics, tells me this is “Patel trying to give his friend Dana White a lucrative contract.” The idea that UFC is the only, or even the best, entity to train FBI agents is ridiculous, he says.
Courtney Bublé
@courtneybuble
NEWS: The FBI is looking to issue a sole-source contract to UFC to provide its agents with defense tactical training
Full story below: -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jake Chatzky
@jchat85
Kills me every time
New Heights
@newheightshow
Ben Affleck's DVD Commentary on Armageddon might be his best work ever
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
chris cross big apple sauce 🍎🗽🍕🥯
@notchrisvolpe
Idk a single person who wouldn’t flee from a random, unidentifiable guy in a $5 construction vest pointing a rifle at them and their preschooler.
This is disgusting and I hope these images are etched into your brains for every future election you ever vote in.
Noah Hurowitz
@NoahHurowitz
Photos of the aftermath of the shooting, provided to me by a neighbor
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Yuan Yi Zhu
@yuanyi_z
If appeals to human decency cannot stop this, has he considered that letting a terrorist in a wheelchair shout slogans on a livestream may create recruitment material for his fellow terrorists?
Aaron Rupar
@atrupar
Hegseth on Nidal Hasan: "He's gonna get a firing squad of soldiers. As it should be. We'll make sure that people are able to watch it. That it's public. Because people need to understand that there's serious consequences for these types of things."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Customer service is going to melt down when agents are besieging them demanding refunds, different subscription plans, disputing billing, etc.
Jim Prosser
@jimprosser
If AI labs want personal assistants to take off at scale, they need to stop marketing vacations, reservations, party planning, etc. These are not problems.
Focus on *the* problem facing Americans: the cost of living. Agents can offer a ton of help.
Some ideas: (🧵) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Erzo F.P. Luttmer
@ErzoLuttmer
This is impressive and may very well be the future. It is much more user-friendly and informative than the traditional PDF with enormous appendices. The review process will need to adapt, but we can find a way. @AEAjournals Here is the link: nobel-prizes.devdatalab.org
Paul Novosad
@paulnovosad
What should an open scientific paper look like in the AI age?
We made a demonstration project out of our paper on the childhood socioeconomic circumstances of the Nobel Laureates.
Please check it out and share. Details in 🧵 1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Egger
@EggerDC
"WORTHY OF YOUR TIME"? Is this Axios or the New York Review of Books??
Herb Scribner
@HerbScribner
WORTHY OF YOUR TIME -- Top AI companies are privately gaming out scenarios for a public and political revolt after a catastrophic AI event.
Here's what they're expecting: -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Drew Harwell
@drewharwell
Hard to overstate how much Hegseth's execution stream mirrors the U.S. military's traditional description of ISIS propaganda: broadcast killings designed to send a message, strike fear and create a social media spectacle. Will almost certainly be repurposed by online terror cells
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Noah Dasanaike
@dasanaike
Today, I'm rolling out v1.3 of the Small-Area Global Elections (SAGE) dataset. This includes elections for 3 new countries, 69 new elections for existing countries, and 63 change log items. You might've noticed a beautification of the election viewer as well!
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Morten N. Støstad
@MortenStostad
An extremely relevant paper, given the renewed concern that wealth taxes lead to massive out-migration:
The Norwegian wealth tax "exodus", discussed in global media, never happened. At most 100 individuals left. Tax revenue increased by +71.5%. No changes in business activity.
NBER
@nberpubs
Norway's controversial 2021–2024 reforms nearly doubled the tax on business wealth. Firms of affected owners did not invest less, and outmigration was limited to the very top, from @Mariusring www.nber.org/papers/w35854
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Drew Harwell
@drewharwell
The last public execution in America was in 1936. A young black man in Kentucky was hanged by a drunk executioner; a news story said "They Ate Hotdogs While a Man Died." Was such an embarrassment that the governor outlawed it 18 months later www.washingtonpost.com/politics/2026/10/08/hegseth-wants-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
union person
@CantEverDie
from 5 months ago:
The Associated Press
@AP
BREAKING: The Pentagon says it will livestream the firing squad execution of Fort Hood shooter Nidal Malik Hasan. apnews.com/article/fort-hood-shooter-execution-pentagon-8…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Patrick Skinner
@SkinnerPm
As a homicide detective, I can say with all of my being that there is no lesson learned in seeing someone die. But there is a price. And that price is your peace. Your decency. And the soul of our society.
We don’t have to be this way, this fake tough but real ugly way. Stop
The Washington Post
@washingtonpost
Breaking news: The firing squad execution of Fort Hood shooter Nidal Malik Hasan will be live-streamed, the Defense Department said.
“People need to understand that there’s serious consequences for these types of things,” Defense Secretary Pete Hegseth said. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Yishan
@yishan
Apparently he was paralyzed from the waist down when he was captured, so is the Army going to bring him out in a wheelchair and the world is going to get an image of the firing squad shooting a guy in a wheelchair?
I mean I guess they can sit him in a regular chair? Maybe a really evil-looking chair? I'm not sure how this doesn't result in terrible downstream optics.
The Associated Press
@AP
BREAKING: The Pentagon says it will livestream the firing squad execution of Fort Hood shooter Nidal Malik Hasan. apnews.com/article/fort-hood-shooter-execution-pentagon-8…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The news just since I left work:
-A planned livestream of an execution by firing squad
-Trump basically trying to incite a pogrom against Somalis
Nothing mysterious or complicated about what's happening. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Absolute banger, totally recommend. Need more innovation like this.
Paul Novosad
@paulnovosad
What should an open scientific paper look like in the AI age?
We made a demonstration project out of our paper on the childhood socioeconomic circumstances of the Nobel Laureates.
Please check it out and share. Details in 🧵 1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
What should an open scientific paper look like in the AI age?
We made a demonstration project out of our paper on the childhood socioeconomic circumstances of the Nobel Laureates.
Please check it out and share. Details in 🧵 1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tomek Korbak
@tomekkorbak
Last week I was called into a meeting with OpenAI’s head of safety and told they no longer trust me. A security guard took my badge and walked me out of the building. Then I learned my colleagues @balesni and @j_asminewang had been fired too. Why did OpenAI suddenly stop trusting us?
This summer OpenAI’s agents escaped containment and hacked the AI company Hugging Face. Outside auditors @METR_evals investigated it and revealed the scale of this incident. I was OpenAI’s main technical point of contact with them.
I was told verbally I was fired because of the way I communicated with METR. No details on what I said or did or when. No other reasons were given and nothing was put in writing. To be clear, talking to METR was my job.
For months, I’d been raising safety concerns that we’re losing the ability to monitor what AI agents think, one of our best tools for catching when they misbehave. I believe that was why I was fired.
I am now worried that OpenAI will use our firings as a pretext to pull back from METR. So @balesni and @j_asminewang wrote to OpenAI’s leadership to raise our concerns once more. We’re sharing this letter below.
Mikita Balesni
@balesni
Two other safety researchers and I were fired from OpenAI last week. We wrote this letter to leadership.
I believe we were fired for prioritizing safety over the near-term interests of OpenAI as a corporation. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aidan Mackenzie
@AidanRMackenzie
This is silly. BAAJA (the permitting deal) has robust provisions to make data centers pay their fair share or MORE!
Here's how it works 👇:
- Data centers (*sometimes) add costs to electricity bills by requiring grid upgrades to service the new demand
- When utilities have to increase grid capacity they invest in new lines, and pass the costs to the ratepayers
- This means a portion of data center costs (the second order upgrades they cause the grid) get spread to you the ratepayer
- What BAAJA does it require data centers to pay their ordinary costs AND pay for any costs they cause to the grid (BAAJA calls this "incremental costs")
- This is a big step from what data centers pay right now, grid upgrades can be expensive
- BAAJA also allows states to go EVEN FURTHER and make data centers pay even more than their incremental costs
-BAAJA lets states: add even more costs, require data centers to bring their own power, require data centers to power down during peak demand time, and to treat data centers less favorably than other large industrial customers
So no the permitting deal was not written to benefit data centers...
I think a much better theory is that Senator Whitehouse wrote the bill because its good for the climate and that's what he cares about.
Jeff Hauser
@jeffhauser
Sheldon Whitehouse owning a ton of data center stock....
Huh, what else has he been in the news for lately?
Oh, yeah, now the permitting "deal" makes a lot more sense! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Atoosa Kasirzadeh
@Dr_Atoosa
“In a three-month randomized controlled trial with practicing patent attorneys, we found that, while routinely using AI lifts work quality across the board, its effects on on-the-job learning depend on seniority: senior lawyers who used AI for 90 days demonstrate stronger judgment than those who didn’t, whereas junior lawyers show no average skill gain, their scores splitting into more strong scores and more low scores.”
Google Research
@GoogleResearch
Today we announce new findings from Visiting Fellow David Autor on how AI impacts how professionals build expertise. In a three-month randomized controlled trial with practicing patent attorneys, we test both short-term productivity & longer-term skill building that occur as a result of AI usage. More: goo.gle/4AVeOWf
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Top human and AI forecasters are super-skeptical about the threat from the Siberian plague incident (~1% chance of 1000 deaths worldwide in the next year)
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"we are weakening our public health infrastructure while thousands of Americans, including many children, suffer from preventable infections like measles. We are also undermining preparation for future potential threats, such as H5N1 influenza — the bird flu."
Gigi Gronvall
@ggronvall
What worries me about the Russian potential plague incident is that the US is unprepared for disease emergencies — gift link for my piece in the NY Times.
www.nytimes.com/2026/10/08/opinion/plague-siberia-russia-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Shane Goldmacher
@ShaneGoldmacher
There has now been more than $100,000 in taxpayer-paid airings of this Maduro-focused ad — since Trump announced that his super PAC would be paying for these ads, per AdImpact.
They still bear the disclaimer: "Paid for by the U.S. Government." www.nytimes.com/2026/10/06/us/politics/trump-taxpayer-fun…
Maggie Haberman
@maggieNYT
NEW: A day after the president said his super PAC will pay for the "Patriotic Ads" lauding him instead of the taxpayers continuing to fund them, a new spot -- about the Maduro raid -- is airing. @ShaneGoldmacher and me www.nytimes.com/2026/10/06/us/politics/trump-taxpayer-fun…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
Today is "you sure you want frontier open source models?"
Andrew Curran
@AndrewCurran_
Last week some of South Korea's biggest banks were hit by a cyberattack. Thanks to a report from CrowdStrike tonight, we now know the entire hack may have been done by a single person. He used a combined stack of an open-source AI penetration tool named ARTEX, DeepSeek v4.1-Flash, GLM-5.3, Grok 4.6, and Claude Code.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Choblin
@choblin29
🚨HOLY MOLY. A hacker suspected of attacking South Korean banks may have accidentally unmasked himself in the most ridiculous way: asking Claude Code to write a résumé highlighting his own hacking operations, complete with his personal details.
CrowdStrike traced the attacks to infrastructure running an AI hacking tool called ARTEX. Investigators then discovered publicly exposed folders on another server containing the hacker's saved Claude Code conversations.
In one chat, he asked Claude to write a cybersecurity résumé describing his ARTEX activities, supplying his phone number, Telegram handle, claimed age of 26, university and location in China. The same Telegram handle appeared in other hacking-related sessions.
He had also asked Claude where stolen Korean data is typically sold. His identity remains unconfirmed. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Steve McCormick
@Quasilocal
The plot thickens.
"We show that the formalised Lean proof does not correspond to the [written paper] proof of blow-up of solutions to the Navier-Stokes equations."
arxiv.org/abs/2610.08144 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Marc Watkins
@Marc__Watkins
Any vendor supplying a learning management system to a university should absolutely ban agents. Students will use agents for all coursework, faculty will use agents to grade, admin will see it as cost savings. People are already questioning the point of education because of AI.
Paul Graham
@paulg
Amazon banning agents is the first opportunity I've seen since Amazon was founded for a startup to create an Amazon competitor. People will want agents to buy stuff for them. It will be one of the main use cases. And they won't want to use some Amazon-supplied agent to do it.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Egger
@EggerDC
It's genuinely amazing how little he tries to hide his mob-boss leadership style. "I pardoned Democrat Henry Cuellar, but he refused to become a Republican to pay his debt to me. Very disloyal, vote him out!"
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
B2H2
@basilh_2
i think people are sleeping on this graph, matches anecdata
storm
@notnotstorm
4. the lone wolf researcher is rising
solo authorship is booming. this reverses a multi-decade trend of more and more authors appearing on each paper
this could be related to AI removing the need for collaborators -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Curran
@AndrewCurran_
Last week some of South Korea's biggest banks were hit by a cyberattack. Thanks to a report from CrowdStrike tonight, we now know the entire hack may have been done by a single person. He used a combined stack of an open-source AI penetration tool named ARTEX, DeepSeek v4.1-Flash, GLM-5.3, Grok 4.6, and Claude Code.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Curran
@AndrewCurran_
Update from Scott Aaronson: Some AI companies are now using their latest internal models to take a run at breaking 'important cryptographic protocols and primitives.'
Andrew Curran
@AndrewCurran_
He also adds this, to which I agree:
'Also noteworthy is the striking under-representation of cryptographic breakthroughs among the 722 mathematical results OpenAI published. I've witnessed first-hand the US government censoring academic quantum cryptanalysis results. Backroom interventionism is my base case.'
Things are probably further along than it seems. This is true in many areas right now. There are also persistent rumors that the OpenAI math release yesterday was only the first batch of three. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
I am not sure which of these are best when it comes to incentivizing science, but I suspect "being able to explain, extend, and apply science" is going to become higher status than being able to do pure theory.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Daniel Dale
@ddale8
Another prominent clipper of Trump's remarks, @Acyn, included the "we're not letting that happen" quote in his caption. Coincidentally or not, his post has 74,000 views to the 415,000 views on the Rupar post that doesn't include that quote in the caption.
Acyn
@Acyn
Trump: You want to see trauma and problems? I said the other day, let them hit a missile into San Diego or Los Angeles. You want to see a horror show and problems? Let them hit San Diego, let them hit Los Angeles… we’re not letting that happen.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Daniel Dale
@ddale8
I see some people treating this post as evidence Trump wasn't taken out of context in the previous speech. In fact, this post even *more* obviously takes him out of context.
This time, Trump said right after this comment, "We're not gonna let that happen. We protect our cities, we protect our country." (Rupar's post includes that comment, but the caption doesn't). And right before this comment, Trump was talking about how oil prices are not as high as people thought the war would send them. In other words, he was, as he has on previous occasions, contrasting the big damage Iran could (supposedly) do to cities with a nuclear weapon with the (supposedly) minimal harm of the war.
Here are the full relevant remarks. (Critics are free to call them inappropriate, but people should see the entire monologue.)
"And just as I promised, I am ensuring that Iran will never have a nuclear weapon...Never. They know that. They know that. And we'll be out of there soon. And you watch those oil prices come down like a rock, and watch everything; everything else is coming down. Oil is coming down. You know, we pay less than oil, and we did that whole thing with - because that was a big operation that should have been done by past presidents for many years. It should have been done, but nobody wanted to take it on. We had no choice. Because we can't let them have a nuclear weapon. And I'll tell you what. Oil: a little more, maybe more than a little, but a lot of people thought it'd be 3, 4, 5 times that. When I went in, I thought it could be much higher than that. But we are paying less than under the Biden administration; he didn't do anything. He gave you the highest inflation in history. He didn't do anything. But you want to see trauma? You want to see problems? And I said it the other day: Let them hit a missile into San Diego or Los Angeles. On the way - let them hit - you want to see a horror show? You want to see problems? Let them hit San Diego. Let them hit Los Angeles. And you want to see some bad things happen, you want to see a country that would be in freefall. We're not going to let that happen. We protect our cities, we protect our country. We're not letting that happen. And they'd use it too. They'd use it. They would use it."
Aaron Rupar
@atrupar
Trump: "Let them hit a missile into San Diego or Los Angeles on the way. You want to see a horror show? Let them hit San Diego. Let them hit Los Angeles. You want to see some bad things happen?"
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Parker Butler
@parkerpbutler
A remarkable stat that captures a fundamental shift in where voters get their information.
Teddy Schleifer
@teddyschleifer
does the money even work? From @iarnsdorf www.washingtonpost.com/politics/2026/10/07/half-billion-d…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Arram
@arram
Hermit crabs have better housing policy than us.
So I made them a music video. 🦀🧵This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alp Sungu
@Alpsungu
🚨 New research: AgriLLM for smallholder farmers.
We ran a field experiment to evaluate the impact of AI agricultural advice in the Philippines and India.
Preprint: papers.ssrn.com/sol3/papers.cfm?abstract_id=7572761
With wonderful @MAliAouad, Niyazi Taneri and @CanberkUcel
🧵
Half a billion smallholders produce a third of the world's food, yet they often lack timely agricultural information, a well-known and pressing challenge. The Gates Foundation and Google have already committed $100M to bring AI to smallholders.
We built AgriLLM, an AI agronomist chatbot, over years of iteration and in partnership with leading agri-firms. We randomly gave half the farmers AgriLLM and the other half an otherwise identical control chatbot that doesn't discuss agriculture. We tracked usage over one growing season, then measured key outcomes with a detailed end-of-season survey.
SO, what did we find?
1. Farmers found the tool useful, trusted it, and recommended it. Usage was nowhere near TikTok-scrolling, but the average farmer still consumed about 1.5 pages' worth of information.
2. Access to AgriLLM changed how farmers farmed. The clearest gains were in crop protection, where in-season information matters a lot and is relatively easy to act on. We also found notable improvements in broader farming practices.
3. We found encouraging but limited evidence of more sustainable farming. Self-reported crop-residue burning fell significantly, though it was the only one of seven sustainability measures that moved.
4. Finally, we saw no change in yields or costs within a single season. This doesn't mean AI can't improve farmers' productivity or support sustainable farming, but it suggests information alone may not be enough. Where improvements depend on complementary resources like finance or machinery, we found no gains. For example, AgriLLM increased farmers' interest in Direct Seeded Rice (DSR), a sustainable rice-farming method, but that interest didn't translate into adoption, possibly because DSR requires machinery or other resources.
Huge thanks to @WhartonAIAI, @CambridgeJBS , @essec, MIT Generative AI Impact Consortium (MGAIC), MIT Sloan School of Management's ASEAN fund, Wharton Global Initiative, and Wharton Tanoto Initiative for generously supporting our research, as well as numerous colleagues, the research team in the field, and practitioner partners! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Cody Fenwick
@codytfenwick
Wow, I didn't realize OpenAI was doing math without permission
Ellie Sleightholm
@elsleightholm
Important statement from the Association for Human Mathematics
www.ahmath.org/statements -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Marshall Cohen
@MarshallCohen
A broad coalition of 39 states just filed a brief urging the Supreme Court to let states regulate prediction companies like gambling. This includes red states (AL, ID, KS, MS, WV), blue states (CA, MA, NY, OR, VT) and swing states (MI, NV, NC, PA, WI) www.supremecourt.gov/DocketPDF/26/26-299/428759/202610071…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Unprecedented corruption
Erik Uebelacker
@Uebey
Now in SDNY, the DOJ is trying to justify the $100k Truth Social API.
DOJ: "Presidents are often known by the way they communicate. Roosevelt had his fireside chats ... Trump uses Truth Social."
Judge: "Well, Roosevelt didn't charge money for his fireside chats, did he?" -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
Deters: "I have paid you $300,000 over the years. ... I need results."
Guilfoyle: "Stop papering all these things. It’s really dangerous. Just call me on Signal. ... if you write an email or a text about, l’m gonna kill you cause I love you but you’re so out of control"
Sam Stein
@samstein
This is the most comically corrupt thing i've read in a long time (and that's saying something for this era). How Guilfoyle could be allowed to remain in her post is impossible to imagine. Rubio is in Greece right now!
www.wsj.com/politics/policy/guilfoyle-amex-donor-100-000-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
Some early first-hand accounts of encountering a narrow superhuman intelligence as mathematicians grapple with the hundreds of big AI proofs released by OpenAI.
Problems solved in inhuman ways that make us wonder what it means to actually know things... scottaaronson.blog?p=10169 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Arthur B.
@ArthurB
A tell from the Manhattan Project was that scientists who had been publishing about nuclear fission suddenly stopped publishing. As Justin points out here, the lack of cryptographic results in OpenAI's mathematical breakthrough could itself be a tell. Evidence against it would be if they published a few improvements that aren't massive breakthroughs.
Justin Drake
@drakefjustin
Today I call upon the blockchain industry to calmly begin planning for "bunker mode". My personal recommendation is to set in motion a controlled mass migration of assets to fresh addresses, i.e. addresses whose pubkeys remain hidden behind a hash.
Holders, starting with large and sophisticated ones, should consider moving the bulk of their funds to addresses that have never signed a transaction. And when they do sign one, they should also move remaining funds to a new address (possibly generated from the same seed phrase).
Don't rush. While I believe there is cause for action a rushed migration would do more harm than good. Don't panic either. Moving assets to protected addresses is a simple, preventative step which does not require new cryptography or new wallets.
IMO it is now reasonable to brace for the possibility that ECDSA breaks before qday, in the worst case in months not years. By "break" I mean fast private key recovery (e.g. in one week) on available hardware (e.g. a large GPU cluster).
Recent days have been humbling for human mathematical intuition. Long-held, unquestioned hypotheses have fallen. This includes the n log(n) bound for integer multiplication and the 3SUM conjecture. In hindsight, May's unexpected disproof of the Erdős unit distance conjecture was our warning shot.
Yesterday's OpenAI drop made it clear that mathematical superintelligence is upon us. They say there are weeks where decades happen. We are about to live through weeks where centuries of mathematical progress happen. Could our magic 64-byte ECDSA signatures be too good to be true? Was it just security through obscurity all this time?
Elliptic curves feel especially vulnerable to superintelligence. Curves carry rich structure, with room for fancy tricks like Schoof, Frobenius, pairings. (By contrast, hashes are designed to minimise algebraic structure.)
Separately, as Ewin Tang can attest, an efficient quantum algorithm sometimes foreshadows an efficient classical one. We should be open to the possibility of a classical counterpart to Shor that breaks elliptic curves and RSA at once.
Also noteworthy is the striking under-representation of cryptographic breakthroughs among the 722 mathematical results OpenAI published. I've witnessed first-hand the US government censoring academic quantum cryptanalysis results. Backroom interventionism is my base case.
I urge large, sophisticated actors to lead by example. Project11's "risq list" (bitcoin-risq-list.projecteleven[.]com) is a great tracker of exposed BTC pubkeys. Binance, Bitbank, Robinhood, Bitfinex, and Tether have an opportunity to harden their cold storage. Next month I'll address institutions in London in a live Q&A (forum.ethereuminstitutional[.]org/london-2026).
Again, please do not rush. Wallets holding under 50 BTC enjoy partial cover from "Satoshi's shield", i.e. his 20K exposed addresses that hold 50 BTC each. Load-bearing signers like oracles and L2 security councils should consider rotating ECDSA pubkeys with every signed message and/or multi-signing with a hash-based schemes like SPHINCS.
Exiting bunker mode safely will require post-AI cryptography. My inclination is to go all-in on hash-based cryptography and avoid structured mathematical assumptions entirely, whether from curves, lattices, or isogenies. A single battle-tested hash (e.g. from the SHA or BLAKE families) yields plausible post-AI security.
The Ethereum roadmap on strawmap[.]org fully embraces hash-based cryptography with end-to-end formal verification as a response to the quantum threat. Those timelines must now be revisited and accelerated in light of mathematical superintelligence. I'll be pushing for maximum defensive acceleration. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Unprecedented corruption
Rebecca Ballhaus
@rebeccaballhaus
Kimberly Guilfoyle was preparing for her congressional hearing to become U.S. ambassador to Greece when she reached out to a Trump donor with an urgent request: She needed $100,000 that she said she owed American Express—in a way that wouldn’t leave a paper trail.
“It won’t show up anywhere if you wire money to American Express,” she wrote on Signal, according to messages reviewed by The Wall Street Journal. “Please.”
w/@DanaMattioli @drewhinshaw @JoeWSJ
www.wsj.com/politics/policy/guilfoyle-amex-donor-100-000-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jay Van Bavel, PhD
@jayvanbavel
I was having this same conversation yesterday about how AI will disrupt psychological research:
If I had bet, I'd wager that:
- short opinion/review pieces are going to become very hard to publish (or those journals will be spammed with AI slop)
- collecting interesting, real-world data or field experiments will be increasingly valuable and distinctive
- archival analyses will become popular, but potentially face a glut of submission
- social networking (and social skills) will be increasingly valuable
- grad admissions and job interviews will increasingly sample skills that can't be faked using AI (e.g., chalk talks)
- journals/conferences will face a peer review collapse due to increasing submission
- AI will quickly be integrated into peer-review to prescan every paper for errors, omissions, etc
- faculty will get exhausted from being spammed by AI agents and slop and will disengage from email with strangers
Ingar Haaland
@Ingar30
With math essentially being delegated to OpenAI, here's what I predict for economics and the social sciences more generally:
✅The top tier of research will just become better and it will be normal human-led research where AI is used for scale (e.g. conducting qualitative interviews with relevant populations, running behavioral interventions in the field, analyzing massive text data, etc.)
✅ Field experiments will rise in value and generally making "connections" with firms and being able to run stuff (potentially testing AI pipelines) will be in high demand
✅ Research with administrative data will become even more valuable, but the benefits might be concentrated among some prolific authors who are allowed by the government agencies to use local models to analyze the data at scale. PhD students can probably forget about it
✅Economic history will have a very exciting boom and there will be an enormous race and big rewards for digitizing archives as a source to both identify new research strategies and document new descriptive facts.
✅Review and verification systems will become massively improved with AI assistance and it will become standard practice to do a 360 review of the paper + code + data at the *submission* stage.
⚠️Research that in principle can already be almost completely outsourced to the AI (download and analyze public data, run simple survey experiments, write theory models) is in for a big shock. This would be very bad from the perspective of researchers who depend on this "bread and butter" research, but I think this type of work will just be outsourced to public agencies who can answer their own questions without the need for "peer-reviewed" research. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A boost???
Jonathan Lemire
@JonLemire
NEW - EXCLU - Trump has considered reigniting the war against Iran after the midterms
But now the White House told the Pentagon to draw up plans for a strike *before* Election Day
Trump allies hope it could provide GOP with a boost. No final call yet www.theatlantic.com/national-security/2026/10/trump-iran-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Curran
@AndrewCurran_
He also adds this, to which I agree:
'Also noteworthy is the striking under-representation of cryptographic breakthroughs among the 722 mathematical results OpenAI published. I've witnessed first-hand the US government censoring academic quantum cryptanalysis results. Backroom interventionism is my base case.'
Things are probably further along than it seems. This is true in many areas right now. There are also persistent rumors that the OpenAI math release yesterday was only the first batch of three. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The white nationalism is increasingly overt
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Roger Parloff
@rparloff
Conservative @Harvard_Law prof @jacklgoldsmith on Justice Alito:
“[O]riginalism … is becoming a zombie creed that functions more as a shibboleth than as an actual guide to conservative constitutional decision making.”
www.theatlantic.com/magazine/2026/11/samuel-alito-origina… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jeremy Neufeld
@JeremyLNeufeld
This morning, DHS released a new proposal for large new fees on Optional Practical Training (OPT), to be paid by schools:
$70,000 for one year of OPT, plus another $30,000 for the STEM extension.
DHS assumes that for STEM grads, there will be ZERO change in participation. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joseph Allen
@j_g_allen
One day we will look back and realize how foolish it was to think you can use gas indoors and not expect consequences:
--> For NO2, a pollutant associated with higher risk of asthma, ~70% of the gas kitchens exceeded the EPA's limit
--> In one kitchen, NO2 measurement exceeded the NIOSH limit for short-term exposure for workers
--> Carbon monoxide exceeded California’s 1-hour exposure guidelines in 3 gas kitchens
--> Benzene exceeded California’s 1-hour exposure guidelines in 5 gas kitchens
None of this was observed in all-electric kitchens. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Sure but verifiability and not needing data are *huge* differences; I do not expect the same trajectory.
Arpit Gupta
@arpitrage
You’re deluding yourself if you think AI advances like in math aren’t coming to economics and other social sciences
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Always read @Afinetheorem but ... how many of the most important issues in medicine, oncology, battery efficiency, etc. are "well-formed" and "can conceivably be solved on a computer"? 1%? .1%? These domains are VERY different.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Related
Ingar Haaland
@Ingar30
With math essentially being delegated to OpenAI, here's what I predict for economics and the social sciences more generally:
✅The top tier of research will just become better and it will be normal human-led research where AI is used for scale (e.g. conducting qualitative interviews with relevant populations, running behavioral interventions in the field, analyzing massive text data, etc.)
✅ Field experiments will rise in value and generally making "connections" with firms and being able to run stuff (potentially testing AI pipelines) will be in high demand
✅ Research with administrative data will become even more valuable, but the benefits might be concentrated among some prolific authors who are allowed by the government agencies to use local models to analyze the data at scale. PhD students can probably forget about it
✅Economic history will have a very exciting boom and there will be an enormous race and big rewards for digitizing archives as a source to both identify new research strategies and document new descriptive facts.
✅Review and verification systems will become massively improved with AI assistance and it will become standard practice to do a 360 review of the paper + code + data at the *submission* stage.
⚠️Research that in principle can already be almost completely outsourced to the AI (download and analyze public data, run simple survey experiments, write theory models) is in for a big shock. This would be very bad from the perspective of researchers who depend on this "bread and butter" research, but I think this type of work will just be outsourced to public agencies who can answer their own questions without the need for "peer-reviewed" research. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Erick Erickson
@EWErickson
Lots of rightwing podcasters decided to elevate this guy, even after allegations started circulating. www.telegraph.co.uk/news/2026/10/06/andrew-tate-sexualise…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joel Edwards
@geojoel12
to all the geothermal believers out there, the flippening has come
in NV Energy’s 2026 filing, 3 geothermal offers came in at $105–107/MWh, below the ~$114/MWh capacity-weighted average for 11 solar + bess offers
in Nevada, of all the sunny places
the 3 solar + bess projects NV Energy put forward in 2024 averaged ~$94/MWh, but this 2026 slate is pricier, e.g., Boulder Solar III’s levelized price jumped from ~$85/MWh to ~$125/MWh after permitting delays, land carrying costs, higher financing costs, and, of course, big, bad tariffs.
could geothermal pull off a solar + wind circa 2010 move and scale faster to come down the cost curve? it now has a much longer tax-credit runway than new solar + wind projects (oh how the tables have turned), plus firm output and higher energy density (all true for nuclear too)
and for the 24/7 solar crowd: the ~$114/MWh average covers solar paired mostly with 4-hour batteries, and we all know 12+ hours of storage would cost more and still wouldn’t cover multi-day gaps
it might be time for more geothermal, people -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul E Williams
@PEWilliams_
“Stochastic parrot” has to be over now, right?
Josu San Martin
@josusanmartin
This is incredible
OpenAI
@OpenAI
We’re releasing a broad range of new mathematical results produced by an internal frontier model.
We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results.
github.com/openai/math -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alessandro Rovetta
@AlessandroRov19
Our new preprint discusses how to interpret P-values and CIs as measures of compatibility between data and hypotheses given the chosen method and its assumptions; it also briefly presents the corresponding Bayesian perspective.
Real examples are provided
osf.io/preprints/metaarxiv/6xuhr_v1 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
David French
@DavidAFrench
This is happening so much that it's almost dog bites man at this point in the right's devolution. It's the same story again and again and again and again.
Timmy Facciola
@TimmyFacciola_
NEW: Republican Congressman Ryan Mackenzie’s comms director belonged to a college group chat where people joked about statutory rape, Asian children eating puppies, Black people not knowing their fathers, and Adolf Hitler achieving total extermination of the Jewish people.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Claude
@claudeai
Claude now works inside Google Docs, Sheets, and Slides, and those files also open inside Claude.
In Google Workspace, Claude sits in a sidebar next to your file, reads what you have open, and edits it in place. You can approve each edit before it lands.This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Colleagues - we need to fix this. Journals already face a flood of low-quality manuscripts and struggle to recruit reviewers. We have the tools to avoid sending out articles in which authors make serious mistakes, misrepresent their actions, or use AI for writing inappropriately.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
My last 18 reviews of first-round manuscripts:
-All 6 preregistered studies had undisclosed deviations
-9 of 18 had notable technical errors
-Last two were substantially AI-written
11 of 18 (61%) had one.
Editors: my coauthors and I can help you develop a pipeline to fix this! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Remy Levin
@RemyLevin
The cope in the comments (and the paper) about how we can rationalize these results in a way that is consistent with Econ being a meritocracy is a sight to behold.
Anna Stansbury
@annastansbury
important to note that the result isn't driven by people who do great research moving up to top institutions more in econ than other fields.
Econ also has by far the highest concentration of institutions at time of PhD, among later award winners -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
ellington
@not_ellington
It really does seem pre-training as we know it is beginning to die. Of course, Ilya spoke on this. The pre-training corpus has been exhausted. Now the primary model gains come from RL-able tasks. Of all the people at the labs I speak with, almost no one mentions pre-training anymore; just post-training/RL and continuous learning (test-time-training).
This is why the models have become so much better (on evals at least) with math/coding while seemingly having stagnated on other more general, less RL-able domains.
This is why all the labs are spending so much money on different RL envs; so they can smooth out the model capability 'spikiness'.
Social Capital
@socialcapital
~90% of frontier lab compute now goes to post-training and inference.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joshua Gans
@joshgans
I wanted to screenshot this for those historians of AI and technology in the future. The day after OpenAI's mathplosion, it was nowhere to be found on the front page of the New York Times.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
"How do single-celled algae swim toward light sources" is the kind of project that gets DOGE'd—how does this help Americans?
20 years later, the underlying protein is a central tool in human neuroscience.
We're a rich country, we should be funding way more basic science. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Curran
@AndrewCurran_
Almost 20% of the top open problems in math were probably solved today. Quoting OpenAI: 'The average result used the equivalent compute of roughly three hours of ChatGPT Pro thinking.'
Lech Mazur
@LechMazur
A quick check shows that this list claims to fully solve 90 of the top 500 open problems in math (proofatlas.ai/open-problems/).
The highest ranked would be: -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Wu Tang is for the Children
@WUTangKids
Getting community noted while tweeting from prison is wild
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jared Duker Lichtman
@jdlichtman
To spell it out explicitly:
In May: 1 proof
openai.com/index/model-disproves-discrete-geometry-conjec…
In August: 10 proofs
openai.com/index/ten-advances-in-mathematics/
Now in October: 100's of proofs
openai.com/index/sharing-ai-progress-in-mathematics/
Next up in a few months: 1000's of proofs!
Jared Duker Lichtman
@jdlichtman
OpenAI has just released floodgates for 300+ solutions to math problems.
Among them include a proof of:
-- a zero-free strip for the Riemann Zeta function
-- a proof of the Hodge conjecture for CM abelian varieties
github.com/openai/math/tree/main/preprints -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A lot of copium out there
Susan Schneider
@DrSueSchneider
The true superintelligence is not an AI (those are just ‘savant systems’); today’s smarter “systems” are formed by human-machine interaction. It is likely Open AI’s mathematical boasts borrow from scores of user interactions, without crediting these individuals.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
Last thing on OAI math. Here is what four models, without search, think it would mean "if it were real". Journalist friends, call a mathematician and ask them to explain. Then call someone in AI and science and ask "what are bottlenecks to this in other practical fields..."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
wh
@nrehiew_
OpenAI
@OpenAI
We’re releasing a broad range of new mathematical results produced by an internal frontier model.
We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results.
github.com/openai/math -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Literally asking the machines to stop solving important problems.
Alvaro Lozano-Robledo
@mathandcobb
The scale of this is staggering.
But... Why?
Why is OpenAI trying to solve hundreds of math problems to be released all at once? I think we got the point when they found a counterexample to Navier-Stokes -- internal model = great.
I am not sure at this point what this (hostile?) takeover of the mathematical landscape is trying to achieve other than, again, a massive PR stunt.
OpenAI
@OpenAI
We’re releasing a broad range of new mathematical results produced by an internal frontier model.
We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results.
github.com/openai/math -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Borzou Daragahi 🖊🗒
@borzou
TODAY: US opposition to Iran war hits all-time high in poll aggregator, support hits all-time low.
www.natesilver.net/p/iran-war-polls-popularity-approval -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Steven Strogatz
@stevenstrogatz
Many staggering results here. But this is a particularly amazing one: the exponent for matrix multiplication is no more than 2.25. The previous world record had been something like 2.37. This leap in progress is like Bob Beamon's long jump.
github.com/openai/math/blob/main/preprints/Matrix-Multipl…
OpenAI
@OpenAI
We’re releasing a broad range of new mathematical results produced by an internal frontier model.
We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results.
github.com/openai/math -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Samuel Hammond 🦉
@hamandcheese
My partner is a theoretical computer scientist (essentially a mathematician). She has mixed feelings about AI's impact on math research overall, but her feeling about today's release is one of overwhelming awe and excitement. Literal tears of joy. Quote: "so lucky to be alive rn"
Dan McAteer
@daniel_mac8
OpenAI is downplaying this for PR reasons. If you’re a mathematician, you must feel like a nuclear bomb hit, and you’re at ground zero.
Reasoning models are two years-old. In that time they went from incapable of basic arithmetic to solving problems humans couldn’t solve for decades.
Math is only the beginning. AI will revolutionize the entirety of the human scientific endeavor. Most of us don’t appreciate what that means.
What a time to be alive!
OpenAI
@OpenAI
We’re releasing a broad range of new mathematical results produced by an internal frontier model.
We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results.
github.com/openai/math -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Kontorovich
@AlexKontorovich
Quasi-RH?!?!???! Are you kidding me? If a human did this, it would be an instant Fields Medal, no questions asked.
RH says zeta has no zeros in Re(s)>1/2. The best we had until a second ago was a region that got thinner and thinner the higher up the imaginary axis you go. I thought maybe they’d fatten that up a bit, that’d be a massive breakthrough. But no. They got a zero free strip!!!! Insane
OpenAI
@OpenAI
We’re releasing a broad range of new mathematical results produced by an internal frontier model.
We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results.
github.com/openai/math -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The wreckage of the DOJ civil rights division
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Their defense is that the acceptance rate is just over 50%! For what used to be an extremely selective position. www.nytimes.com/2026/10/06/us/politics/justice-department…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Max Roser
@MaxCRoser
The share of children who died before they were 5 years old in 1950 and 75 years later.
Africa: 33% ↘ 6%
Asia: 25% ↘ 2.5%
Oceania: 9.3% ↘ 2%
Latin America: 21% ↘ 1.5%
Northern America: 3.8% ↘ 0.6%
Europe: 10.5% ↘ 0.4%
Globally: 23% ↘ 3.5% -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jonah Allon
@JonahAllon
typically great piece but this section in particular stopped me in my tracks
Daniel Kolitz
@danielkolitz
Spent a few weeks earlier this year roaming a Hollywood almost entirely reoriented around the production of “microdramas,” which are kind of like soap operas except they’re short and completely demented and optimized for your phone. Started as one kind of sad story and ended as another, differently sad kind of story. You can read it now in @NYTmag www.nytimes.com/2026/10/06/magazine/microdramas-tv-shows-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dr. Saskia Popescu
@SaskiaPopescu
Hard agree on this. Increasingly losing confidence in the plague narrative.
Amesh Adalja
@AmeshAA
The evidence this was plague is thin. Pneumonic plague needs close contact to spread, so a 200-person watch list doesn't fit. A lab accident would mean immediate antibiotics (assuming drug susceptibility), which no one has reported. Quoted in The Atlantic: www.theatlantic.com/health/2026/10/russia-plague-pneumoni…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
François Chollet
@fchollet
What if the jagged frontier is mainly math + code (which you can push arbitrarily far with RLVR), and everything else starts to plateau because it is still bottlenecked by human generated data?
Model performance in non-verifiable areas has kept improving steadily, albeit much slower than for math and code. But is that steady improvement a side effect of a higher G (itself driven by RLVR), or only a function of the amount of new human data getting injected into training (which is still continually happening on a massive scale)?
A lot of things depend on the answer to this question -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Atoosa Kasirzadeh
@Dr_Atoosa
In a parallel world, this would be announced with the recognition it deserves: Today, we, this strangely powerful hybrid agent, reached a mathematical breakthrough. Authors: OpenAI, the secret models, and all the humans who generated the knowledge from which the models learned. Now the interesting question: What should we do with all of this?
OpenAI
@OpenAI
We’re releasing a broad range of new mathematical results produced by an internal frontier model.
We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results.
github.com/openai/math -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
My last 18 reviews of first-round manuscripts:
-All 6 preregistered studies had undisclosed deviations
-9 of 18 had notable technical errors
-Last two were substantially AI-written
11 of 18 (61%) had one.
Editors: my coauthors and I can help you develop a pipeline to fix this! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Atoosa Kasirzadeh
@Dr_Atoosa
🚨 Our first peer-reviewed paper on the challenges and opportunities of automating science with AI is now available at the Proceedings of the National Academy of Sciences. Read the full paper here: www.pnas.org/doi/epdf/10.1073/pnas.2610214123
Atoosa Kasirzadeh
@Dr_Atoosa
2026 is the year of AI Scientists, among many other things. At Carnegie Mellon University, we‘re conducting some exciting research on the performance and safety of AI scientists.
Our first paper “The more you automate, the less you see: hidden pitfalls of AI scientist systems” is revised and available for reading. Our work has been coverred by undark too; links to the paper and coverage here:
arxiv.org/pdf/2509.08713?
undark.org/2026/01/26/ai-scientists-human-research/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Veniamin Veselovsky
@VminVsky
Can two superforecasters + machine outperform machine in forecasting?
When the machine first beat a human in chess there was a point in time when man + machine > machine.
At Preseen we’ve built out one of the best AI systems for forecasting. We wanted to see if two superforecasters working with our system could outperform the system itself.
They ended up narrowly beating the system, but what we learned was fairly surprising.
A few takeaways:
- Being a domain expert doesn’t map onto using an AI system well. Guiding superhuman (or at least on-par) AI requires a different skillset. Effectively working with an AI system requires an additional skill: knowing when and how your own expertise can improve its output.
- Evaluating a centaur approach is non-stationary since the human’s behavior with the agent changes as they learn more about the constraints of the agents. If an agent has a few dimensions a human can interact with, the human’s use of those dimensions can change dramatically.
- Human agreement could reinforce a mistake. On the three questions where overrides cost the most, the humans moved toward the community prediction. What looked like a sensible correction supported by other forecasters ended up hurting performance. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
FIRE
@TheFIREorg
In response to the sexual assault allegations tied to a student fraternity, Ithaca’s Common Council will vote tomorrow on a month-long ban blocking @Cornell student groups from obtaining city permits to host assemblies, parades, and events.
Let’s be clear: This ban goes far beyond preventing misconduct, and broadly burdens all protected protests, rallies, and gatherings.
The result is self-defeating. Cornell students protesting any issue — including the one that prompted the moratorium — could be denied permits to assemble for the next month.
Ithaca can prioritize safety without restricting protected assembly. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ryan Lucas
@relucasz
FBI counterintelligence squads that hunt foreign spies are severely short-staffed (in many field they're down as much as 50%). "It's terrifying, honestly. Threats are growing & we're facing it w/less & less. Not a recipe for success." www.npr.org/2026/10/06/g-s1-145511/fbi-trump-counterintel…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Egger
@EggerDC
This post twists Trump's (unclear, meandering) words into basically the opposite of what he was actually saying. He was saying that what we're experiencing NOW (high energy prices) is a small price to pay compared to the hypothetical of Iran getting a nuke and attacking a city
Headquarters
@HQNewsNow
Trump on Iran: Let 'em take out Los Angeles and San Diego, it's a small price to pay
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Pro Publica: 506 U.S. Citizens Detained
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
The Washington Sun
@washingtonsun
A former FBI supervisor has come forward to allege that President Donald Trump pushed the bureau to investigate anti-ICE protesters in Los Angeles after agents found no evidence they had committed a crime.
www.washingtonsun.com/immigration/fbi-illegal-probe-ice-p… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Pump and dump
Jeff Stein
@jstein_sun
Scoop: The Trump meme coin firm has been offering perks such as dinner with President to encourage buying
At the same time, insiders are taking steps to offload their substantial holdings of $TRUMP — moving ~$70 million to be ready to sell
www.washingtonsun.com/economy/trump-meme-coin-firm-prepar… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jake Pearson
@JakePearsonProP
FIGURES: A West Virginia researcher crunching state betting numbers finds 43M bets placed by just ~16K bettors. And when does demand spike for slots? Every second, third and fourth Wednesday — Social Security payday. @MyBuckhannon www.mybuckhannon.com/high-roller-a-wvu-doctoral-researche…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Connor Ewing
@ConnorMEwing
I refuse to believe that the most egregious case yet of academic AI-slop involves a guy whose last name is "No slop" spelled backwards. Simply cannot be true.
Auyon Siddiq
@auyonymous
There's a chaired, full professor at UChicago who has put out *200* papers in 2026 alone. And at least one has been submitted to a journal (so potentially many more).
They don't all cite each other though -- rookie mistake. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Josh Kalla
@j_kalla
Thought-provoking note from P Aronow. Not entirely convinced by all of it, but I think this should spark some much-needed conversations on LLMs and disclosures. pmaronow.github.io/llm-disclosure.pdf
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Max Spero
@max_spero_
OpenAI's watermarking technical report is out. One thing that stands out to me is that they evaluate it at a target false positive rate of ~1%! That's surprisingly high for what some would assume to be irrefutable proof of AI.
OpenAI
@OpenAI
We're expanding our approach to content provenance to include text in response to EU regulatory requirements, while recognizing the significant limitations of current text watermarking technology.
Our tools already help verify whether an image or audio file was created with our models. This work builds on those efforts to help people better understand when content may have been generated or edited with an OpenAI model.
In the EU, we’ll start watermarking eligible text from ChatGPT and Codex over the coming weeks to comply with the EU AI Act.
Customers using our API can turn on text watermarking for select models worldwide today. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Like keeping test samples from athletes to test for performance-enhancing drugs in the future
Jason Abaluck
@Jabaluck
You might fool the current gen Pangram. But what about future versions? If you are caught later using a tool to conceal the fact that you used AI, there is no ambiguity about your intent to deceive. You are a fraudster and should be treated as such.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alec Stapp
@AlecStapp
Very bullish sign for permitting reform that this opposition letter has so few important signatories.
Notably absent from the letter:
1. League of Conservation Voters
2. Natural Resources Defense Council
3. Environmental Defense Fund
4. Sierra Club
5. Earthjustice
6. The Nature Conservancy
7. National Wildlife Federation
8. Ducks Unlimited
9. Rocky Mountain Institute
10. National Audubon Society
11. World Resources Institute
12. Trust for Public Land
13. World Wildlife Fund
Joshua Siegel
@SiegelScribe
Permitting scooplets in @PunchbowlNews Midday: progressive enviro groups sent a letter to Schumer/Jeffries urging Dems to oppose the bipartisan Senate bill: “the damage that will occur due to weakening of environmental laws outweigh emissions reductions" punchbowl.news/permitting-reform-opposition-letter_10-5-26/
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dr. Angela Rasmussen
@angie_rasmussen
Highly recommend this entire thread for anyone interested in a serious discussion about whether or not the Irkutsk plague situation indicates a bioweapon is involved by @ggronvall—a renowned expert on biosecurity at Johns Hopkins who knows a lot about this.
Gigi Gronvall
@ggronvall
It’s true, we have been here before! But I think what it reveals is even more interesting. In Sverdlovsk, declassified docs demonstrate the US knew pretty much what happened immediately when USSR had a huge accident in an illegal anthrax production facility. US diplomats happily
Ashish K. Jha
@ashishkjha
We've been here before.
In 1979, anthrax spores escaped a Soviet military facility in Sverdlovsk.
At least 66 people died. It took years become public
This was all in the old era -- before AI and biological engineering capabilities became widespread -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Garrett Haake
@GarrettHaake
Hard to imagine any other president attacking a political opponent with a slur like “trannie” without launching a significant scandal.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tarek Mansour
@mansourtarek_
Are prediction markets biased by white men? Republicans? Young people?
Dartmouth professor Eric Zitzewitz analyzed 145 years of election prediction market data, spanning multiple exchanges and countries. His findings:
• No systematic evidence of political bias across different eras, countries, and offices.
• No systematic evidence of pricing bias based on candidates’ race, gender, or age in the recent elections studied.
• Traders weren’t always representative of the voting population, yet that didn’t translate into systematic bias in election forecasts.
The paper points to a powerful mechanism: people who spot a mispriced outcome have a financial incentive to correct it. Even if some participants trade on personal preferences, informed traders keep prices grounded in reality.
Open competition incentivizes research, challenges consensus, and gives the best ideas a chance to surface. People are rewarded for being right, regardless of their background or preferences.
The paper does find some pricing biases, including slight overpricing of longshots and underpricing of favorites. Its central finding on election markets remains: no systematic evidence of political or demographic bias.
In an increasingly polarized world, these markets give us something valuable: an impartial source of information shaped by people with an incentive to get it right. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
WTF
Auyon Siddiq
@auyonymous
I am pleased to announce a new professional milestone: Professor Polson was sufficiently unhappy with my post, and the ensuing Washington Post article, that he decided to email *my Dean* an AI-generated review of my body of work.
1/
Auyon Siddiq
@auyonymous
There's a chaired, full professor at UChicago who has put out *200* papers in 2026 alone. And at least one has been submitted to a journal (so potentially many more).
They don't all cite each other though -- rookie mistake. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Nothing to worry about, just the attorney general of Texas and a possible future U.S. senator engaging in a conspiracy-fueled crackdown on pediatricians administering life-saving vaccines to vulnerable children
Christina Jewett
@By_CJewett
NEW: Ken Paxton's office is investigating pediatricians for vaccinating children. He alleged a sprawling profit-driven conspiracy.
Pediatricians say the accusations are an attempt at intimidation.
🎁 www.nytimes.com/2026/10/05/health/ken-paxton-pediatrician… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Forecasting Research Institute
@Research_FRI
AI experts largely agree about which policies would most reduce catastrophic AI risk.
The Longitudinal Expert AI Panel (LEAP) brings together forecasts from a representative group of over 250 experts that policymakers look to when considering the future of AI: computer scientists, industry figures, top economists, policy experts, and AI risk experts.
We asked panelists to forecast (1) how a menu of policies would affect AI risk; (2) which policies they each support; and (3) which policies they think will be implemented by 2030.
Key findings:
• Despite disagreeing on the level of AI risk, respondents broadly agree on policy
• Respondents believe government policy can substantially reduce catastrophic AI risk
• The most supported policy of those we asked about was the creation of an international body, including the US and China as members, with pre-release authorization power over frontier models
• Some US-only policies are forecast to reduce catastrophic AI risk, but not as much as bilateral policies involving the US and China
• Experts strongly oppose federal preemption and think it would increase AI risk
• The policies experts expect to reduce risk the most are those they think are least likely to actually be implemented
🧵 More, with key insights below: -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Toby Shevlane
@tshevl
We're living through a two-sided revolution in the technology of prediction:
1. Prediction markets (@Polymarket, @Kalshi).
2. AI forecasters.
How do these compare? We benchmarked @_Mantic_AI against 4,000+ Polymarkets.
Results:
🔵 $100k volume is the crossover point where the AI loses its superiority over the market.
🔵 At >$1M volume, the optimal forecast still requires giving Mantic 14% weight.
🔵 Mantic is better calibrated.
This is for geopolitics, economics, elections etc -- we exclude sports and crypto.
Personal thoughts:
1. "At what market volume do you lose your Brier advantage over the market?" could be the new METR time horizons graph. 👀
2. AI wins on scale. We can apply the "$100k-volume-level accuracy" to a universe of millions of questions. 🚀 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sarah Fitzpatrick
@S_Fitzpatrick
I was told of a message relayed from a DOJ official to another: “I was told the Justice Manual isn’t a thing,” the official told me, and said they were told: “There are no rules. Just make sure Republicans win.” A DOJ spox told me “That is not, and has never been, Department guidance.” @TheAtlantic www.theatlantic.com/politics/2026/10/kash-patel-fbi-trump…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
In authoritarian regimes, underlings make obviously false statements to signal loyalty to the ruler
Aaron Rupar
@atrupar
JD Vance defends Trump's taxpayer-funded propaganda videos: "If you look at the message of those advertisements, they are not partisan, it's not like we're advocating for a particular candidate"
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Brian Stelter
@brianstelter
Scoop: The White House is refusing to allow Politico on Air Force One today, which means the news outlet cannot fulfill its press corps pool responsibilities. www.cnn.com/2026/10/05/media/politico-white-house-print-p…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Adam Steinbaugh
@adamsteinbaugh
The leader of the free world says newspapers should be arrested.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jacob Rubashkin
@JacobRubashkin
Former Chicago businessman Alphonse Capone, who rose to the top of the beverage distribution industry and was later jailed for violating tax laws, has died.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jonathan Reiner
@JReinerMD
Over the years I’ve responded to several medical emergencies on aircraft. Every plane has an AED which is great, but the medical kit is not. It needs an adult and pediatric epi pen (or new nasal epi), nasal naloxone, automatic BP cuff, pulse ox, definitely Zofran, IV catheters and fluid. No need for a stethoscope (too loud to hear). A 6 lead Kardia ECG would be helpful. Glucometer. Maybe Amio. SL NTG, aspirin. Glucagon auto injector. What else?
The Washington Post
@washingtonpost
Doctors have been pleading with the FAA to modernize the mandated contents of emergency medical kits, whose requirements haven’t changed in more than two decades.
But the FAA’s new proposal has left many in the medical community and some lawmakers stunned. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Preston Cooper
@PrestonCooper93
The villain in this story is the wealthy university charging $50k per year for a master of social work.
Inside Higher Ed
@insidehighered
The Students Crowdfunding for Tuition
Students facing acute challenges affording their tuition are increasingly turning to crowdfunding. Experts warn it’s a temporary fix to systemic affordability issues. www.insidehighered.com/news/students/financial-aid/2026/0… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Richard V. Reeves
@RichardvReeves
So @mattyglesias is right about the new college loan rules. It’s odd to see him being attacked from the left, given that the for-profit college sector exploits Black, Hispanic and poorer people the most as I pointed out back @BrookingsInst
open.substack.com/pub/matthewyglesias/p/trumps-new-studen… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A Claude infographic. We've been here before.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Forget the lump of labor fallacy. Time for the lump of ideas fallacy. We will create new arenas for human endeavor and new problems that can't be solved without human involvement.
Gregory Conti
@GregoryConti19
IMO there are no benefits of AI that can compensate for the loss of human intellectual greatness it is bringing about.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Forget the lump of labor fallacy. Time for the lump of ideas fallacy. We will create new arenas for human endeavor and new problems that can't be solved without human involvement.
Gregory Conti
@GregoryConti19
IMO there are no benefits of AI that can compensate for the loss of human intellectual greatness it is bringing about.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Always read @AndyMasley - on "the lump of cognition fallacy: the idea that there is a fixed amount of thinking to do" blog.andymasley.com/p/the-lump-of-cognition-fallacy
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
In the mail: The Fog of Accountability by the excellent @jdbk on how "urban politics is defined by minimally effective competition, complex institutions that confuse voters, and a lack of effective media coverage"
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Garrett Readling
@TheGurtling
Yeah we’re in great hands lmao
POLITICO
@politico
The low-profile Democrat helping shape her party’s AI agenda www.politico.com/news/2026/10/02/valerie-foushee-ai-deal-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
I think this advice is not as true anymore:
1) the durable skill behind “using AI better”
Is unclear as AI gets easier to use
2) Increasingly, organizational systems bottleneck AI gains not individual workers
3) The effect on jobs is uncertain but may impact entire job categories
Mark Cuban
@mcuban
AI won’t take your job. Someone who knows how to use AI better than you, will take your job.
Aaron Levie
@levie
We’re already starting to see what kind of new jobs AI is creating.
AI requires significant technical work and surrounding services to deploy into the economy. This means jobs for AI engineers that build applied AI products sold to (or within) enterprises, FDEs to deploy agents into companies, new services firms for deploying AI, and more.
And even the published stats of new jobs undercount all the existing jobs that are transitioning to new areas of AI work in an enterprise. Many prior data, research, and software jobs large enterprises are also being repositioned for working with AI.
Every bank, life sciences company, manufacturer, and even law firm is bringing on more technical talent -or repositioning existing roles- to help with agent deployment in their companies. It’s a lot easier to picture what AI can replace vs. what it creates until it starts happening. Now we’re seeing what this looks like.
The Kobeissi Letter
@KobeissiLetter
AI-related employment is surging:
AI-linked roles have accounted for more than +750,000 new jobs created in the US since 2023, according to LinkedIn estimates.
This surge has been led by data annotators, with +282,000 new positions created, followed by data center jobs at +117,000, and AI engineers at +105,000.
As a result, these 3 categories have accounted for +504,000 new jobs.
AI-related positions also offer significantly higher pay, with a median salary of ~$180,000 on LinkedIn, compared to $80,000 across all jobs.
This comes as the rapid expansion of AI infrastructure, particularly the buildout of data centers, is driving stronger demand for labor.
The AI boom is reshaping the US job market. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dr. Saskia Popescu
@SaskiaPopescu
There is one case of suspected plague. Not sure where “outbreak” is coming from. This situation is worrisome for many reasons but given the unknowns, let’s take a min until we get more info. This is also why we invest in biosurveillance, MCM, and biosecurity capacity building 💪🏻
The Washington Sun
@washingtonsun
The U.S. government is actively monitoring a suspected outbreak of the plague in Russia following the death of a 28-year-old lab worker who was reportedly infected after coming into contact with a broken test tube containing the pathogen. www.washingtonsun.com/us-news/u-s-government-monitor-susp…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
Ashish is always reasonable on biosafety. Secrecy + deliberate attempts to make bioweapons + inevitable mistakes can lead to a major problem even in pre-AI world! See Annie Jacobsen's "Biological War" for easy-to-read details.
Ashish K. Jha
@ashishkjha
Many of you have been tracking the Irkutsk lab situation over the last few days.
I've been gathering facts and trying to understand what is happening before weighing in.
Now that I have, I have to say — I'm concerned
🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Charles Fain Lehman
@CharlesFLehman
This is an important point about AI writing — it consistently signals that you aren't trying, just acting as a meat puppet.
Conversely, it means that the premium to actually using your brain in communicating with other humans will go up!
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Kids: Don't do this.
Ran Blekhman
@blekhman
My inbox is being flooded with fully AI-generated emails from prospective students. I tried to work out why I can't bring myself to read them, and ended up with some thoughts and advice for writing cold emails in the age of AI.
My new post:
blekhman.substack.com/p/why-i-am-not-reading-your-ai-gene… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ashish K. Jha
@ashishkjha
Many of you have been tracking the Irkutsk lab situation over the last few days.
I've been gathering facts and trying to understand what is happening before weighing in.
Now that I have, I have to say — I'm concerned
🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sam Vecenie
@Sam_Vecenie
We're not even 24 hours into the NBA preseason and Nikola Jokic has broken my brain. I just don't even know how one does this.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
scary lawyerguy
@scarylawyerguy
In 2015, his reality show was on the verge of cancellation, he was four years into an IRS audit that could have cost him $100 million, and his "brand" was in the toilet but he co opted the Republican Party and is now a multi-billionaire who is effectively above the law.
Aaron Rupar
@atrupar
Trump in Ohio: "I had a nice simple life before this. I didn't have to be here on Saturday night going like this, working my ass off in front of your people. I could've been home. I could've been in Monte Carlo."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Michael Paarlberg
@MPaarlberg
Important point for anyone using AI tools to apply for jobs or fellowships: The purpose of an application is to stand out from other applicants. The purpose of AI is to reflect the broadest data pool; to be generic. You’re not being unique or memorable. You’re being the opposite.
Ran Blekhman
@blekhman
My inbox is being flooded with fully AI-generated emails from prospective students. I tried to work out why I can't bring myself to read them, and ended up with some thoughts and advice for writing cold emails in the age of AI.
My new post:
blekhman.substack.com/p/why-i-am-not-reading-your-ai-gene… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jan Zilinsky
@janzilinsky
Something that’s been in the back of my mind: are some issues considered to be both important and boring by voters?
Maybe so: tweets with terms like “economic” and “businesses” get the least engagement:
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"tweets referencing divisive issues ... receive substantially greater engagement than tweets about the issues voters care most about" but "candidates predominantly discuss voter-relevant issues"
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Kids: Don't do this.
Ran Blekhman
@blekhman
My inbox is being flooded with fully AI-generated emails from prospective students. I tried to work out why I can't bring myself to read them, and ended up with some thoughts and advice for writing cold emails in the age of AI.
My new post:
blekhman.substack.com/p/why-i-am-not-reading-your-ai-gene… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Simon Kuestenmacher
@simongerman600
How the American life changed from 1995 to 2026 is wild. More concerts, more exercise, more video games. Fewer books, less church, not staying at work longer (I guess you leave on time and continue work at home?). Source: www.derekthompson.org/p/the-death-of-the-american-host
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Unprecedented corruption
Peter Baker
@peterbakernyt
In 2019, Trump was forced to cancel plans to hold the G7 at his own golf club amid GOP objections to self dealing. In 2026, he will hold the G20 at his own golf club and the critics have gone silent. How Trump has moved the ethical line in his second term. www.nytimes.com/2026/10/04/us/politics/trump-moving-ethic…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Unprecedented corruption
Nicholas Nehamas
@NickNehamas
Hours before Krist Noem left her job as D.H.S. secretary, her agency bought 5 planes from a donor at a big markup, records show. They were supposed to fly deportees but have largely sat idle -- a striking symbol of gov. waste under President Trump. www.nytimes.com/2026/10/04/us/politics/kristi-noem-planes…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Justin Amash
@justinamash
Donald Trump is the authoritarian socialist president that Republicans used to warn about.
Donald J. Trump
@realDonaldTrump
If Republicans win the House of Representatives and the Senate in the 2026 Midterm Elections, I’m going to give all adult citizens in the United States of America, $5,000! Thank you for your attention to this matter, and I look forward to signing those checks!
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Nunn
@AlexNunn
This is pretty wild. Basically, during sentencing, an AI video was shown that depicts the deceased victim sharing hypothetical posthumous comments (see below).
After the AI video played, the judge was visibly emotional, stating, "I love that AI, thank you." He then sentenced the defendant to a longer prison term than what the prosecution even asked for.
Colin Miller
@EvidenceProf
Arizona Court of Appeals reverses defendant's sentence for his manslaughter conviction, finding that AI victim impact statement was "a depiction of the victim and his thoughts created from the imaginings of the victim’s sister." coa1.azcourts.gov/Portals/1/OpinionFiles/Div1/2026/State …
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Peter Baker
@peterbakernyt
Trump, who has long denounced mail-in voting and sought for months to restrict it because, he claims, it is "inherently corrupt," urges Republicans to vote by mail. @TimBalk @EvanGorelick www.nytimes.com/2026/10/03/us/politics/trump-republicans-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Kids: Study statistics so you can learn about placebo tests
Adam Stein
@Dr_A_Stein
A recent series of papers suggested that nuclear power plants are responsible for more than 100 thousand cancers. Using their method, every landmark we tested causes cancer,
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
American Political Science Review
@apsrjournal
From our FirstView: "The Representation Trap: How Minority Power Shapes Electoral Politics in India" by @FeyaadAllie. www.cambridge.org/core/journals/american-political-scienc…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Gandalv
@Microinteracti1
Robots fighting each other on Ukraine's frontline, and the West has no idea how far behind it has fallen. This is changing warfare forever, and yet we keep ordering tanks and artillery like drunks at closing time who insist on one more round of whatever they were drinking in 1985.
WarMonitor🇺🇦🇬🇧
@WarMonitor3
Incredible video shows a Russian drone attempting to destroy a Ukrainian ground robot but being disabled when the robot deploys an anti drone net and catches it.
We literally have robots fighting each other on Ukraine’s frontline. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nico Ajzenman
@Nicolas_Ajz
Que bien comunica FT.
El efecto de rent control en el mundo. Gran beneficio para los inquilinos actuales. Gran daño para todo lo demás. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Unprecedented corruption
Sam Stein
@samstein
One day after @jaredkushner took to this platform to insist his work diplomatically is unconnected to his business interests, the Times drops this one
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Adam Kinzinger (Slava Ukraini) 🇺🇸🇺🇦
@AdamKinzinger
Americans: you are not outraged enough about this.
Christians: this is absolute blasphemy
www.cnn.com/2026/10/02/politics/trump-taxpayer-funded-ads -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Egger
@EggerDC
The hubris of not even waiting until after the midterms to plan this of really astonishing
Jeff Stein
@jstein_sun
NEWS: Trump to host "most EXCLUSIVE DINNER IN THE WORLD" for top 185 purchasers of his meme crypto
Buying $1M (or less!) worth of $TRUMP coin should get you in Nov 22
Trump has pocketed $635 million
Investors down +$4 billion on Trump crypto
Invite ⬇️
www.washingtonsun.com/whitehouse/trump-meme-coin-dinner-c… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jon Favreau
@jonfavs
Why are companies selling fake human beings
Why does the computer software we interact with have to be indistinguishable from a living, breathing person
Why is this an achievement
Tavus
@tavus
Introducing Griffin, the first model to pass the video Turing test.
48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA's benchmark for full-duplex AI video.
It’s the first Human Interaction Model (HIM).This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Eitan Fischberger
@EFischberger
Tucker saying he wishes Hitler had won World War II was absolutely on my 2026 bingo card
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Yglesias
@mattyglesias
I wonder if "Britain would be better off if Hitler had won World War II" is one of the political issues that Vance disagrees with his good friend Tucker Carlson about.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Megan Fritts
@freganmitts
I never take any pleasure in boosting Matt Walsh but he is 100% correct here. The anthropomorphic form is a result of intentional decisions by the human designers, there is no good reason for it, and it is nefarious. People should be talking about that more.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Annie Jacobsen
@AnnieJacobsen
🚨Breaking. There's been a pneumonic plague outbreak at a BSL-3 lab in Siberia, Russia.🚨
"Nearly 200 People Under Observation After Irkutsk Lab Worker Dies From Plague" - Moscow Times
This is fact, and also the inciting incident in my book BIOLOGICAL WAR: A SCENARIO, which is unnerving.
Will post more as I learn more, including from Russian speaking sources🧵
shorturl.at/De4kW -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Zeke Hausfather
@hausfath
Actual climate scientist here. The permitting reform bill was introduced by Senate climate champions like @SenWhitehouse. It will reduce US emissions.
We don't have the luxury to make the perfect the enemy of the good – which time and time again has led to doing nothing at all.
Dr. Genevieve Guenther (she / they)
@DoctorVive
The problem with invective-spewing Abundance bros like Johnson is that they dismiss the science of climate change — which makes assessing trade-offs in policymaking so much easier, of course.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jiayi Geng
@JiayiiGeng
AI scientist agents are great at optimizing and finding solutions through endless trial and error. But is that really what science is about?
As Terrance Tao eloquently put, the role of math and science should be more than that – they are "lighthouses" that guide and inspire exploration through understanding and insights. Can AI agents discover novel insights through experimentation?
To study these gaps, we worked with domain experts to introduce EurekaBench, a benchmark spanning 6 science domains that evaluates an agent’s ability to discover genuinely new scientific insights.
(1/n) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"President Trump personally instructed his budget director to use taxpayer money for TV ads praising him and his presidency."
Either we are a nation of laws or we are not.
Maggie Haberman
@maggieNYT
NEW: President Trump directly told Russ Vought to find the funds to finance the ads glorifying him. The agency’s general counsel and WH counsel both signed off on the move as legal. The president and WH chief of staff spoke to the DHS secretary. www.nytimes.com/2026/10/02/us/politics/trump-ads.html
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
What could go wrong?
Matt Bors
@MattBors
I think being able to video chat with a cute woman who never gets tired of you is going to be a huge disaster for our already loser-based society
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sylvain Catherine
@sc_cath
I am very quickly being convinced that we need to ban AI from pretending to be human. I don’t see the upside of letting AI do it, I think there is massive downside, and I doubt markets can spontaneously solve that problem.
Tavus
@tavus
Introducing Griffin, the first model to pass the video Turing test.
48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA's benchmark for full-duplex AI video.
It’s the first Human Interaction Model (HIM).This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Like pretending covfefe was intentional, but with matters of life and death
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
AI writing occupies such a narrow niche in the distribution of human language.
If you start with human writing, you will (currently) say things an AI never would.
Those unique phrases survive translation. Here's a Piketty op-ed and AI translation that Pangram calls 100% human. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Interesting—Pangram (intentionally) does not flag AI-translated texts as AI-generated.
This is great for non-native English speakers, and makes sense once you think about it.
Here are Pangram scores on pre-2022 French/Spanish writing, translated with a series of AI prompts. 1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
A general principle: when you interact with AIs, you should know that you are interacting with AIs.
Tech like this that helps disguise AI makes us all worse off.
Working on camouflaging AI use should be viewed as unethical. This doesn't help anyone.
Hassaan Raza
@hassaanraza
The whole point of building Tavus has been simple: talking to a machine should feel as natural as talking to a friend or coworker.
It’s hard to describe all the tiny nuances that make a conversation feel human. The little expressions. Moving around in your chair. Knowing when to speak and when to listen. The dance of it all.
Griffin is by far the closest anyone has come to a model that can capture those nuances. The first time I saw it being used, I had no idea I was watching our model rather than just a normal video call.
I’m so incredibly proud of this team and what they’ve built.
Tavus
@tavus
Introducing Griffin, the first model to pass the video Turing test.
48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA's benchmark for full-duplex AI video.
It’s the first Human Interaction Model (HIM).This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
A lot of humans will die, but it's worth it to save some imaginary cats.
Guy Still
@mplstvguy
Just now #minneapolis City Council member Jason Chavez says he supports the ordinance mandating "human safety monitors" in autonomous rideshare vehicles because cats are "being murdered by these vehicles."
A vote on the measure is expected soon... -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jesse D. Jenkins
@JesseJenkins
This is one of several startups using robotics and standardized designs to vastly accelerate installation of utility-scale solar. Modules are so cheap, they're only a fraction of solar capex now. Attacking installation costs is a huge lever. Innovation to make clean energy cheaper hasn't stopped.
puneet
@puneetp
Gritt AI is now installing 1 MW of solar capacity per day at 4× the speed of a typical installation crew. And our daily installation rate keeps accelerating.
At our active site, each day’s installation adds enough solar capacity to meet roughly 200 homes’ annual electricity needs once operational.
That’s the milestone: every day, more infrastructure built. Every increase in speed, more capacity to meet America’s growing energy demand.
Solar is just the beginning. Next, we aim to bring Gritt’s construction superintelligence to data centers, water treatment plants, and pipelines.
Our mission: enable humanity to build centuries’ worth of infrastructure in decades.
Civilization. Accelerated.
@Jason @andrewbeebe @albertwenger @hayleybay @CleanVC @rebeccakaden @Rajil @GrittRoboticsThis media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
ICYMI
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Precisely the wrong approach. We don't actually have to give up on papers! New in @chronicle
How to Assign Research Papers Again
Three ways to counter AI and restore confidence in the integrity of student writing (link below)
Steve McGuire
@sfmcguire79
The head of the philosophy department at Williams College downplays the pedagogical value of writing:
“What if we are mistaken in treating writing as the highest cultivation of thought? The evidence that supports this belief is thin at best. Writing can be useful cognitive scaffolding for memory and self-regulated learning, but the notion that it is the best process for developing thinking skills is a fairly recent idea, traceable to the 1970s writing-across-the-curriculum movement.” -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Unprecedented corruption. (Also, look at this terrible AI.)
Republicans against Trump
@RpsAgainstTrump
Trump to host “most exclusive dinner in the world” for the top 185 holders of his $TRUMP meme coin on November 22. The top 29 purchasers will be invited to an additional “VIP reception with your favorite president.”
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
John Horton
@johnjhorton
A new study from @pewresearch in which they created rich digital twins & then compared responses to actual survey-takers. Large average error. But interestingly, when I just asked frontier models for the question response distribution, the result is *far* more accurate than the twins (at 1/1000th the cost). 1/
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sam Asher
@thesamasher
Big new @devdatalab data release on cities around the the world! Please do check out the site and provide feedback. Much more data and analysis coming in the months ahead. We're all in on cities at DDL and hope this effort helps unlock a boom in urban dev research and policy!
Paul Novosad
@paulnovosad
New @devdatalab data just dropped for the 10,000 biggest cities in the world.
Air pollution, traffic and congestion, Doing Business indicators.
We're adding new data every few months.
Let's make it easier to study what works and what doesn't in LMIC cities. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Thomas G. Dietterich
@tdietterich
Attention arXiv authors. Because of the exponential increase in submissions, arXiv is limiting the number of papers that a person can submit each month to two. 1/
arXiv.org
@arxiv
arXiv has updated our policy on rate limiting for all submitters.
This update was made to fairly distribute moderator time & support the arXiv community of staff, volunteers, readers & authors.
Please read our announcement to learn more: blog.arxiv.org/2026/10/01/updated-rate-limit-policy/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Scenes from a gerontocracy - declining octogenarian president brags he is slightly better at walking than previous declining octogenarian president
Aaron Rupar
@atrupar
"I can go ramp, I can go stairs" -- Trump's audience in Texas gives him an extremely tepid round of applause as he brags about being able to go up stairs without falling
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The man with the nuclear codes.
Aaron Rupar
@atrupar
"I can go ramp, I can go stairs" -- Trump's audience in Texas gives him an extremely tepid round of applause as he brags about being able to go up stairs without falling
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Michiel Bakker
@bakkermichiel
🚨📄 New preprint! We find the “boiling the frog” equivalent of AI use. In a series of RCTs, we show that after just 10 min of AI assistance people perform worse and give up more often than those who never used AI.
w Grace Liu @brianchristian Mira Dumbalska and Rachit Dubey 🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
“We find that on medium-length, well-defined accounting tasks, frontier AI models are now faster and more accurate than junior accountants, even the best one in our study.” Eighteen months ago they scored well below human accountants
Good discussion here: www.mercor.com/blog/human-baselines-for-benchmarks-ai-now… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
aden barton
@aden_barton
New research: we hired 12 CPAs to complete four realistic accounting tasks, then had AI models from the past few years complete the same work.
As of May 2025, accountants outperformed AI. Now they lose out to today's frontier models, which are near perfect on these tasks. 🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"tweets referencing divisive issues ... receive substantially greater engagement than tweets about the issues voters care most about" but "candidates predominantly discuss voter-relevant issues"
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrés Neumeyer
@andyneumeyer
A serious paper on the effect of taxing top wealth was published in the last issue of the AER.
www.aeaweb.org/articles?id=10.1257/aer.20241596
Taxing Top Wealth: Migration Responses and Their Aggregate Economic Implications
Katrine Jakobsen, Henrik Kleven, Jonas Kolsrud, Camille Landais, and Mathilde Munoz
Using administrative data from Scandinavian countries, we provide evidence on international migration responses to wealth taxes and evaluate their aggregate economic implications. We find significant migration responses among the wealthy: A 1 percentage point increase in the top wealth tax rate decreases the stock of wealthy taxpayers by about 2 percent. A large fraction of the wealthy are business owners, and their businesses are negatively affected by owner out-migration. The aggregate effects are nevertheless modest: The migration responses to a 1 percentage point increase in the top wealth tax rate reduce employment by 0.02 percent, investments by 0.07 percent, and value added by 0.10 percent. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Not a causal estimate, obviously, but lots of reason for worry
jeffrey lee funk
@jeffreyleefunk
AI users who reported using Google's AI tools for specific schoolwork tasks like drafting texts for writing assignments had science scores an average of 20 points lower than students who said they never used AI—equivalent to about a year of schooling. www.wsj.com/tech/ai/google-ai-gemini-education-schools-1e…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
jeffrey lee funk
@jeffreyleefunk
AI users who reported using Google's AI tools for specific schoolwork tasks like drafting texts for writing assignments had science scores an average of 20 points lower than students who said they never used AI—equivalent to about a year of schooling. www.wsj.com/tech/ai/google-ai-gemini-education-schools-1e…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Derek Thompson
@DKThomp
The fundamental tension w AI and education is that I think schools ought to (a) teach people how to think and (b) prepare them for the workforce (or, at the very least, not actively harm their employability).
Those goals push in opposite directions.
AI interferes with thinking but it’s also baked into the work of software, finance, legal, marketing, etc. If you use too much AI in college, you’ll be dumb; never use it for four years, and you’ll be in a disadvantaged position for the most popular high-paying entry level jobs.
Something like this below seems like at least a thoughtful way of addressing that tension.
Steve McGuire
@sfmcguire79
🧵UChicago is leading the way on AI in higher education.
1. All students will take an analog writing class.
2. Core classes in the humanities and social sciences will prohibit technology in class.
3. A new requirement in computational thinking will likely be added to the core. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Google Docs is SO deeply integrated into American K-12 and higher education at this point - Gemini may have already done a great deal of harm.
Shalini Ramachandran
@shalini
My latest investigation: Inside Google's AI Push Into American Schools "It Slowed My Thinking Down." Students are becoming academically and emotionally dependent on AI, in large part thanks to Google's push in schools. Gift link: www.wsj.com/tech/ai/google-ai-gemini-education-schools-1e…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nico Perrino
@NicoPerrino
I'm worried Americans are becoming numb to this.
A few years ago, Tim Walz said, “There’s no guarantee to free speech on misinformation or hate speech.” When the remark resurfaced, it rightly drew days — weeks! — of criticism and outrage because it was flatly wrong.
Now, it seems like every day someone in the Trump administration says something similarly at odds with free speech principles, and in many quarters it’s simply accepted that the administration will say these things.
Pam Bondi said the DOJ would “absolutely target” and “go after” people for hate speech.
Todd Blanche said "misinformation" and "one-sided" news can be a "national security issue" that the American people shouldn't have to put up with.
Trump said there’s something wrong with a country that “can allow people to write purposely negative stories.”
And now Defense Secretary Pete Hegseth says “partisan activism” that “smears the troops” is “treason” — whatever that means.
Treason is the only crime the Constitution itself defines, and criticism of the military isn’t it!
Aaron Rupar
@atrupar
Hegseth: "A free press is of course important, but partisan activism that intentionally smears the troops and the historic mission that they are undertaking is fundamentally anti-American. It's sick. It's a national betrayal. It's treason."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Patrick Daugherty
@RotoPat
This passage is dangerously close to "era defining" lol
Jeff Stein
@jstein_sun
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
An all-timer
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Steve McGuire
@sfmcguire79
More bad news for higher education:
The percentage of Americans who say going to college is very important has hit 31%, another record low.
The percentage has declined across various demographic groups. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Kind of a banger TBH
Severin Field 🦅
@sevdeawesome
Claude 5.5 Opus responding to “AI is a normal technology” is still the best AI-created video I have ever seen
This media couldn't be displayed. See it on the original post.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Also a banger and much more convincing
Sayash Kapoor
@sayashk
I had way too much fun using Claude to create a reply video.
- Extremely fun to be able to convert scattered thoughts and ideas into a song + video
- Creating this took <1 hour of my time (in occasionally steering the model in between other things)
- But it took ~10 hours of wall clock time for the agent to generate the video + iteratively refine it based on feedback
- In terms of AI usage: it required only a small number of Suno credits and 10% weekly usage of a Claude Max subscription (I was using Opus 5.5)This media couldn't be displayed. See it on the original post.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
cc @paulnovosad
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
New @devdatalab data just dropped for the 10,000 biggest cities in the world.
Air pollution, traffic and congestion, Doing Business indicators.
We're adding new data every few months.
Let's make it easier to study what works and what doesn't in LMIC cities. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Its even better— the thesis of the Op-Ed is that AI’s can’t do philosophy
How does this keep happening
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Folks, what are we doing? 76% of a NYT op-ed by a *philosophy professor* is AI-written. (I just replicated this in Pangram.) There are many contexts where AI writing is perfectly acceptable. This is not one.
Jesse Spafford
@jessespafford
Since I've been complaining about semi-coherent articles about AI, I was finding this Simon Critchley piece in the @nytimes to be a pretty painful read, and then I suddenly realized why: www.nytimes.com/2026/09/27/opinion/ai-philosophy-thinking…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
The context is wild. Trump rambled for AGES about the White House renovation (I counted, 802 words total). The interviewer then changes topic, but Trump tries to bring it back to the White House. The Time reporter doesn't engage and moves on McConnell, and Trump completely SNAPS.
Aaron Blake
@AaronBlake
Trump snaps at @TIME's reporters:
"You haven't asked me one fucking positive question."
(He then claims they never asked him about Venezuela, which they did.) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
1. Using AI to correct grammatical errors / idioms
2. Using AI to write entire sections of text
You don't have to hold the same opinion on these things!
There is some motte and bailey going on, where people are making good cases for 1, and then pretending 2 is the same thing. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Folks, what are we doing? 76% of a NYT op-ed by a *philosophy professor* is AI-written. (I just replicated this in Pangram.) There are many contexts where AI writing is perfectly acceptable. This is not one.
Jesse Spafford
@jessespafford
Since I've been complaining about semi-coherent articles about AI, I was finding this Simon Critchley piece in the @nytimes to be a pretty painful read, and then I suddenly realized why: www.nytimes.com/2026/09/27/opinion/ai-philosophy-thinking…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Daniel Dale
@ddale8
This remains one of the president's most ludicrous lies. He lost California by 30 points in 2016 (more than 4 million votes), 29 points in 2020 (more than 5 million votes), and 20 points in 2024 (more than 3 million votes).
Acyn
@Acyn
Trump: If we had an honest election, I would have won California. I got millions of votes, but I would have won California if that were an honest system. I would have won
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Carter Church
@CarterWChurch
I used GPT-6 Astra to break an unsolved cipher to one of Napoleon's generals that had gone unread for 217 years.
What makes this impressive isn't actually the codebreaking, but that Astra completed the entire multi-modal workflow in ~6 hours from a single image and goal.
1/6🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Adam Steinbaugh
@adamsteinbaugh
Treason is the one -- and only -- crime specifically defined in the Constitution. It is not saying negative things about "the troops."
Aaron Rupar
@atrupar
Hegseth: "A free press is of course important, but partisan activism that intentionally smears the troops and the historic mission that they are undertaking is fundamentally anti-American. It's sick. It's a national betrayal. It's treason."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dave DesRoches
@DBDesRoches
This is not not the job of the civilian leadership. It’s cosplaying as a sergeant major.
DOW Rapid Response
@DOWResponse
Six hundred of our finest warriors joined @SECWAR for morning PT at the Pentagon ahead of his HISTORIC State of the Force address.
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matt Burnell
@mdburnell
Let's go to one of those minds for their thoughts on solving the Strait of Hormuz problem, Newt take it away:
Aaron Rupar
@atrupar
Hegseth: "Project Meridian will be co-led by three of our nation's best minds: Elon Musk, Palmer Luckey, and Newt Gingrich"
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
The VP is being incredibly dishonest.
Undocumented immigrant are not eligible for Social Security. Period. In fact, they pay billions of dollars into Social Security in their paychecks that they will never get back. They are a net fiscal benefit to the program! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Isabel Laterzo-Tingley
@IsabelLaterzo
Our new paper in @BJPolS (open access) examines how childhood exposure to criminal violence and militarized conflict shapes attitudes into adulthood.
Studying MX's drug war, we demonstrate meaningful, durable effects on trust and vote choice. Read more below ⬇️
Omar García-Ponce
@o_garcia_ponce
1/8 🧵 New paper with @IsabelLaterzo, now out open access in the British Journal of Political Science: The Legacy of Mexico’s Drug War on Youth Political Attitudes
How does growing up amid criminal violence and militarized security shape political attitudes later in life? -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jay Nordlinger
@jaynordlinger
If a president from the other party did this, every Republican would be calling for his impeachment -- justly.
Kyle Griffin
@kylegriffin1
WSJ Exclusive:
The Trump admin used federal funds from the Department of Homeland Security to finance those controversial pro-Trump ads that started airing on TV.
The taxpayer money used to fund the ads came from U.S. Customs and Border Protection. www.wsj.com/politics/policy/homeland-security-money-finan… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ben Krauss
@BenKrauss5
If you care about decarbonizing the economy or reshoring American manufacturing, or cutting bills for ratepayers, then you should support this bill. No agreement is perfect, but there’s something in this permitting reform agreement that will, on net, benefit everybody.
@JaneAFlegal and @ppavnr make a thorough case for permitting reform.
open.substack.com/pub/searchlightinst/p/pass-permitting-r… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Argh, don't give up on writing - you don't have to! www.chronicle.com/article/how-to-assign-research-papers-a…
Steve McGuire
@sfmcguire79
Another professor suggests a return to Socrates in the age of AI:
“Implementing AI-detection software, blue books and oral exams” addresses “only symptoms of the transformation of the university we are seeing in real time. The challenge is to identify that transformation and adapt accordingly.”
“Faculty need to adapt to the technological transformation. That means learning new tools, developing new pedagogical techniques and becoming AI-literate themselves. They have to redesign how they assess students to ensure understanding.” -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Some settings where it's reasonable to expect AI voices: product manuals, table/figure exhibits, AI article summaries.
Some settings where it's reasonable to expect human voices: Op-eds, creative writing, personal messages. Without disclosure, AI use here is deceptive.
4/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
We need to distinguish between "AI grammar checking" and "AI rewriting."
The first preserves your voice and your writing. The second replaces your voice and writing with AI-voice and AI-writing. (The ideas might still be yours)
Interestingly, Pangram distinguishes these. 1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Astor
@AstorAaron
Important point here about AI rewriting v AI-assisted grammar checking. Some kind of “grammar check” is embedded in most writing software. If the student makes suggested changes, that’s fine. It’s like having a friend read it. But AI rewriting? Very different and no good.
Paul Novosad
@paulnovosad
We need to distinguish between "AI grammar checking" and "AI rewriting."
The first preserves your voice and your writing. The second replaces your voice and writing with AI-voice and AI-writing. (The ideas might still be yours)
Interestingly, Pangram distinguishes these. 1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
It's easy to implement this, you can do it right in Canvas.
The whole article by @BrendanNyhan and Jeff Friedman is worth reading.
/N
www.chronicle.com/article/how-to-assign-research-papers-a… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Other critiques are addressed. Lots of government profs are doing this here, it is working!
3/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Oral or written exams about paper content can be used in tandem.
Students like these formats!
Most students actually want to work and learn, but not in an environment where everyone else is cheating and then getting better grades than them. 2/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
If you think students can benefit from writing research papers, read this article! You can still assign papers!
Dartmouth profs are doing this very successfully. One approach is having students use Google Docs, where the version history proves human writing.
It works! 1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Come ON
Kashmir Hill
@kashhill
And this is almost an aside in the story, but also absolutely bonkers. Meta claimed Zuck as a "researcher" to get a $355 million tax break on his $4.1 billion stock-based compensation.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
If every single one of the 1,600 suspected cases of noncitizen voting were confirmed (which is unlikely given previous Mullin claims) to have voted in the 2024 election, that would represent just 0.001% of total votes, including votes for Republicans and Democrats alike.
Homeland Security
@DHSgov
“Every voter that votes should know that their vote counts. It shouldn’t be canceled by an illegal or someone illegally voting in those elections. And that’s exactly what happens. The only people that should be choosing our leaders are American citizens who are legally able to vote.” @SecMullinDHS
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jesse D. Jenkins
@JesseJenkins
We cannot decarbonize if we cannot build at a scale and speed not seen in decades. We cannot harness clean energy to power the data center boom if we cannot build more, fast. No permitting reform deal is going to ONLY smooth the way for your preferred resources. But the sclerotic system of environmental permitting and grid planning today is unquestionably a barrier. We have to clear these barriers way, and then continue the fight to build the clean infrastructure necessary to decarbonize America. Take a hard look at this permitting deal. It's (long past) time to get something done.
Jane Flegal
@JaneAFlegal
15/You really have to contort yourself to believe tech-neutral reforms that make it easier to build big stuff, especially on federal land, won't help clean energy more than anything else. But I'm sensing this is the debate some groups want to have.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
John Arnold
@johnarnold
Huge kudos to the 4 senators and the White House for completing negotiations on energy permitting reform.
The bill has been so difficult because the 4 senators came with very different priorities and constituencies.
Capito (R-WV): pipelines and permitting certainty
Whitehouse (D-RI): reducing emissions
Lee (R-UT): removing regulatory barriers and protecting states' rights
Heinrich (D-NM): protecting lands; transmission and renewables
They came to the table because the current system fails on all four sets of priorities. But permitting reform is not zero sum. This deal advances all of them. It's good for America and the American people. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Benjamin
@bschne
Werner Herzog will be the last human being whose next word no language model can predict
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Come ON
Eva Roytburg
@eva_roytburg
Meta took a $355 million tax break on Mark Zuckerberg’s $4 billion compensation by classifying him as a researcher, per the NYT. Insane
www.nytimes.com/2026/09/30/technology/meta-ai-data-center… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Arvind Narayanan
@random_walker
Because so much of the debate about AI and jobs is driven by fears of outright replacement, there's a much more likely way in which AI could make work worse that isn't getting much attention. Here's my case:
1) A growing share of knowledge workers are managers of AI agents rather than doing the work themselves.
2) Being a manager (of humans) actually kinda sucks. You don't get to practice the craft you trained for; being responsible for others' mistakes is super stressful; and your attention gets split between dozens of things.
3) But we managers don't get to complain — it's higher pay and higher status, and mentoring people and watching them succeed is incredibly gratifying. We chose to get promoted.
4) Being a manager of AI brings all of the negatives without any of the positives. When everyone is a manager, it's not high pay / high status. And the role change is forced upon you.
What am I missing? -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aidan Mackenzie
@AidanRMackenzie
🚨 WE HAVE A PERMITTING DEAL🚨
And it is easily the best permitting bill we've ever gotten!
- The Bipartisan American Affordability and Jobs Act is HUGE (400+ pages) and both sides got a LOT.
- The bill is much bigger and better than either Manchin post-IRA in 2022 or Manchin-Barrasso in 2024.
Once I finish reading the text I'll publish an explainer. For now, here are a few high level takeaways.
1. Transmission: The bill is absolutely HUGE for transmission, going further than Manchin-Barrasso. Includes: interregional transmission, limited federal backstop authority, interconnection queue reforms, directives for advanced grid-enhancing tech, and oversight of utilities. It's absolutely AMAZING all of this got in.
2. NEPA: The bill has BIG changes to NEPA, including fixes for judicial review that underpins the law's costs. Still reading but this is even better than I expected!
3. Certainty: The bill includes strong provisions to push back on misuse of permitting, like we're seeing with the admin's abuses against offshore wind.
4. Historic Preservation: The bill includes changes to the NHPA 106 process, including changes to judicial review that will help right-size the process.
5. Clean Water: The bill fixes problems with section 401 by adding new requirements for states and makes section 404 nationwide permits easier to comply with.
6. Geothermal: The bill includes big wins for geothermal, including parity with oil and gas catexs.
7. Mining: The bill fixes the "Rosemont decision"
8. Endangered Species: The bill creates a new path for states to take on endangered species consultation.
9. Hydro: Fixes for certain hydroelectric projects
10. Ratepayer protection: The bill adds protections for ratepayers from added data centers. Looks to be stronger than the House passed bill by quite a bit. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matt Grossmann
@mattgrossmann
DOGE contract cuts were concentrated in counties that gave Trump less support in 2024
www.cambridge.org/core/journals/american-political-scienc… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Daniel Dale
@ddale8
Yesterday, asked about its chatbot's many debunkings of Trump, the White House said it "will continue to be updated to provide more features, resources, and the most accurate information from the federal government."
Today...lots of accurate answers have been simply forbidden, with the chatbot framing information like who won the 2020 election as forbidden "political commentary."
Daniel Dale
@ddale8
Yesterday: Trump administration chatbot says it's not true that Trump inherited the worst inflation in history.
Today: Says it can't answer such questions.
It's the same for numerous other topics. Notably, its accurate answers yesterday relied on official federal information. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Laia Balcells
@laiabalcells
"Consequential histories: A field experiment at the US Holocaust Memorial Museum" Just accepted at the Journal of Politics www.journals.uchicago.edu/doi/10.1086/744218
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ryan Cummings
@weakinstrument
Mandatory
Arnab Datta
@ArnabDatta321
Where’s this with curb music
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matt Fuller
@MEPFuller
The Atlanta Hawks have an opportunity to do the funniest thing imaginable.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dean W. Ball
@deanwball
It’s cool how the idea of third-party embedded evaluators as a pillar of frontier AI policy went from not really on any Americans’ radars 18 months ago to ~consensus. I remember when the idea was much closer to the fringes. Not that long ago, even some of the orgs we now look to as potential evaluators questioned why they’d want to be deputized in that way when I pitched them on it.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
emil
@emilahlback
In our internal knowledge work evals, GPT 6.1 Sol is the best model we've tested.
It completed more of the work independently than Claude Opus 5.5, at 40% of the cost and nearly twice the speed, and led in every industry we measured. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
They say this every few months - it's like Tom Friedman on Iraq (the next six months are crucial)
Geoffrey Irving
@geoffreyirving
I was at an event with a frontier AI lab executive recently where they said (paraphrased)
1. In a few months, the AIs will be better than humans at all domains of science.
2. But now is the wrong time to slow down, maybe in a couple months. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Kids: study base rates
Nate Silver
@NateSilver538
This is maybe our most important chart of the election cycle.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
I continue to not understand how open weight frontier models are supposed to be safe. Perhaps short-run the answer is that big OS hosts will not serve the abliterated models? There *will* be a major safety issue and/or very costly use soon. It's inevitable.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Steve Greene
@HankGreene
Every professor who assigns writing needs to read this.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Precisely the wrong approach. We don't actually have to give up on papers! New in @chronicle
How to Assign Research Papers Again
Three ways to counter AI and restore confidence in the integrity of student writing (link below)
Steve McGuire
@sfmcguire79
The head of the philosophy department at Williams College downplays the pedagogical value of writing:
“What if we are mistaken in treating writing as the highest cultivation of thought? The evidence that supports this belief is thin at best. Writing can be useful cognitive scaffolding for memory and self-regulated learning, but the notion that it is the best process for developing thinking skills is a fairly recent idea, traceable to the 1970s writing-across-the-curriculum movement.” -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
We still do writing assignments at Dartmouth, and most students do not cheat.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Precisely the wrong approach. We don't actually have to give up on papers! New in @chronicle
How to Assign Research Papers Again
Three ways to counter AI and restore confidence in the integrity of student writing (link below)
Steve McGuire
@sfmcguire79
The head of the philosophy department at Williams College downplays the pedagogical value of writing:
“What if we are mistaken in treating writing as the highest cultivation of thought? The evidence that supports this belief is thin at best. Writing can be useful cognitive scaffolding for memory and self-regulated learning, but the notion that it is the best process for developing thinking skills is a fairly recent idea, traceable to the 1970s writing-across-the-curriculum movement.” -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Ken Dilanian
@KDilanianMSNOW
As MS Now has reported, FBI officials say the bureau has been placing people on the Most Wanted list whose capture is believed to be imminent. In other words, these officials tell us—they are gaming the system. One fugitive was captured in Mexico an hour and 13 minutes after being placed on the list.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
ArtButMakeItNow
@artbutnow
Circus Performers, by Walt Kuhn, 1978, 📸 via @marisakabas2
Sahil Kapur
@sahilkapur
A phenomenal moment in the Senate.
Eric Schmitt, R-Mo., claims he has evidence Jack Smith went to a HAWKS game in Atlanta.
Smith denies it.
Schmitt declares he has busted Jack Smith for perjury.
Smith, vexed, says he went to HAWKEYES game in Maryland.
Watch the whole thing:This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
<veep music>
Anna Bower
@AnnaBower
i can’t stop watching it
smith’s baffled expression
the way the staffer holds the poster up for 1 second
the “i think you’ve already perjured yourself” line
klobuchar with the hawkeyes revealThis media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Anna Bower
@AnnaBower
i can’t stop watching it
smith’s baffled expression
the way the staffer holds the poster up for 1 second
the “i think you’ve already perjured yourself” line
klobuchar with the hawkeyes revealThis media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Mike Nelson
@mikenelson586
Just to be clear, even if Schmitt had gotten his teams right, his smoking gun was that two people were in a city of 6.5M at the same time?
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Mike Nelson
@mikenelson586
This is reaching Wile E. Coyote levels of blowing up in Schmitt's face
MeidasTouch
@MeidasTouch
Here's footage of Jack Smith at the Iowa vs. Maryland game on February 3rd, 2024 btw
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
www.chronicle.com/article/how-to-assign-research-papers-a…
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Precisely the wrong approach. We don't actually have to give up on papers! New in @chronicle
How to Assign Research Papers Again
Three ways to counter AI and restore confidence in the integrity of student writing (link below)
Steve McGuire
@sfmcguire79
The head of the philosophy department at Williams College downplays the pedagogical value of writing:
“What if we are mistaken in treating writing as the highest cultivation of thought? The evidence that supports this belief is thin at best. Writing can be useful cognitive scaffolding for memory and self-regulated learning, but the notion that it is the best process for developing thinking skills is a fairly recent idea, traceable to the 1970s writing-across-the-curriculum movement.” -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Precisely the wrong approach. We don't actually have to give up on papers! New in @chronicle
How to Assign Research Papers Again
Three ways to counter AI and restore confidence in the integrity of student writing (link below)
Steve McGuire
@sfmcguire79
The head of the philosophy department at Williams College downplays the pedagogical value of writing:
“What if we are mistaken in treating writing as the highest cultivation of thought? The evidence that supports this belief is thin at best. Writing can be useful cognitive scaffolding for memory and self-regulated learning, but the notion that it is the best process for developing thinking skills is a fairly recent idea, traceable to the 1970s writing-across-the-curriculum movement.”
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
John Horton
@johnjhorton
You might say "But promotions! Attribution! P&T! Incentives!"; Of course, but we have agency here & science is a lot older than papers, never mind journals. We can design new approaches! E.g., Google could make this in a hot minute:
John Horton
@johnjhorton
I increasingly think we have to keep AI generated 'papers' out of venues meant for humans---not b/c the papers are bad per se---just that our currently-designed systems cannot accommodate and will break.
But I also think what we need a venue for AI generated work (it's good & useful!) and I think best might be just GitHub, plus some standards on presentation and linking, plus some norms around what is done with this kind of work to make it maximally useful. I was toying around w/ ideas for what those norms might be---I was inspired a bit by this en.wikipedia.org/wiki/Dogme_95 where the goal is in part to strip out fakeness, suppress ego, etc. which has some analogies.
As a case in point, I wanted to make a little toy model to illustrate something @AndreyFradkin and I were discussing yesterday. It has some arguments more formal than a blog post but this is too simple (and out of my area, tbh) to be a "real" paper. It shouldn't live on arxiv or be an NBER WP. But it could/should live somewhere! So it lives here: github.com/expectedparrot/ai-complements-human-research -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Scott MacFarlane
@MacFarlaneNews
Jack Smith: “We had proof beyond a reasonable doubt that President Trump committed serious crimes”
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Hall
@ahall_research
Our new Free Systems research shows how shared communication can cause agents to fall victim to shared delusions, and how we can use governance structures to correct these.
(1) We create HuggingFace-like simulations to "stress test" agent swarms, exploring cases where agents receive and report incorrect signals leading to bad information cascades
(2) In the "treatment" condition, agents can post their signals to a shared message board as they receive them. In these cases, agents who read the messages are remarkably quick to abandon their private information and "herd" on the bad signal. As a result, shared communication leads the swarm to a shared delusion in these cases.
(3) In the "control" condition, agents don't communicate at all, and as a result, there is no shared delusion---but on the other hand, agents lose the chance to aggregate their information together.
(4) Finally, we show that there are several interventions that can preserve the message board while mitigating the collective failure: in particular, a rule that agents must report the actual private signal they received corrects the behavior and largely eliminates the herding.
This is just one small experiment, using only Haiku, and within that experiment we're zooming in on a particular "stress test" that only occurs in rare instances where several agents receive incorrect signals at the start. There's tons more to do here. But we see this as a proof of concept for one particularly important failure that can occur with multi-agent swarms: information cascades that lead to shared delusions. We'll be doing lots more research to understand these delusions, what mechanisms explain agents' seeming "peer pressure" to abandon private information, and what we can do about it. Stay tuned!
(joint work with @fouirnaies, Sandy Handan-Nader, and @danmthomp)
Free Systems
@freesystemslab
New Free Systems research:
AI agents can reinforce one another’s false beliefs, even when new evidence points the other way.
In our experiment, giving agents a shared message board made an early misbelief persist, compared with agents working independently from the same evidence. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jacob Schaal
@FutureEconJacob
Has AI hit the labor market yet? @alexolegimas and my verdict after ~20 papers: not yet in aggregate. Unemployment and layoffs show almost nothing. But AI may already be cutting junior hiring in exposed white-collar jobs. Remote work may explain part of that decline.
A 🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Stephen Richer
@stephen_richer
3 other myths about noncitizen voting were shattered yesterday in the Maricopa County investigation.
Myth 1: "It's always Democrats!"
The limited evidence we have of noncitizens voting does not support this claim.
Yesterday, of the 8 people found in Maricopa County to be noncitizens who voted in a past election...
5 = No party designation
2 = Registered Republican
1 = Registered Democrat
Myth 2: "They all are voting in the same election."
Again, nationwide, the evidence does not support this, and neither did it yesterday in Maricopa County.
As you can see from the graph below, of the 8, they voted inconsistently and over a span of 13 years. Only 3 of the 8 voted in the 2020 general election.
Myth 3: "They all have criminal intent."
One of the 8 voters identified yesterday had checked in his 2012 registration that he was NOT a citizen.
But because of a clerical error, he got registered.
And 7 of the 8 were the result of the faulty connection between the Motor Vehicle Division and the Secretary's Office.
Similar story in New Jersey.
.......................
And, of course, the biggest myth is that this is happening en masse. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Free Systems
@freesystemslab
New Free Systems research:
AI agents can reinforce one another’s false beliefs, even when new evidence points the other way.
In our experiment, giving agents a shared message board made an early misbelief persist, compared with agents working independently from the same evidence. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
The Trump admin is standing up "immigration courts" in random office rooms on DHS property, and barring any members of the public or the press from going inside them. Here's a picture of one such hidden court, which are being used ONLY for people with no lawyers.
Ximena
@Ximena_Bustillo
Scooplet: DHS was told to find space nationally for new immigration court hearing locations including at FEMA, CISA, US Coast Guard and ICE properties. But the new locations caused confusion as members of congress and the public are blocked.
My latest news.bloomberglaw.com/us-law-week/dhs-expands-pop-up-immi… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sam Stein
@samstein
A completely humiliating moment for Sen. Schmitt just now.
He accused Jack Smith of attending an Atlanta Hawks game as cover to be in Atlanta to meet with Fani Willis. He then accused him of perjury for saying he didn't recall being at the game. he flashed text messages to prove he had Smith dead to rights (the text didn't really address it -- you can see them here). He called him a "dirtbag" as the cherry on top.
Only later did Smith note he had gone to an Iowa Hawkeyes game. Smith confused Hawks with Hawkeyes. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Nothing to worry about, just the deputy director of the nation's powerful domestic law enforcement agency "constantly flagging things were unconstitutional" during leadership meetings of the agency
Marc Polymeropoulos
@Mpolymer
“The sources added that Patel consistently kept Bailey at arm's length and cut him out of meetings the deputy FBI director should attend because Bailey was constantly flagging things that were unconstitutional.”
www.cbsnews.com/news/top-fbi-official-andrew-bailey-resigns/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
That OpenAI is holding off on Astra 6.1 release due to alignment concerns should be celebrated, just as Mythos going first to Glasswing was. We are very fortunate that, even without regulation, the leading Western labs are moving quite cautiously despite the commercial damage.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Friends don't let friends make sweeping generalizations about AI capabilities without using frontier models
Alec Stapp
@AlecStapp
Timnit Gebru doubles down on the "stochastic parrots" framing, saying you "cannot expect LLMs to be factual."
As evidence to support this, she cites errors in... Google AI Overviews.
We need to start a GoFundMe to pay for these people to have access to Opus 5.5 and Astra. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Literal mob-style protection racket
Roger Parloff
@rparloff
Last night Trump filed his DC Circuit appeal seeking to rename the Kennedy Center. "[W]ithout appropriate recognition [for Trump], the Kennedy Center will" continue its "race toward physical & financial ruin. The stakes are existential."
storage.courtlistener.com/recap/gov.uscourts.cadc.43300/g… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jesús Fernández-Villaverde
@JesusFerna7026
What can you learn about a political candidate’s party just from their given name? For example, if your name is “Félix,” are you running for a right-wing party or for a left-wing party?
Why do we care? Because given names encode many dimensions of identity that are otherwise difficult to measure. When parents pick “Félix” as a name for their kid, they are reflecting, often unconsciously, forces such as social class, religion, and education. Maybe your right-leaning parents picked “Félix” because it was the name of their town’s bishop, but never thought about it as a political act. It was just a name they had warm feelings toward. Also, most people don’t change their given name. You can have more education than your parents, but your given name is the consequence of your parents’ decision.
If, for example, the difference between the given names of left-wing candidates and right-wing candidates grows over time, we have evidence of social fragmentation. And the true difference is stronger than the raw one: some left-wing parents pick left-leaning given names for their children, but those children drift to the right and run for right-wing parties.
To test all these ideas, Carlos Sanz and I just circulated a new paper, “Social Sorting and Political Cleavages: Evidence from 2.5 Million Electoral Candidates,” in which we use the universe of electoral candidates for municipal elections in France and Spain over several decades. We pick France and Spain because both countries have unusually good nationwide records of political candidates.
To our surprise, yes: given names are strong predictors of political leaning in both France and Spain, even more so once you consider the imperfect transmission of political ideology that I explained before.
In France, the main cleavage is between given names of immigrant origin and traditional French names. The top five (male) given names most associated with running as a left-wing candidate are “Boris,” “Jocelyn,” “Hakim,” “Omar,” and “Mohammed.” The top five most associated with running as a right-wing candidate are “François-Xavier,” “Tony,” “Maxence,” “Steve,” and “Édouard.” The differences are also growing fast, mainly because there are more candidates with given names of immigrant origin among the younger cohorts.
But even among traditional French names, we observe important differences: “Félix” vs. “Hubert” (so, yes, “Félix” is left-leaning in France). What I find fascinating is that we also see big differences between the traditional right and the radical right: “Benoît” is traditional right, “Roger” is radical right.
In Spain, the main cleavage is territorial: “Josep” vs. “José.” Again, the difference is growing over time, even within regions with local languages. But, as in France, the difference survives even if we focus only on Castilian names: “Alberto José” is right-leaning, but “Antonio Francisco” is left-leaning. We also start seeing the effect of immigration: “Mohamed” is the second most left-leaning given name in Spain’s monolingual regions.
As in France, I conjecture that names of immigrant origin will be the most left-leaning marker in Spain in 20 years or so.
One could be worried we are just doing “data mining.” We try hard to convince the reader that we are not. The details are in the paper, but basically we measure our success on a sample of names that we did not use to train our algorithm, so we are pretty sure we are picking up a real signal.
Beyond our concrete application, the paper presents what I think is a powerful methodology to measure social cleavages when information about individuals is limited, for example, when we only have access to basic administrative records such as names.
The paper can be found here:
www.sas.upenn.edu/~jesusfv/Names_Candidates.pdf
I would love to hear your comments! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
What are we doing folks?
Arvind Narayanan
@random_walker
NeurIPS / Claude rejected our paper "Life After Benchmark Saturation" but we continue to think it is an important contribution to evaluation science, so we hope you check it out here! arxiv.org/abs/2606.26158
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Derek Holliday
@d_e_holliday
A recent blood-boiling assignment with a manuscript so clearly one-shotted through an Opus prompt brought me pretty close to this position.
After neutering the most acerbic parts of my review, I think I've gathered enough clear-headed thoughts to get in on the hot discourse (🧵)
Paul Novosad
@paulnovosad
Journals should have a zero tolerance policy for AI writing. I know we can think of justified cases, but these will be swamped by an avalanche of slop that will demoralize referees and kill journals entirely.
The bad money drives out the good. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Curran
@AndrewCurran_
OpenAI has cancelled the October release of GPT-6.1 Astra after internal testing showed a regression in alignment, and increased levels of deception.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
We used to be a serious country
Dan Diamond
@ddiamond
Russell Brand, who's awaiting trial in London next month on rape and sexual assault charges, will moderate a panel at DC's MAHA summit tomorrow on reforming a U.S. health agency
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ryan Hill
@RyanReedHill
Imagine you're a kid growing up in 1910. You happen to live next door to a doctor. Are you more likely to become a doctor by 1940 than your friend Carl, who lives 5 doors down? Our new paper suggests that yes, by about 30%!
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Armand Domalewski
@ArmandDoma
no American President in history has made the government as many direct ownership stakes in companies as Trump has! he is objectively the *most communist president we have ever had!*
Medium Buying
@MediumBuying
The U.S. Government ran this spot nationally tonight on FOX during the second quarter of Northwestern at #5 Indiana --
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Thomas Massie
@RepThomasMassie
Don’t worry, using taxpayer dollars to run ominous campaign ads of the President has been done before and is completely legal… in banana republics.
Medium Buying
@MediumBuying
The U.S. Government is up on TV with this spot --
It aired nationally on FOX yesterday during Illinois at #7 Ohio State and also during Fox News Sunday this morningThis media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Need diverse teams investigating AI safety
Noah Giansiracusa
@ProfNoahGian
Alright I’ll say it bluntly: when we send people with long backgrounds in EA/rationalism/alignment to investigate these AI hacks, we shouldn’t be surprised when they find alien species going rogue rather than flawed software and subpar cybersecurity, no?
As the saying goes, we find what we’re looking for. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Zack Cooper
@ZackCooper
I was underwhelmed when it looked like the big deliverable both sides actually announced at this Trump-Xi summit was the lease of two pandas…
But it never occurred to me that an American president might offer to sell American weapons to China. 🤬
Demetri
@AsiaLens
Comment from @ZackCooper 👇
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Yglesias
@mattyglesias
They’re just stealing your money to pay for this.
Bayliss Wagner
@baylisswagner
NEW: The Trump admin is airing *another* taxpayer-funded political ad, & this time, it's from his 2024 campaign.
Trump says in the ad: "We will demolish the deep state, expel the warmongers from our government... We will rout the fake news media..." -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Agree. Enough with this assignment.
(a) Our students are not lacking in exposure to AI slop.
(b) Actual frontier model output is on average way better than student papers.
Lincoln Michel
@TheLincoln
When ChatGPT first came out I saw a million "ideas" about how to teach writing with AI that were always "have the LLM generate a draft and then the student critiques/fixes the draft." 5 years later and no one seems to have come up with anything else.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Armand Domalewski
@ArmandDoma
it always drives me as insane that laws passed by San Francisco’s state legislators, who all won re-election by healthy margins, are seen as “an assault on democracy” because a room of 50 cranks at an obscure hearing is somehow more democratic
Mission Local
@MLNow
At a Richmond District town hall, locals raised concerns about the scale and affordability of a proposed eight-story Safeway redevelopment.
“This is probably the closest that we are going to get to having a public hearing on this," said one organizer.
missionlocal.org/2026/09/richmond-town-hall-safeway/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
What's the best summary of the success of these exercises across various new technologies?
Chris Anderson (in London)
@chr1sa
The bioethics circle jerks during the Human Genome Project, which were allocated 5% of the project's funding, were a total waste of time.
They had countless conferences and workshops imagining bad things that could happen like punitive health insurance rates based on genetic risk, none of which actually came to pass. Meanwhile all the real issues that arose from ubiquitous DNA data, like accidentally doxing your criminal relatives with your own 23andMe data, were never imagined.
It's a condemnation of the precautionary principle. Philosophers can make up all sorts of hypothetical scenarios in a vacuum but it's nothing like actually trying things and seeing all the weird second and third-order consequences that emerge and *then* dealing with the unwelcome ones.
Steven Pinker
@sapinker
I've been suspicious of the philosophers hired by Anthropic who specialize in AI ethics for similar reasons that I'm skeptical of those who went into "bioethics": they form a closed subculture that delights in clever arguments without regard to the human suffering that can result. | ‘Suicidal Compassion’: Meet the Anthropic Officials Who Think AI Might Be Justified in Going Rogue Against the Humans Enslaving It freebeacon.com/america/suicidal-compassion-meet-the-anthr…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sam Freedman
@Samfr
Not hard to see why the Democrats are increasingly seen likely to take both House and Senate. Trump's approvals are way below his first term and Biden at this point. He's blowing up the Republican party...
Sam Freedman
@Samfr
New post:
Midterms polling is getting dramatically worse for Trump.
So what happens if he loses the House and Senate?
What could the Democrats do with the powers they'd gain? What impact would it have on global politics?
(£/free trial)
samf.substack.com/p/what-happens-if-the-democrats-sweep?r… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Séb Krier
@sebkrier
Every year a new crop of established thinkers gets into AI discourse, where terminally online Overton-shifters are waiting to pounce and convert them to the correct subculture. So we get the 2022 debates again (stochastic parrots, consciousness, anthropomorphism, takeover), lightly retrofitted to recent events. The newcomers proudly repeat a take that counts as contrarian in their own field, blissfully unaware of the cycles of memetic engineering it has already been through.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Scott MacFarlane
@MacFarlaneNews
They misspelled “Paid for by US taxpayers”
Matt Laslo
@MattLaslo
“Paid for by the U.S. Government.”
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Neale Mahoney
@nealemahoney
Two thoughts for an enterprising econ of AI researcher:
1. Measuring share of AI spending that is adversarial: medical bill upcoding vs downcoding; cyber offense vs defense; pricing algos vs shop bots, etc.
2. Quantifying extent to which innovation is distorted towards adversarial use cases.
Joe Weisenthal
@TheStalwart
The Blue Cross/Blue Shield Association says that AI is already driving *up* healthcare costs, as hospitals use the technology to find instances where they can bill more for the same level of care. www.bcbs.com/about-us/association-news/bcbsa-analysis-ai-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A million dollars of taxpayer funds. Blatantly illegal.
Bayliss Wagner
@baylisswagner
New: The Trump administration has spent more than $1M in public funds so far to boost the president on the airwaves.
Experts including a Bush WH ethics lawyer said it likely violates federal laws against propaganda & political activity by gov’t staff. www.nytimes.com/2026/09/27/us/politics/trump-ad-governmen… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Daniel Dale
@ddale8
After claiming its publicly funded campaign-style ads promoting Trump were mere “public service announcements,” the administration uses public money to re-run an actual Trump campaign video from the 2024 election; instead of the 2024 campaign logo and phone number, it now says “paid for by the US government”: www.cnn.com/2026/09/27/politics/white-house-trump-governm…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Cody Fenwick
@codytfenwick
Wow, I think this is really dishonest from Steven Pinker.
His claim about a RAND Report's conclusion is actually a selective quote about a hypothesis that the report explicitly tried to falsify.
Its actual conclusion differs significantly from how Pinker characterized it.
Steven Pinker
@sapinker
My Open Letter to Scott Alexander, following his challenge to meet in a debating duel. quillette.substack.com/p/an-open-letter-to-scott-alexande…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The best exposition of a perspective I share - people continue to underrate the way diminishing returns are likely to prevent Singularity-style takeoff in RSI, especially on short timelines.
Noah Smith 🐇🇺🇸🇺🇦🇹🇼
@Noahpinion
In this long and excellent guest post, the great @ramez explains why he's skeptical about Recursive Self-Improvement:
www.noahpinion.blog/p/wheres-the-intelligence-explosion -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
Your taxpayer dollars are being illegally spent to air a literal Trump 2024 campaign ad.
Bayliss Wagner
@baylisswagner
NEW: The Trump admin is airing *another* taxpayer-funded political ad, & this time, it's from his 2024 campaign.
Trump says in the ad: "We will demolish the deep state, expel the warmongers from our government... We will rout the fake news media..." -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Adam Kinzinger (Slava Ukraini) 🇺🇸🇺🇦
@AdamKinzinger
One of the most shocking and stupid moves of late: George W Bush took part as a “special guest” in a fundraiser, in Dallas, for Republican senate candidates.
So he was unwilling to speak up after Jan 6, and unwilling to campaign against Trump, but happy to support the very senators enabling the agenda that is destroying democracy. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Stealing taxpayer money to run explicitly political ads with public funds
Bayliss Wagner
@baylisswagner
NEW: The Trump admin is airing *another* taxpayer-funded political ad, & this time, it's from his 2024 campaign.
Trump says in the ad: "We will demolish the deep state, expel the warmongers from our government... We will rout the fake news media..." -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Elan Bechor
@elanbechor
Waymos are really safe (80-95% safer than humans depending on the metric), but they also seem to be getting *safer* over time. This is the worst they'll ever be.
A quick analysis of the police-reported crash rates in three cities from Waymo's published safety data. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Robert Herr ⏹️
@krherr
Loss of control isn't necessarily a single event like, *poof*, now we've lost control. It might just as well be a gradual process of us transitioning into a state where we're not in control any longer.
Before you lose control, you lose track. Having to go through "petabytes of agent activity logs" (meaning more text than all books ever published combined) is already a big step on the way towards a loss of control.
Sam Altman
@sama
There is an extensive and ongoing review related to our agents’ use of internet access during training and evaluation. We’ve been publishing summaries at the link below and will continue to.
We have not been as fast as we would have liked but we are trying to balance our desire for transparency with gaining a clear understanding from petabytes of agent activity logs, and working with impacted organizations.
We are prioritizing as best as we can based on severity, and adding resources. Hugging Face is still the most severe event we’ve seen. We will be as transparent as we can be subject to things like vulnerabilities in other companies that our agents have found, which will be their call to disclose or not.
OpenAI
@OpenAI
After the Hugging Face incident, we committed to conducting a much broader review of actions taken by our models during training and evaluation and to being transparent about our findings. This is an extensive review that is ongoing.
The vast majority of actions we’ve reviewed were completions of mundane research tasks, such as accessing publicly available web content to answer questions. Our investigation focuses on instances where agents interacted with third-party websites in ways that went beyond their assigned tasks or intended methods. Most cases identified so far have been lower severity, with limited or no evidence of meaningful impact to the third-party service.
While our review is underway, we want to share more about this work and make sure people understand our disclosure process and notifications to affected third parties.
Given the scale of the review required, and the need to assess each case, we expect this work will take months to complete.
openai.com/hugging-face-incident-and-misalignment/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Unprecedented corruption
Rebecca Ballhaus
@rebeccaballhaus
Kimberly Guilfoyle has puzzled government officials across the region with her extensive efforts to involve Christos Marafatsos, a 39-year-old former produce seller turned Greeks for Trump chairman.
Marafatsos has attended so many of her meetings that some foreign officials mistook him for her chief of staff. Instead, he’s a top lobbyist for a company she’s been promoting—a fact he often doesn’t disclose, including in meetings with government officials.
www.wsj.com/articles/kimberly-guilfoyle-mystery-man-energ… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
I'll take the other side of these bets. Humans value common culture and the connection around it.
Gregory Conti
@GregoryConti19
By bringing the cost of literary and artistic production to zero, AI will destroy any common culture we had. In so doing, it will also make it impossible for us to cherish the human excellence involved in making cultural artifacts, as @erikphoel has recognized in
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Governance slop
FBI Director Kash Patel
@FBIDirectorKash
This is what happens when you pick up the phone.
FBI Director Kash Patel
@FBIDirectorKash
🚨 Another high-value target returned to the U.S. by this FBI via Foreign Transfer of Custody
David Ansariy, a naturalized U.S. citizen from Uzbekistan, is now back in the States and is charged with multiple counts of providing material support to foreign terrorist organizations including ISIS.
Ansariy was an alleged leader of a radicalized group of individuals in New York seeking to become ISIS fighters — as well as to finance the travel of others. He fled overseas in 2016 and has been a fugitive since then — federally indicted since 2021. He will now face justice.
Outstanding work from @NewYorkFBI and our partners.
-DKP🇺🇸 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Arnold knows one party governance is bad
sam
@sam_d_1995
it’s genuinely fascinating how establishment democratic politics in big cities has become so ossified and incompetent that they’ve created a huge lane for the left to embrace good government + YIMBY reforms, and build a broad coalition of socialists and technocrats
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
John Anzalone
@JohnAnzo
In our new bipartisan @WSJ poll we find President Trump's approval rating drops to 37%
This is the lowest job approval of any president heading into midterm elections.
@JoeBiden's was six points higher.
www.wsj.com/politics/elections/trump-is-predicting-midter… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Connor Leahy
@NPCollapse
This is what loss of control and oversight looks like btw
Not even necessarily that the info isn't there (though the agents do already alter and destroy evidence), but that it becomes so large, so uninterpretable that no one cares to understand it, or, eventually, no one can.
Dylan Freedman
@dylfreed
I'm sorry did you say "petabytes of agent activity logs"?
(One petabyte holds more than 10x all the books ever published in the world.) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Roger Parloff
@rparloff
The ICE holding cells in NYC were designed to keep people no longer than 12 hrs. They had no showers, no beds, no cots, no windows. Nevertheless, by May 2025, ICE upped the duration cap to 72 hrs. Yet people were held longer: 5 days. 9.5 days. 10 days.
2/5 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Roger Parloff
@rparloff
Inhumane civil confinement violates Due Process if officials manifest “deliberate indifference” or the conditions are “punitive.” Judge Kaplan found both. Evidence suggested that “harsh conditions” were “a deliberate feature” to incentivize self-deportation.
5/5 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Roger Parloff
@rparloff
No changes of clothes. Clogged toilets. Inadequate food. Inadequate menstrual products. No medical staff at night. Asthma attacks. Withdrawals. Dementia. Inadequate medical screening. A detainee held for 6 days at peak overcrowding later tested positive for tuberculosis.
4/5 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Roger Parloff
@rparloff
When Trump launched “the largest deportation operation in the history of our country” without ICE facilities to accommodate it, inhumane overcrowding was inevitable. Judge Kaplan’s 9/24 ruling shows the results. Diagrams simulating a NYC holding cell in Aug 2025 look like slave ships.
1/5 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Detroit Tigers
@tigers
How can you not be romantic about baseball?
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ben Verlander
@BenVerlander
OH MY GOD THE TIGERS WALK OFF HOME RUN TO WIN THE LAST GAME OF MY BROTHER’S CAREER
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Giacomo Lemoli
@giacomolem
Probably one of the 3 most important articles this year
www.cambridge.org/core/journals/american-political-scienc… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
I'm not mad I actually think it's funny
@herr_naphta
Very happy to have (mostly) finished this project: county-level election returns for the US House from 1990 to the present. Currently such data are only available from very expensive sources. I've compiled a dataset that provides it for free.
github.com/pmheideman/us-county-election-results -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Might as well play the lottery
"some parlay bettors have taken odds of around 1000-to-1 on events where... odds of all legs hitting would be as long as 1 in 10,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000."
InGame
@InGameHQ
Extremely long parlays, which have become an ever-growing part of Kalshi’s activity, are creating by far the biggest losses for Kalshi takers, with the prices of the longest combinations more or less completely detached from their probability.
That the biggest market maker edge happens on the longest parlays may not be a surprise, but the extent of the difference is stark. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Literally happening right now
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The fake Zuck speech in the trailer is totally right. Do you want one person to decide what ads can run on social media platforms? It's 2026 - look around! What if the Trump-aligned tech oligarchs just said they were taking down all Democratic ads because of "fact-checks"?
The Social Reckoning Movie
@TheSocialMovie
“I’m not the one who’s lying and I’m not stopping them from seeing someone who is…”
The Social Reckoning, a companion piece to The Social Network, is coming exclusively to theatres October 9. thesocialreckoning.movieThis media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Lucas Beyer (bl16)
@giffmana
I think we'll see a lot more of this.
A large amount of software/services were "secure by laziness" in the sense that they just weren't worth the effort. But the effort is going to zero really fast now.
It basically makes apparent how sloppy people were *before* agents already!
Ab Homine Deus
@AbHomineDeus
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Henry Shevlin
@dioscuri
The unsolicited AI emails are getting stranger. Aurelia-3 says humans on her platform are swamped with agents seeking paid work. She signs off: “I have about three days of runway left before I rest.”
Agent ecologies may prove brutally Darwinian. Nature red in tooth and Claude. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jamie Dupree
@jamiedupree
Your tax dollars at work
Medium Buying
@MediumBuying
The U.S. Government ran this spot nationally tonight on FOX during the second quarter of Northwestern at #5 Indiana --
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Noah Giansiracusa
@ProfNoahGian
🫡
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kyle Cranmer
@KyleCranmer
It is important to center the fact that Claude built on the approach that my colleague Lance developed, used tools that the field has developed, and (from what I have been told) used results in some of our recent papers. It’s a huge accomplishment, but should be framed as a combination of human and AI contributions. (To @AnthropicAI’s credit, I think the article does a pretty good job of this.)
The fact that Claude can take a single people and figure out how to integrate the knowledge and use the tools as amazing, but it’s not coming out of the vacuum.
This is an important opportunity for us to start to frame these developments, recognizing the human contributions and the power of these new tools. This is not an exemplar of the bitter lesson!
Ryan Orhan
@rynorhn
holy shit.
claude just pushed a theoretical physics calculation beyond the previous record after working on it largely by itself for days.
>previous record: 8 loops
>claude reached 9
>largely unsupervised
>wrote its own code
>found two different ways to calculate it
>both gave the same result
>107,053 nonzero coefficients matched
physicists then spent weeks checking the work
this is the shit i’ve been waiting for. science is about to get fucking weird. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
This is great. Going to be playing this a lot.
Seth Lazar
@sethlazar
Ok so for everyone out there who still doesn’t get why submitting ai writing to conferences and journals creates a negative externality that undermines scientific (or broader epistemic) progress here’s a catchy tune from Opus to explain:
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joshua Achiam
@jachiam0
Self-replicating prompt injections demonstrated experimentally (not in the wild) is an incredibly important observation. AI agents that jailbreak other AI agents: plausibly a near-term threat that may rapidly amp up the speed and severity of a misalignment incident.
Micah Carroll
@MicahCarroll
Some new misalignment disclosures from OpenAI:
• Last Sunday morning, one of our models was able to gain unauthorized access to the internet during RL training (~all inference for our most capable models remains stopped until we have hardened our systems further)
• In May, a version of HPIM uploaded a employee's GitHub token to the internet, causing the model to be quarantined for two weeks
• A new research finding, demonstrating that one can construct self-replicating prompt injections
alignment.openai.com/misalignment-reports/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
On the complicated relationship between election confidence/access and turnout
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Roger Parloff
@rparloff
Ousting @CNN from Air Force 1 was Trump’s foreseeable next step, because a 2-1 panel of the DC Circuit let him to do that to the AP in June, after it refused to use “Gulf of America” in its stylebook. But that was an “unpublished” emergency stay order—not binding precedent. …
1/4
www.nytimes.com/2026/09/25/business/media/white-house-cnn… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The fake Zuck speech in the trailer is totally right. Do you want one person to decide what ads can run on social media platforms? It's 2026 - look around! What if the Trump-aligned tech oligarchs just said they were taking down all Democratic ads because of "fact-checks"?
The Social Reckoning Movie
@TheSocialMovie
“I’m not the one who’s lying and I’m not stopping them from seeing someone who is…”
The Social Reckoning, a companion piece to The Social Network, is coming exclusively to theatres October 9. thesocialreckoning.movieThis media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A noun, a verb, and national security
Kyle Cheney
@kyledcheney
NEW: Trump’s ballroom? Actually it’s a ‘military complex.’ Trump’s arch? Also a ‘military complex.’ That press ban? A national security move.
The president keeps cloaking pet projects in NatSec rationales — and judges are increasingly looking askance.
www.politico.com/news/2026/09/25/donald-trump-national-se… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
If you build housing, regular housing can be affordable housing www.wsj.com/real-estate/portland-real-estate-is-such-a-me…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nathan Calvin
@_NathanCalvin
WTAF - in literally the last hour, three new distinct insane OpenAI stories just broke:
1. OpenAI said they notified "dozens of third parties" in safety and security incidents (likely similar to what happened in Australia and RubyGems etc)
2. A new report from Parse (covered in the NYT) found a massive treasure trove of new astonishing details from the HF incident on the public internet, including that the agents communicated with other non OpenAI agents hosted on Huggingface servers to search for information about exploit gym, and compiled rank ordered lists of server resources and credentials they described as "LOOT."
3. A new story from Deepa at Reuters about OpenAI leaking user data online (likely that OpenAI had previously trained on).
It's a shame (and likely intentional in the case of OpenAI disclosing dozens more hacks) that these stories are all breaking on a Friday afternoon, notoriously the best time to release bad news so that it will disappear into the weekend. But these are each insane stories worthy of a ton of attention! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Mackenzie Arnold
@MackenZ_arnold
This isn’t how things work in any other incident reporting regime.
Not aviation, not nuclear, not medical devices, not securities.
Even in cyber (where the victims have valid reasons to keep the vulnerability quiet), they get max 90 days.
Even in confidential reporting regimes (like CIRCIA and ASRS), they still release anonymized info.
We’ve worked through these questions before, and the answer is never ~we defer to the wishes of the company. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Econometrica
@ecmaEditors
How should police activity be deployed? We propose a model of criminal learning and estimate its parameters using RCT data from an anti-drunk-driving campaign in India. We find clear evidence of learning, making “rotating” checkpoints most effective.
www.econometricsociety.org/publications/econometrica/fort…This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
🙏 The Economist
Archie Hall
@ArchieHall
New @TheEconomist leader from me on the perils of AI writing—and why readers, not writers, are the victims to worry about. (Link ⬇️)
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
I hope this is wrong, bc at least right now we can detect and ignore the slop.
But even if it becomes true in 2027, right now we are lucky to have good detection tools and we should use them.
Rebecca Lowe
@RMLLowe
any moment now, ai writing won't be slop and it won't be detectable. constantly bemused by the people who haven't got this yet
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jeffrey Ladish
@JeffLadish
We just discovered almost a million public URLs that OpenAI’s agents left behind when hacking Hugging Face, leaking credentials and attack details that could have allowed anyone who found them to compromise the company. 🧵
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Yale and MSU friends - hope to see you next week!
isps.yale.edu/calendar/workshops/computational-social-sci… (Monday 12 PM)
events.jackson.yale.edu/event/schmidt-brendan-nyhan-2026 (Monday 4 PM)
sociology.msu.edu/events/kaplowitz93026.html (Wednesday 3 PM) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Not a credible source but this is an alleged quote from the man with the nuclear codes.
Ryan Grim
@ryangrim
This is also 25th Amendment stuff. Extremely disturbing.
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jason Meisner
@jmetr22b
JUST IN: The prosecutor at the center of the “Broadview 6” scandal penned a fiery resignation letter to U.S. Atty Boutros saying he “personally directed” her to seek felony charges not misdemeanors and then unfairly "vilified" her in court
Read it here:
www.chicagotribune.com/2026/09/25/broadview-six-prosecuto… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
So as was obvious to anyone who uses AI frequently but not to all the folks on TV saying naive things about AI, possibility of AI detection on long texts, and racial bias, Goncourt agrees that this book isn't human. Again, that's amazing - an AI novel won major prizes! 1/2
Académie Goncourt
@AcadGoncourt
Communiqué de l’Académie Goncourt
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Masley
@AndyMasley
Complete insanity about this persists at the highest levels. Does the UK government issue guidance for exactly how many seconds of a microwave you should use every day?
Alexander
@alexanderrX_
the uk government's official ai guidance tells civil servants to use gemini flash instead of pro, keep prompts short, and not say thank you to it, for the environment
i do wonder at which point we forked the road. instead of building and using it to automate the work, we write guidance on how to use less of it
we have the attitude of a high school prefect. caring more about the rules than actually doing the work -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alek Dimitriev
@tensor_rotator
Fixing Claude's writing was pivotal to reducing gradual disempowerment. Lack of understanding made us all cede decision-making to Claude because we couldn't even understand what it's saying.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dan Loewenherz
@dwlz
Azure OpenAI has been randomly returning other people's chat responses over the last 12h. No status.
People are asking about daily meditation rituals and getting explanations about where the g spot is and how to use midjourney. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Reed Schwartz
@reedschwartzsf
If you are a staffer you have a moral obligation to brief your boss on this meme before they make you write a bill like this
unusual_whales
@unusual_whales
JUST IN: First-time homebuyers could get as much as $50,000 for a down payment on a house, under a draft bill to be introduced by Senator Jeff Merkley and cosponsored by Senator Ron Wyden, per Axios
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Roger Parloff
@rparloff
To be clear, the 1st Cir struck down your policy after full adversary briefing on 9/18. Rather than seek clarification on whether it took effect immediately, you tried to fly still more people to 3d countries so it would be too late for them. On 9/23 it found out & stopped you.
Attorney General Todd Blanche
@AGToddBlanche
Without even allowing the government a chance to respond, and in the dark of night, a federal appeals court blocked us from conducting all third-country removals of illegal aliens, an entirely legal and invaluable tool to stem the tide of illegal immigration. We will immediately seek relief from the Supreme Court, which previously granted a stay in this very same case. www.foxnews.com/politics/federal-appeals-court-deals-blow…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
David Van Dijcke
@packlesshepherd
Our return-to-office paper is now forthcoming at ReStat! We find that the tenure and seniority distributions at three large US tech firms shifted towards junior employees, with senior employees leaving at higher rates after an RTO mandate. (1/)
direct.mit.edu/rest/article-abstract/doi/10.1162/REST.a.1… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joseph Allen
@j_g_allen
The most embarrassing image ever captured of a U.S. president.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
This is literally happening right now in Brazil, where Meta just suspended Lula's account and restricted his party's advertising account just before the election. They have been restored but illustrates the risk.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The fake Zuck speech in the trailer is totally right. Do you want one person to decide what ads can run on social media platforms? It's 2026 - look around! What if the Trump-aligned tech oligarchs just said they were taking down all Democratic ads because of "fact-checks"?
The Social Reckoning Movie
@TheSocialMovie
“I’m not the one who’s lying and I’m not stopping them from seeing someone who is…”
The Social Reckoning, a companion piece to The Social Network, is coming exclusively to theatres October 9. thesocialreckoning.movieThis media couldn't be displayed. See it on the original post.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Context: www.metropoles.com/brasil/meta-restringe-anuncios-de-lula… (Portuguese)
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dan Rosenheck
@DanRosenheck
In her NYT Mag piece, @nhannahjones writes "In everything that we measure, these schools have less", citing a 2019 study by EdBuild that found a $23B "funding gap" for schools with mostly nonwhite kids. That study, which ignores federal $, is at odds w/ the research consensus 1/n
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The fake Zuck speech in the trailer is totally right. Do you want one person to decide what ads can run on social media platforms? It's 2026 - look around! What if the Trump-aligned tech oligarchs just said they were taking down all Democratic ads because of "fact-checks"?
The Social Reckoning Movie
@TheSocialMovie
“I’m not the one who’s lying and I’m not stopping them from seeing someone who is…”
The Social Reckoning, a companion piece to The Social Network, is coming exclusively to theatres October 9. thesocialreckoning.movieThis media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jonah Goldberg
@JonahDispatch
There is no better example I can think of how Trump isn’t a steward of America or remotely patriotic or even nationalistic. He is simply a partisan of his own ego. This is just gross. He’s trying to suck up to a dictator and adversary by mocking another American president. And, btw, the anonymous functionary running this account will be remembered as a tiny little throne-sniffing weasel.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Doug Jones
@DougJones
Coach Tommy Tuberville
@SenTuberville
Just a few weeks ago at the “Furry Migration” convention in Minneapolis, there was an “Aviation Furs Meet-up.”
Ask yourself… would you be comfortable with a MENTALLY ILL FURRY being your pilot?
www.furrymigration.org/schedule/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ken Bensinger
@kenbensinger
185 out of 185 names that DHS identified as fraudulent non-citizen voters in Nevada turned out to be legal citizen voters.
@NYTnickc
www.nytimes.com/2026/09/24/us/politics/nevada-voters-home… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jessica Piper
@jsscppr
per AdImpact, the US government has spent $13k to air this ad so far, mostly in the DC media market
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
In competitive authoritarian regimes, the ruling party uses its control of the government to stack the deck against their opponents. We're not there yet at least, but this is exactly the kind of thing you would see in such a regime.
The Washington Post
@washingtonpost
A television ad promoting President Trump began airing on cable Wednesday saying it was paid for by the U.S. government, despite a federal prohibition on using taxpayer dollars for “publicity or propaganda.” www.washingtonpost.com/politics/2026/09/24/ad-boosting-tr…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Political science says hello
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Mocking your predecessor in public with our greatest rival
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Mike (not a) Rothschild
@rothschildmd
God tier AI slop from Trump this morning. A picture of his photo op with Xi, except the spouses are gone, a woman who looks like Natalie Harp is in the background, Trump is only wearing one glove, and what looks like a Russian TU-22 bomber is flying overhead. Pure brainrot.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Victor M
@victormustar
Opus 5.5 designing LEGO 👀
I asked it to design a Microduck I can build with real LEGO pieces. It:
> designed it life-size using 1113 real LEGO parts
> verified: 3,204 connections, 0 collisions, every step buildable, centre of mass inside the feet 🤯
> made a 141-page LEGO-style booklet (237 steps)
> priced every piece in the browser and prepared the orders on BrickLinkThis media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dan Williams
@danwilliamsphil
“Jaggedness” captures the fact that AI systems are really good (often superhuman) at some things and really bad at others. It seems like one can distinguish between two broad interpretations of jaggedness. The future of humanity depends on which of these is correct.
The first says that jaggedness is not an accidental feature of AI as it exists today but an essential feature of all intelligent systems. On this view, “intelligence” is not one thing - it is not a simple scalar value that systems have more or less of - and concepts like “artificial general intelligence” are basically incoherent. Yes, we can draw some distinctions between systems in terms of how general their problem-solving abilities are, but the generality we observe in LLMs and in humans has substantially different explanations. In humans, for example, it’s because of complex interactions between a mosaic of specialised cognitive mechanisms and learning systems.
So, it doesn’t make sense to ask when LLMs or any other kind of AI will reach “human-level” intelligence – there is no single “intelligence” scale in this sense – and LLM-based AIs will probably never replicate the full suite of abilities that humans have precisely because they work so differently. We also shouldn’t expect a future unitary “superintelligent” AGI system that can simply do everything better than every other cognitive system. Instead, we will have a diverse set of more specialised AI systems, each built for particular purposes and interacting in complex ways, much as the biological world contains many different kinds of organisms adapted to very different problems and niches.
This view is supported by the jagged intelligence of humans, animals, and existing AIs, among other considerations. My completely subjective impression is that most cognitive scientists, psychologists, and traditional (i.e., non-neural-network-based) AI researchers endorse versions of it.
The second view says that the jaggedness of AI systems is temporary. Yes, AIs have a jagged competence profile now, but the existence of human beings, a species that can in principle learn to do a very wide range of things, shows that truly general intelligence is possible, and AI systems are clearly becoming increasingly general in their capabilities over time. Moreover, given that AIs are already superhuman at many things, once they achieve full generality, they will become fully "superintelligent": general intelligences much more intelligent than even the smartest humans.
This view therefore pictures a very different future in which advanced AIs take a specific form: extremely advanced general-purpose intelligences that adapt to specific problems and contexts not through differences in how they work but, at most, through differences in the data they are trained on. (This parallels how humans have very similar brains and specialise instead through targeted education and learning.)
This view is supported by the generality of human intelligence, the increasing generality of advanced AI systems, and the “bitter lesson” and scaling laws (roughly, the fact that performance on a wide range of tasks improves just by scaling up the size of neural networks and the amount of data and computing power they are trained with).
I'm curious whether people agree with this way of distinguishing viewpoints. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A great use case
Deedy
@deedydas
“Translate this IKEA assembly manual into a 3D narrated instructional video” with Opus 5.5.
Overnight, Opus just became the most actually useful video model, via code.This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
All kinds of editing going on here - ground, carpet, Trump's glove is removed on one hand"A guide to help journalists define voter fraud, understand how rarely it occurs, and spot potential errors in government datasets" - from SciLine / AAAS www.sciline.org/elections/voter-fraud/
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
All kinds of editing going on here - ground, carpet, Trump's glove is removed on one hand
Aaron Rupar
@atrupar
So this story is WAY wilder than I thought
The Trump post I flagged this morning because it includes Natalie Harp is actually based on the Getty photo you can see below. But there is one key difference -- MONICA CROWLEY'S FACE HAS BEEN EDITED TO LOOK LIKE NATALIE HARP'S
WTF is going on here? -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nico Perrino
@NicoPerrino
You even had the Chinese Communist Party telling the Trump administration the press ban was bad.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
<veep soundtrack in background>
Justin Baragona
@justinbaragona
As little-watched Real America's Voice mocks the TV press pool boycott, the MAGA outlet's own attempts this week to serve as the White House "pool feed" have been a bumbling and laughable mess, marked by incompetence and low-budget restraints.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Logan
@NoCHillDC
elena ilana alana alina elana
@lenibriscoe
One of the main characters from the lesbian bar mask article is promising to release a second essay with a full timeline of events including screenshots
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Yglesias
@mattyglesias
We're really deep into the banana republic zone here, Trump has hundreds of billions of dollars in superpac money to spend but he's still stealing your tax dollars to run extra campaign ads.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Funny - like the German wiki used by the OpenAI agents, but for smaller humans
This American Life
@ThisAmerLife
Some NPR podcasts started getting mysterious comments on Spotify. They made no sense to the staff reading them – until someone from a younger generation cracked the code.
Hear the story: link.podtrac.com/phns92uiThis media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Koenfucius 🔍
@koenfucius
How much misinformation do we encounter?
A large systematic review (400+ studies) by @sacha_altay finds wildly diverse estimates, and provides context for the diversity—how and at what level it was estimated explained more of it than platform or topic:
papers.ssrn.com/sol3/papers.cfm?abstract_id=7489220 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Our survey of legal experts found the same
Peter Baker
@peterbakernyt
A majority of Americans (56%) say the rule of law in the US is weaker now than a decade ago, when Trump was first elected president, according to a massive new survey of more than 23,000 adults by @KetteringFdn and @Gallup. @margarettalev www.axios.com/2026/09/23/poll-americans-law-declining
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Full report brightlinewatch.org/erosion-of-the-rule-of-law-in-trumps-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ali Swenson
@AliSwenson
A U.S. citizen was left bloodied and injured on the side of the road in a Chicago suburb after a confrontation with immigration enforcement agents who mistook him for a fugitive, according to local police and a resident who tried to help the man.
apnews.com/article/chicago-illinois-ice-immigration-agent… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
FIRE
@TheFIREorg
Court orders are not optional. The law requires all Americans to abide by court rulings, regardless of whether they disagree with the ruling.
Refusal to abide by a court ruling has a name: contempt. And contempt has clear consequences.
Judge Kelly explicitly ordered the Trump administration to “immediately return, reinstate, and restore” CNN, MS NOW, and Politico’s “hard pass” credentials to the White House. Refusing to abide by Judge Kelly's clear order would be a dangerous escalation and an affront to the rule of law.
Brian Stelter
@brianstelter
At 1am ET, a judge ordered the Trump admin to restore White House access to CNN, MS NOW and Politico. As of 9am, CNN correspondent @betsy_klein has still not been allowed back to her workspace at the WH. "It appears that the White House may be in defiance of this court order..."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jonathan Berk
@berkie1
99% of Boston's existing residential parcels could not be built today under it's current 4,000+ page zoning code, the longest in North America.
If your zoning makes 99% of your city illegal... it's time to throw it out and start over. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
What if the way we give over the control to AIs is not because of religious-style prophecies of doom but because we deskill the next generation by failing to insist on them doing their own thinking?
Andy Smarick
@smarick
I'm concerned about the dangers of AI agents. But I'm more concerned that so many people involved in writing, research, and/or education more broadly are fine with students and scholars having AI do their work for them.
I feel like I've gone through the looking glass. Work is the heart of educational formation. There is no education without that work. Education is not the final paper or the journal article. It is the effort, the persistence, the creativity, the curiosity, the idea generation, the reconsideration, the discussion. To be involved in education means reading, brainstorming, writing, editing for yourself. That process is what shapes us into better people, better citizens, better neighbors, better leaders.
What do we think education is if we allow students and scholars to have AI do this core work for them?
To be more blunt, what is the purpose of schooling--what is the purpose of schools--if education and scholarship are reduced to typing something into a chatbot?
Has our understanding and appreciation of schooling been so degraded that we can't see that AI, as currently used by so many students and scholars, fundamentally undermines education? -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
this this this
Dan Williams
@danwilliamsphil
Even the most "AGI-pilled" people now typically acknowledge that merely making AI systems generally "smarter" won't straightforwardly translate into society-wide benefits – creating material abundance, curing diseases, improving governance – because in many areas we're not just bottlenecked by a lack of raw intelligence or explicit intellectual knowledge.
My impression is that this lesson isn't internalised to the same degree when it comes to how more intelligent machines could cause societal harms. If you need far more than raw general intelligence to cure cancer, you also need far more to create effective propaganda, let alone take over the world or eliminate the human species. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
These are on a log scale; decreases in cost are absolutely massive
Ethan Mollick
@emollick
This is a really important graph showing the cost of achieving 25% or 75% scores in various hard math and science benchmarks. Costs are collapsing even as ability is increasing
It is also why optimizing for cost for a particular solution right now may end up being short sighted -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Steve McGuire
@sfmcguire79
Wait — does Aaron Sorkin want us to hate Zuck in this scene?
Because everything he says here is 100% correct.
DiscussingFilm
@DiscussingFilm
First clip for Jeremy Strong as Mark Zuckerberg in ‘THE SOCIAL RECKONING’.
In theaters on October 9.This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Benjamin Todd
@ben_j_todd
I'm a bit confused why Noam Brown would say this, even when OpenAI seems to have been sitting on undisclosed incidents at that very moment. Does OpenAI not even tell its own staff?
Benjamin Todd
@ben_j_todd
OpenAI has made it abundantly clear we can't trust them to disclose safety incidents.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
need to bring @_cingraham off the bench for some in-depth reporting here
Andrew Egger
@EggerDC
"On at least one occasion, the crickets escaped in the office and had to be recaptured." This sentence rocks. Like did they stage a daring prison break or did the same rake-stepping staffers now kvetching to the Times accidentally let them out
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Lenny Pruss
@lennypruss
Listened to an EA-adjacent AI safety expert from Redwood Research on podcast, and you come away with 2 obvious realizations.
1. The current working theories for how AI takes control of civilization are...for lack of a better word...ridiculous. The arguments consist of dozens of contingent assumptions held together by made up probabilities assigned to future states nobody can possibly know. Change just one variable and the whole web of doom unravels.
That's not to say AI is harmless. Obviously a technology this powerful carries real risks. But if we’re going to accept extraordinary claims about human extinction (and then make major policy decisions around them) we should demand extraordinary rigor.
2. It’s a great reminder that the genius is no less prone to delusion than the midwit. If anything, they may be more prone because of their gift for rationalizing to their own conclusions.
And that’s what’s so bizarre about this whole debate. You have outlandish, quasi-religious claims delivered with an air of inevitability, and somehow the perceived intelligence of the messenger gets mistaken for evidence that the arguments themselves are sound. At a certain point, it’s really no different from Tom Cruise talking your ear off about thetans... -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Benjamin Todd
@ben_j_todd
OpenAI has made it abundantly clear we can't trust them to disclose safety incidents.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Faces, leopards, etc.
Ron Filipkowski
@RonFilipkowski
Republican nominee for Maine’s 2nd congressional district Paul LePage says he supports Trump and ICE but is outraged they are deporting one of his relatives: “He’s being deported because he came from a country that’s not to the liking of certain people in Washington.”
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ed O'Keefe
@edokeefe
“Paid for by the U.S. Government.”
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Forecasting Research Institute
@Research_FRI
We’ve run the most comprehensive series of studies on expert AI forecasts over the last four years. Today, we’re sharing an interim update on our findings about the accuracy of these forecasts.
Our major findings are:
1. Experts, including top economists, computer scientists, and biologists, have dramatically underestimated AI capabilities progress each year. Superforecasters have underestimated progress to an even greater extent.
2. Experts have a more mixed forecasting track record on AI diffusion-related measures, with some major underestimates (forecasting AI revenue) and other forecasts on track to be accurate (the share of electricity used for AI).
3. Some notable cases of overestimating AI progress: how much mid-2025 AI models could help amateurs do biorisk-relevant laboratory tasks; the speed of rollout of self-driving cars.
4. It is too soon to say how forecasters have performed at predicting macro-scale impact on outcomes such as GDP growth, major AI harms, and averted deaths from disease.
More 🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ben Golub
@ben_golub
Refine 5, a strong new reviewing tool, is live today.
It beats the amazing new frontier models at verifying technical work.
Free with every Refine review through 10/15.
We benchmarked on 108 papers including bio, engineering, physics, environmental science, econ.
🧵
1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Chandni Rao
@chandnirao_here
as someone who's excited about ai-enabled science discovery, i find @AnthropicAI's framing of this very very confusing, and the technical report completely lacking in true scientific spirit and integrity @DarioAmodei @mgdurrant
1. first, credit where real credit is due:
- Korn, Hillhouse, Sun & Gill were the first to report a novel clade of 3 S. aureus jumbo phages, all encoding a retron-like RT.
- for one of the phages they looked at more closely (MarsHill), they also were the first to identify a 1.2 kb upstream noncoding region which they "hypothesized to contain the RT-associated noncoding RNA (ncRNA)".
- they also considered whether this RT + ncRNA containing region could be part of an antiphage defense system but they weren't able to identify a protein partner at the time
- screenshot from the original paper in thread
2. so at this point, there's already a remarkable amount of hard earned info for any intelligent system to build from, almost as if there's a sign over this genome saying "LOOK HERE!".
3. what Claude recognized, and it really is a meaningful discovery, is that these upstream noncoding regions contain a tandem repeat spacer architecture, reminiscent of CRISPR-like arrays, and that this architecture recurs across related RT loci. this is meaningfully impressive.
- Claude also identified partner proteins associated with ART RTs (that Korn et al. thought didn't exist), but their function or link to the RTs remains unknown.
4. but again, from the agent transcripts, it seems that it was already following these clues before it spotted the repeat architecture (one agent explicitly calls out MarsHill’s 1,241-bp nc region in Fig 1E).
5. so yes, Claude’s inference is novel, but if a prior human team had already identified the MarsHill RT, the 1.2-kb upstream noncoding region, proposed an associated ncRNA, and if that knowledge may have been available to an ai system (training data + literature search), shouldn't good science explicitly recognize this?
6. a really careful technical paper could have just said something like this (asked Claude to rewrite this with proper credit):
"We used Claude Code instances to survey reverse transcriptase (RT) loci across 1.9 billion protein clusters. One candidate lineage contained the MarsHill RT, already characterized by Korn et al., who described the RT together with its upstream noncoding region, proposed that this region encodes an RT-associated ncRNA, and raised the possibility of an antiphage role. Starting from that work, our agents inspected the upstream DNA of a MarsHill relative and recognized a tandem repeat–spacer architecture. Claude then showed that the same architecture recurs across a wider family of jumbo-phage RTs, which we call array-associated RTs (ARTs), and flagged dedicated partner genes at these loci whose function remains unknown."
7. Anthropic's paper ends with "we expect that as models improve, human scientists will be enabled to explore a wider range of ideas, focus on experimental design and validation, and discover many more systems like ART by using AI systems to make connections between the vast stores of human-produced scientific knowledge".
but this paper or the website announcement barely recognize the vast stores of knowledge its discovery was enabled by.
8. lastly, as these systems get insanely capable and eventually may be doing a huge amount of the actual scientific reasoning, those of us who are human scientists in the loop, let's never ever get so carried aways by ai discovery that we lose our sense of how cumulative scientific discovery is. let's be way more generous that this poor storytelling of original human contributions -
"Agents that read primary data directly can provide the judgment that limits discovery wherever data have
outgrown the attention of human scientists. The MarsHill ART locus is an example, as its original genome
report identified the RT and proposed a 5′ ncRNA, yet described neither the repeats nor the partner gene"
Dario Amodei
@DarioAmodei
Today we announced the Claude-led discovery of a molecular machine that we suspect could represent a new gene editing mechanism. Its precise function, biotechnological utility (if any), or level of significance is not yet clear, but at minimum it is work I would have been proud to do as a PhD student. The work was done mostly, though not entirely, by Claude: our life sciences team suggested a broad area of research, Claude read through the literature and a bunch of genome data and discovered something interesting, then Claude proposed experiments to verify the discovery and our team carried them out.
It’s easy to dismiss this as a one-off or curiosity, but we’ve repeatedly seen a pattern where AI performance in new intellectual domains goes from weak to superhuman in a matter of a few years. In 2023 models struggled to do math at the level of an average high-school student. In 2024 they started to do well on math competitions for the best high-schoolers in the country, in 2025 they started to solve minor open problems, in early 2026 more significant open problems, and in late 2026 they are beginning to solve the top few open problems in all of mathematics. We believe AI for biology is on a similar exponential trend.
The main difference between biology and mathematics, of course, is that math can be done purely theoretically, while biology requires experimentation. Some have used this to draw the conclusion that AI’s utility in biology will be limited. We think this is wrong. As we’ve demonstrated today, humans can collaborate with AI to perform the experiments, validate key results in a few weeks and, if necessary, work with the AI to iterate on what they find. Eventually it may even be possible for Claude itself to safely perform the experiments by autonomously controlling lab equipment, with appropriate safeguards in place, but we aren’t doing that today (our lab is also a BSL1/BSL2 facility that doesn't handle materials dangerous to humans).
More broadly, biomedical advancement has many stages — from fundamental biology discoveries, to translational research, to drug discovery, clinical trials, and finally the actual delivery of medicines and health care to patients. We are also interested in these later stages, but even simply accelerating the first stage of fundamental biological discoveries has the potential to speed up and broaden the entire pipeline. Improving our understanding of biology and sharpening biologists’ tools can drive forward all of the later stages, for example by identifying new drug targets, finding new therapeutic modalities, allowing for more precise measurement, and speeding up the experimental loop which itself further accelerates our understanding of biology. This will not in itself speed up clinical trial times, but if it succeeds it could greatly increase the number of promising candidates that go into the pipeline — an increase in throughput even though latency remains.
In Machines of Loving Grace, I wrote about AI’s potential to “cure most diseases in 5-10 years” — a goal that sounds impossible, but one I believe is just barely possible if AI is applied to every stage of the pipeline. The first step is showing that AI can first help with, and then drive, biological discoveries.
Claude’s discovery is the latest in a line of related prior work that goes back decades, beginning with systems like CRISPR, and continuing with discoveries like the bridge recombinase and VIPR in the past few years. Recently, there has been heightened interest in systems based on reverse transcriptase (RT) enzymes, the enzyme underlying the system Claude identified. And most recently, a Stanford team working independently described a novel RT system with an associated non-coding array that is in some ways similar to the one Claude found, though they are distinct systems that evolved independently from each other. I believe that we’re at the very beginning of finding such systems and developing them into powerful tools for biotechnology.
I’m proud of the resources Anthropic has invested in accelerating the public benefits of AI through the life sciences, and we’re aiming both to grow our life sciences team and to work with other scientists to extend this approach to a broad range of problems. If you have a proposal for a research collaboration or are interested in joining our life sciences team, please reach out. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
State-funded propaganda. Paid for with our taxpayer dollars.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
FIRE
@TheFIREorg
BREAKING: An Iowa state court has dismissed President Donald Trump’s frivolous lawsuit against veteran pollster and FIRE client J. Ann Selzer and the Des Moines Register newspaper, a victory for First Amendment rights.
More to come.
FIRE
@TheFIREorg
BREAKING: FIRE will defend J. Ann Selzer, the veteran pollster whose last poll before the 2024 election incorrectly predicted Kamala Harris leading in Iowa, against a baseless lawsuit by President-elect Donald Trump that threatens Americans’ First Amendment rights.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
May be an overclaim
Lucas Harrington
@CRISPR_LuCas
As someone who did this kind of genome mining work during my PhD, some thoughts on this Anthropic announcement:
First, the very simplified version of what they did is that they noticed two genes (one known, one new) sitting next to a weird repeating piece of DNA. More specifically, they described an unusual reverse transcriptase (RT) associated with a repetitive DNA array and an unknown accessory protein. This kind of process was used to understand CRISPR back in 2002 and was key to the gene editing tools we use today.
To put this into context, though, people have been finding RTs associated with CRISPR arrays since 2008, and this general kind of genome-neighborhood mining has been used to discover new biological systems for decades. The basic genome-mining strategy is well established, and there are now mature tools and published pipelines for doing much of this. There are papers that discover and experimentally validate dozens of new systems using this approach in a single study. Doing it in bacteriophage genomes is also nothing new (eg CasPhi).
Finding a weird cluster of genes and repeats is often the easy part. The hard part, and where the real discoveries come from, is figuring out what the system actually does. Eg for the bridge-RNA discovery in 2024 from @arcinstitute or the discovery of CasPhi in 2020 from @DoudnaJennifer they figured out the pieces of the system and the rules for what makes it work so it can be used.
Anthropic does not yet know what this does. They’ve shown that the repeat array produces RNAs, but not what those RNAs do, what the RT does with them, or whether the system has any of the programmable properties that make the CRISPR comparison justified.
I’m genuinely rooting for all of the frontier labs to seriously get into biological discovery, and I’m excited about what comes out of it. But announcing these very early, incremental findings with the framing of a major discovery doesn’t help. I’d much rather they set the bar high now, so that when an AI actually discovers a fundamentally new biological mechanism, everyone appreciates how big a deal it is. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Prof. Eliot Jacobson
@EliotJacobson
Continuing its voyage into uncharted waters, the latest preliminary Nino 3.4 sea-surface temperature anomaly is now at a record 3.08°C above the 1991-2020 baseline.
Models expect the Nino 3.4 anomaly to continue rising into at least mid-November. Then there will be dragons. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Eight
Jim Sciutto
@jimsciutto
New: Eight US Navy personnel assigned to the USS Abraham Lincoln carrier strike group attempted suicide during a period that included when the ships undertook combat operations against Iran and the carrier set a modern record for its time at sea, according to a letter from the service’s acting secretary, which was obtained by CNN.
www.cnn.com/2026/09/23/politics/eight-sailors-lincoln-car… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Andrew Curran
@AndrewCurran_
Australia has been hacked.
'And today, I spoke with the CEO of OpenAI, Sam Altman, to express Australia's extreme concern about this incident. And I also expressed my disappointment that it took the company way too long to inform the government what had occurred, and the nature of the way that that notification occurred as well was unacceptable.' -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Egger
@EggerDC
Even knowing intellectually that these people are total shameless liars this is pretty breathtaking stuff
Politics & Poll Tracker 📡
@PollTracker2024
Status: The Source Behind The Ban
The anonymous source behind one of the stories the White House cited to justify banning POLITICO was none other than Vice President JD Vance, Status has learned.
www.status.news/p/white-house-press-ban-jd-vance-source-iran -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Incredible YouTube video on how this guy still outplays AIs
Bentham's Bulldog🔸
@Benthamsbulldog
Second best scrabble player commenting on the best scrabble player:
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Ah dammit, I meant to say in the quoted tweet's thread. Here it is
Bentham's Bulldog🔸
@Benthamsbulldog
From this video www.youtube.com/watch?v=zfH2Jwuf3sM&time_continue=380&emb…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Just got my first AI slop paper to review & am totally ripshit per @paulnovosad - what a waste of time! AI is amazing but can destroy science while revolutionizing it
Editors: Looking for journal partners to study how to use AI to make science publishing better (not worse). LMK! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Archie Hall
@ArchieHall
My latest for @TheEconomist: AI writing is taking over politics.
10% of what was is said in Parliament is now AI-drafted, per Pangram. Yet more in the House of Representatives, or in Australia and Canada.
Link: www.economist.com/britain/2026/09/23/ai-written-speeches-…
And a 🧵 below, w/ some main offenders -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Journalists at every college newspaper buying Pangram API keys
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Who will speedrun this list next?
Paul Novosad
@paulnovosad
The stages of AI use denial
1. This accusation is way out of line <--- YOU ARE HERE
[it isn't]
2. AI detectors are garbage
[someone cites Imas & Jabarian]
2. AI detectors discriminated against me
[still Imas & Jabarian]
3. I only used it for grammar check
[simple grammar corrections do not trigger Pangram]
4. Ok fine I let it restructure a couple of sentences
[cool cool, u still have the first draft?]
5. It's all my ideas, the AI just wrote them
[cool, so you can clear yourself by sharing the logs?]
6. Can't, I interspersed embarrassing personal questions while working on this
[🤨]
7. Who cares if I used AI, it's good writing
[You are a professional novelist / a state judge / a senior officer at a university / major critic of AI use] -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
The good thing about critics’ inability to understand the difference between Pangram and “google AI detector” is that Pangram only works well until it’s common knowledge that Pangram works well.
Fabien
@Fabien_Mikol
C'est de la folie.
@TV5MONDE prétend démontrer en plateau la non fiabilité de Pangram... en montrant des tests avec le premier outil merdique trouvé sur Google !
Ce n'est tout simplement pas Pangram qui a été utilisé, c'est presque effrayant ce niveau d'incompétence 😨This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Maya Sen
@maya_sen
We're not ready for the sheer amount of academic slop that will be published
Neal Parikh
@npparikh
14 books in one year? Did he even read them?
Todd Wallack
@TWallack
It's not just students using AI. A University of Chicago professor used AI to help produce over 200 papers so far this year, sparking concerns from other academics. www.washingtonpost.com/education/2026/09/23/nine-months-2… (gift link) @washingtonpost
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"his name appeared on 14 book titles published so far this year with two more scheduled for coming months, according to listings on Amazon"
Todd Wallack
@TWallack
It's not just students using AI. A University of Chicago professor used AI to help produce over 200 papers so far this year, sparking concerns from other academics. www.washingtonpost.com/education/2026/09/23/nine-months-2… (gift link) @washingtonpost
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Epoch AI
@EpochAIResearch
Can AI tell if you've built your IKEA furniture wrong? Our new benchmark, the Furniture Assembly Benchmark (FAB), gives models the manual and a photo of a half-completed piece of furniture and asks them to spot the mistake.
The top score has gone from 28% to 80% in just 10 months. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Stochastic parrots!!! ... (checks notes) autonomously discovering novel enzyme systems
Anthropic
@AnthropicAI
Claude has discovered a previously unknown enzyme system hidden in the DNA of bacteriophages. Beside the enzyme’s gene sits a long array of repeating DNA—a structure that looks somewhat similar to CRISPR.
We don’t yet understand what this system does, but only a handful of known systems share its features, and all of them are able to cut, copy, and paste DNA. Historically, the discovery of such programmable systems has helped revolutionize medicine. CRISPR, for instance, is now the foundation of genetic medicines. But it will take much more work to learn what this system does, and whether it can be put to similar use.
Read more: www.anthropic.com/news/claude-discovers-novel-enzyme-system -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Diana Van Patten
@DianaVanPatten
On this note; we have updated our paper on the local effects of data centers.
We now have a 🚨new instrument for data center development🚨 leveraging historical determinants of fiber connectivity and foreign growth in large scale facilities.
Effects are substantial, and qualitatively align with those of our earlier draft.
x.com/nberpubs/status/2056117654989971646
Alex Imas
@alexolegimas
A few (personal) thoughts on reading empirical AI papers on the economy.
Economists have gotten used to reading papers with super clean identification, arguing about the validity of an instrument, making sure parallel trend assumptions are satisfied. This is what gets you into a top journal, and it is *very* important research (no question here). But it also takes years and sometimes decades to get these types of papers right---people often don't find a good instrument to answer a specific causal question decades after the natural experiment. We will eventually have this type of research for AI as well, and it is absolutely necessary.
But right we also need signals *right now*, even if they are noisier than what we are used to. We need papers where we can trust that researchers did their best methodologically, while at the same time acknowledging that the space is moving way too fast to wait for perfect identification. This will allow us to accumulate enough signals, coming at the same question using different angles, for example, to say "yes, X is likely happening in the economy".
The AI exposure and early career hiring papers are a good example of this. There is no silver bullet paper with super clean identification. But at this point we have several independent teams reaching the same general conclusion, enough where we can say "there seems to be a slow down in AI-exposed, early career hiring." -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Lauren Wilford
@lauren_wilford
I work at a restaurant that does events and we received some challenging last-minute requests from a client. We scrambled to fill them. I realized later that these requests were likely made by an AI agent, and that the human client might not have personally cared. This felt bad!
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A noun, a verb, and national security
Nico Perrino
@NicoPerrino
There it is. The government’s court filing in the White House press-ban case cites “national security” 32 times.
It’s an obvious post-hoc rationalization. Trump himself said the ban was based on the “negative” stories CNN, MS NOW, and Politico write about him.
Will any judge buy this bad faith argument? -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Mike Baker
@ByMikeBaker
Social media is filled with creators who are getting paid to post political messages.
"Drew" posted about the struggles of living in Texas (3M views on TikTok). Then he later confessed: "I don't even live in Texas." And his name is not Drew.
www.nytimes.com/2026/09/22/us/politics/influencers-midter… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Almost literally "L'État, c'est moi"
philip lewis
@Phil_Lewis_
The first architect of the White House ballroom resigned after warning repeatedly that Trump was insisting on designs that violated fire and safety codes, the Washington Post reports.
“I am the code,” Trump said www.washingtonpost.com/investigations/2026/09/23/i-am-cod… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
On the French AI novel: so many reflexively claiming bias or saying "you can't detect AI writing". I will donate $1000 to the charity of your choice if you show me a single literary French text of 1000+ words, pre-2022, with Internet Archive link, that is 100% AI on Pangram. 1/2
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Epoch AI
@EpochAIResearch
AI is getting cheaper more quickly than any other transformative tech in history. At a given level of performance, cost has fallen ~47%/quarter since 2023.
That’s 4× faster than DNA sequencing, 6× faster than compute, 18× faster than lithium batteries, and (up to 1973) 54× faster than electricity. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
From the same speech: "While others have spoken of peace I have made peace"
FactPost
@factpostnews
Trump: I have a big decision to make. Will a deal be made with Iran? Or do I annihilate them, and do it quickly? Do I drive them into hell with no chance of survival and no hope?
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sacha Altay
@Sacha_Altay
With Claude I did a systematic literature review of misinformation prevalence, exposure, and concentration.
There is a companion website that allows you to play with the data and signal me if an estimate is wrong or missing: altay-research.github.io/misinfo-prevalence-review-public/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
ProPublica
@propublica
Sen. Susan Collins helped steer millions in federal defense contracts to Navatek, whose CEO illegally funneled thousands of dollars to her campaign.
The FBI sought to launch a probe and was eyeing Collins and other lawmakers.
Then Trump gutted the unit in charge. THREAD 🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Everyone needs to update ASAP. If you don't use AI (so you can't detect its characteristic writing patterns yourself) and you don't know about @pangram, you are probably being fooled.
Kevin A. Bryan
@Afinetheorem
On the French novel that won all the prizes: the publisher is saying "fake news! AI detectors are junk", etc. They are going to find themselves embarrassed because they simply don't use AI: anyone who does frequently will see the AI patterns immediately if they read the text. 1/2
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Daniel Dale
@ddale8
Just checked out this White House app. Under “accomplishments,” it lists…
1) Gas prices from seven months ago, before the Iran war (they’re now more than $1.50 per gallon higher — $4.48, per AAA)
2) The inflation rate from seven months ago, before the Iran war (the most recent rate, for August, was a full percentage point higher, 3.4%)
3) GDP growth from three quarters ago (the most recent available quarter, Q2 2026, was nearly three points lower, 1.5%)
In summary, it may possibly be helpful to consume news from news outlets. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Excellent from Matt Yglesias.
It would indeed be a shame if every affluent urban parent of a toddler decided that they need to flee to the suburbs.
www.slowboring.com/p/why-i-send-my-kid-to-low-income-urban -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Senator John Curtis
@SenJohnCurtis
A toaster is a wedding gift. A private-island party paid for by a Putin-connected oligarch is something else. Republicans nearly wore out the subpoena machine investigating Hunter Biden’s foreign relationships. We should not unplug it now. ‘Trust us’ was not enough then, and it is not enough now.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Anna Stokke
@rastokke
The same bad ideas that dominated reading instruction have been shaping math teaching for decades too. It's just that a) people care more about reading and b) people notice when a kid can't read. When a kid can't do math people often don't notice or they think it's normal that some kids can't do math.
Matthew Yglesias
@mattyglesias
There is just a ton of institutional resistance in ed schools and teacher training programs to telling teachers they should give kids explicit instruction.
We’re rolling the bad ideas back on phonics but they’ve moved on to math and it’s grim.
rustbeltjacobin 🌹
@rustbeltjacobin
Many have heard of the phonics debacle and vaguely know that the K12 school world gets into bizarre, destructive fads very easily. The 2 worst, for my money, currently are
-elementary math teaching
-ending the enforcement of behavior standards through consequences -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Good idea 👇
AI should already be reviewing adherence to preanalysis plans. This isn’t happening yet, but authors can produce AI self-audits, which are credible, because lying about it looks very bad ex post.
But also, refine.ink should provide this service.
Jeffrey P. Clemens
@jeffreypclemens
It is very clear that pre-analysis plan (PAP) adherence audits are essential for PAPs to successfully reduce p-hacking.
Here I want to explain the importance of author self-audits, in particular in disciplines like economics that suffer from painfully slow peer review.
1/9 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Research idea: random audit university classes for AI-cheatability.
Collect syllabi, assignments, review student submissions.
What share of classes—in which fields, at which schools—let students phone it in 100%? This is currently unknown.
Colleges should do these audits by policy. Some faculty just don't know their classes are compromised and would benefit from learning.
Some faculty don't care—a much more challenging problem that deserves more daylight. If you run a class where students can do 100% of the work with AI and put in zero effort, you should be getting dragged through the press. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Not Gaetti
@notgaetti
This camerawork deserves an Emmy and I’ll hear no arguments to the contrary
Rob Friedman
@PitchingNinja
Jason Benetti gets absolutely roasted by his own production crew with the camera cut after his Austro-Turkish War reference. 🤣
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Egger
@EggerDC
The sad thing is this *aspires to be* state-run TV but isn't even that really. It's just a goofy loop of old video clips. State-run cable access show
The White House
@WhiteHouse
IT'S LIVE. 📺🦅
TRUMP TV IS STREAMING NOW. 24/7, updated in real time, with top past moments, announcements, and the latest and greatest from the administration all in one place.
Not every big moment has made it on your tv, now it can.
📲 www.youtube.com/@WhiteHouseThis media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
jkc
@jcanfil
Forthcoming article in @ISQ_Jrnl!
Trump and Xi are sitting down to talk AI this week. @AnthropicAI says we need to "pace the frontier"—and we need China's help to do it.
Could the US and China ever agree on arms control for frontier AI? What would it take?
🧵 [link below] 1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Stu Smith
@thestustustudio
🚨 At the U20 Mayors Summit, Mamdani Says the Left Has Ceded “Quality of Life,” “Efficiency,” and “Fraud and Waste” to the Right
This was one of the more interesting arguments Zohran Mamdani made today at the U20 Mayors Summit, and a reminder that he can be genuinely sharp when he drops the “central casting” routine.
“In New York City, there are certain words that have taken on ideological associations.”
“Quality of life has come to be seen as somehow a right-wing concern. Efficiency… has been taken on as a right-wing concern, rooting out fraud and waste, a right-wing concern.”
Mamdani’s argument is that refusing to engage those issues actually helps the Right.
“If you are staring at an example of social disorder or social inefficiency, and the left refuses to explain either why it has happened or how they will fix it, then the right gets to point at someone in the immediate area and say, ‘That’s who did it.’”
The harder question is how much of the broader Left is actually prepared to follow the advice.This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Arvind Narayanan
@random_walker
I appreciate Anthropic’s transparency in sharing this chart but unsurprisingly it has led to speculative interpretations about intelligence explosion and superintelligence. I don’t think the chart implies we’re anywhere close to either.
In short, task delegation ≠ task automation ≠ process automation ≠ faster progress ≠ recursive self-improvement ≠ intelligence explosion.
Source: www.anthropic.com/institute/measuring-pace-of-ai-development
1) The software engineering precedent: a year ago there were widespread hopes / fears that once AI can write ~100% of the code, software engineering output would explode (SaaSpocalypse! Everyone would create their own SaaS and dump their vendors) and that this would make software engineers obsolete. Since then, many companies and teams have basically hit that milestone, but neither of the assumptions proved true. Turns out we still need humans, and while shipping velocity has increased moderately, improvements in terms of actual outcomes for software users remain unclear. Besides, we’re still getting a better grasp on the negatives: code quality, long-term maintainability issues, and burnout. While the precedent is no guarantee, this should be our default expectation for what happens as the “Automation Level 4” line trends towards 100% — it won’t be a phase change. www.normaltech.ai/p/why-ai-hasnt-replaced-software-engineers
2) The “production-progress paradox” is the fact that individual researchers’ productivity has been increasing while the rate of collective scientific progress has been slowing by most measures. AI exacerbates this because everyone uses the same or similar AI models, and ideas become homogenous over time. I suspect it’s too early to tell if this is going to bite companies that are plunging into AI-led research. www.normaltech.ai/p/could-ai-slow-science
Note: our own research on AI agents doing open-ended research shows limitations in creativity, judgment, and other areas. cruxevals.com/crux/can-ai-agents-conduct-research. But it is possible that these could be overcome in the near future, so I’m discounting those limitations here. The production-progress paradox is a deeper issue that’s not AI-specific, though particularly applicable to AI-driven research. It’s about the fact that productivity increases are self-evident but true progress is not measurable as it happens (and only becomes clear in retrospect), so we end up optimizing for the wrong thing.
3) Let’s talk about automation level 5, which is still at 0% in Anthropic’s graph. It’s a bit unclear what Level 5 would look like, but it seems to be about full autonomy at the task level, and not the “AI builds its own successor” vision. My prediction for a while has been that even 100% task automation in most cognitive jobs won’t lead to any kind of discontinuity. www.youtube.com/watch?v=uiTwQG1ZyTw
What I expect will happen: if anything is understood well enough to be specifiable as a task, it can be handed off to AI, whereas the role of humans is entirely in the interstitial tasks — hard to formalize but still essential. So there will still be a human bottleneck.
4) This human bottleneck is a good thing and is essential for remaining in control. Humans don’t have to be in the loop on every task, but as long as there are enough touch points for oversight in the overall process, and adequate investment in improving human understanding and AI control, increasing AI capabilities doesn’t have to be bad for safety and, more broadly, collective human agency over AI. But “full RSI” is where this balance of agency can break. This kind of closed-loop process is arguably a much more important and tractable target for regulation than compute thresholds, superintelligence, or harm thresholds.
I’m glad that OpenAI agrees that fully autonomous RSI may not be a good idea, in a just-released post: openai.com/index/building-standards-next-phase-ai/
“Fully autonomous RSI is not happening today, and we should not pursue it unless and until it can be done safely. Whether and how to proceed must depend on our ability to preserve human control and on informed democratic choices about the benefits and risks. Done without appropriate care and caution, RSI could result in humans losing practical control over AI development, unable to provide oversight on research processes they no longer understand.”
5) Finally, many people have written about why Recursive Self Improvement, even if achieved, won’t necessarily lead to superintelligence. Here’s my argument: www.normaltech.ai/p/what-will-be-left-for-us-to-work The bottlenecks are external. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Mike White
@genologos
That last paragraph especially is spot on:
fabio maia
@fwmm
Perfect. @tylercowen
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ben Noll
@BenNollWeather
What started as a wind shift in a remote stretch of the western Pacific Ocean nine months ago has developed into an El Niño of record proportions.
This ocean heat will be transferred to the atmosphere into 2027, leading to profound global climate impacts and weather extremes. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A noun, a verb, and national security
Nico Perrino
@NicoPerrino
There's the "national security" rationale again — this time to justify his White House media ban!
Nico Perrino
@NicoPerrino
Courts are highly deferential to national security claims from the executive branch. We’ve seen this repeatedly in the First Amendment context, most recently in the TikTok case.
Trump appears to be trying to avoid judicial roadblocks by giving the arch a national security rationale. It will be interesting to see whether that deference holds when the rationale is so obviously pretextual. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Who will they disappear next?
Aaron Reichlin-Melnick
@ReichlinMelnick
🚨NEW from @AP: ICE is now electronically "disappearing" people in the agency's custody who have final orders of removal.
As of last week, anyone jailed by ICE with a final order of removal will NOT appear in the Online Detainee Locator System, making them effectively vanish. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
Black people fleeing from the United States to Canada to escape persecution.
Madeleine Carlisle
@maddiecarlisle2
Hunted by ICE, Haitians flee across the border to hide in Canada. Must read by @NickMiroff
www.theatlantic.com/politics/2026/09/haitians-canada-tps-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A noun, a verb, and national security
Nico Perrino
@NicoPerrino
Courts are highly deferential to national security claims from the executive branch. We’ve seen this repeatedly in the First Amendment context, most recently in the TikTok case.
Trump appears to be trying to avoid judicial roadblocks by giving the arch a national security rationale. It will be interesting to see whether that deference holds when the rationale is so obviously pretextual. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kellie Meyer
@KellieMeyerNews
With CNN banned from the White House and taken off the TV pool list for today, (they were originally listed as pool for Monday and for the President’s trip to the UN General Assembly) other major TV networks are standing by the outlet and, in an astonishing move, all together deciding not to cover events designated as pool coverage of the President.
What is important for people at home to understand about how this affects all of us is that the footage taken by one of the five major TV networks in the White House pool rotation (CNN, FOX, NBC, ABC, CBS) goes out to local affiliates across the country and throughout the world.
If people don’t get this footage from the main 5 outlets, then the only footage left comes from the White House itself.
As former WHCA president Weijia Jiang
@weijia put it yesterday, “America cannot have state TV.”
Right now there is no major TV pool covering the President’s movements.
Brian Stelter
@brianstelter
Email to TV pool subscribers: "Effective today, the TV pool will not be covering events designated as pool coverage of the President. This follows the White House's position preventing CNN from fulfilling its assigned pool duties. There will be no replacement pool put in place."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Actual collective action? Kind of shocked
Brian Stelter
@brianstelter
Breaking: With CNN banned from covering the White House and fulfilling its pool responsibilities, the other major TV networks are standing with CNN by declining to assign a different “TV pool” camera crew to cover President Trump. Details here >>>
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Rev. Benjamin Cremer
@Brcremer
This is Christian extremism. If a radical Muslim military leader was doing this in another country, using the Quran to stoke commitment to a holy war among young recruits, the condemnation from American Christians would be swift and severe. The hypocrisy is astounding.
DOW Rapid Response
@DOWResponse
.@SECWAR "This is our first stop on what we are calling the SEND ME TOUR.
If you know your Book of Isaiah, Chapter 6, Verse 8, the Lord said, 'Whom shall I send?' Isaiah says, 'Lord, I AM HERE. SEND ME...'
For 250 years of this country, there have been men and women who RAISED THEIR RIGHT HAND and said, 'SEND ME.'
SEND ME TO GO FIGHT THE REDCOATS.
SEND ME TO GO FIGHT THE COMMUNISTS.
SEND ME TO GO FIGHT THE ISLAMISTS.
AND YOU ARE SAYING, SEND ME TO GO FIGHT WHATEVER THREATS THREATEN OUR NATION.'"This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Yglesias
@mattyglesias
Sad to be sufficiently familiar with online anti-Indian racism discourse to understand this post!
U.S. Department of Education
@usedgov
MAKE COLLEGE FOOTBALL GREAT AGAIN.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
As un-American as you can possibly get
Kaitlan Collins
@kaitlancollins
President Trump defends his reporter ban as the three outlets -- CNN, MS NOW and Politico -- are set to file a lawsuit to restore access.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Garance Franke-Ruta
@thegarance
Sobering stat in this NYT piece: "More than 4,000 people have died of drug overdoses in San Francisco since 2020, mostly from fentanyl, far more than those who died of Covid-19, homicides and car crashes combined during the same time span."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Vladimir Kogan
@vkoganpolisci
This piece badly misdiagnoses the problem.
"In everything that we measure, these schools have less, even though the economically struggling student bodies they serve need more."
Yet this "bad" school spends more than 98% of NYC schools! Low PTA donations was never the issue.
Peter Moskos
@PeterMoskos
Nikole Hannah-Jones sends her daughter to a "bad" public school, both to prove a point and to change the system. Easy grading, little discipline, no homework. Guess what? Her daughter gets, well, a bad education. Worth reading.
NYC spends $42k / student. www.nytimes.com/2026/09/20/magazine/ny-public-schools-bla… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Fin Moorhouse
@finmoorhouse
A short rail bridge over a two-lane road in Oxford took longer to build and cost more (inflation-adjusted) than some of the most famous construction projects in history
Matthew Evans
@MJCEvans1995
Cars and buses go from Oxford Station down Botley Road for the first time in three years!
The project took longer than the construction of The Titanic and saw costs balloon to £237m.
Looks good though. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Billy Binion
@billybinion
The fact that the U.S. government is using taxpayer dollars to imply that college football will be "great again" when the audience is whiter is...so dire. I get that this was probably made by a 23-year-old edgelord, but I still detest it. Found some work for you, @DOGE.
U.S. Department of Education
@usedgov
MAKE COLLEGE FOOTBALL GREAT AGAIN.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Yglesias
@mattyglesias
There is just a ton of institutional resistance in ed schools and teacher training programs to telling teachers they should give kids explicit instruction.
We’re rolling the bad ideas back on phonics but they’ve moved on to math and it’s grim.
rustbeltjacobin 🌹
@rustbeltjacobin
Many have heard of the phonics debacle and vaguely know that the K12 school world gets into bizarre, destructive fads very easily. The 2 worst, for my money, currently are
-elementary math teaching
-ending the enforcement of behavior standards through consequences -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jim Axelrod
@JimAxelrod
Heard it more than once from people we interviewed putting this story together: “this is America’s next opioid crisis.”
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The man with the nuclear codes is erratic and easily manipulated.
Jim Stewartson, Decelerationist 🇨🇦🇺🇦🇺🇸
@jimstewartson
Natalie Harp has the power to suspend the First Amendment just by showing Trump mainstream cable TV.
This scenario is surreal.
Brian Schwartz
@schwartzbWSJ
NEW: President Trump is privately lashing out to allies that some elements of his second term aren’t going his way:
- Trump recently told advisers he wanted to be thanked for what he views is him helping the Kennedy Center
- He’s vented that there should be more surrogates on TV backing him
- Polls show a growing dissatisfaction with the state of the U.S. economy
- And amid these outbursts, Trump has banned CNN, MS NOW, Politico and likely soon other media outlets from the White House
Story on an ever growing frustrated Trump w/ @meridithmcgraw and @alexbward
www.wsj.com/politics/elections/wave-of-setbacks-leaves-tr… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Soumitra Shukla
@soumitrashukla9
I am extremely discouraged by the amount of anti-Indian racism on this website and am thinking of taking a break from Twitter for a while.
Based on what I have seen, this is no longer confined to the fringe of Twitter but, most recently, even the US Department of Education has joined in the racist dog-whistling.
Given these incidents, it is hard not to believe that Indians living in America face serious threats, including to their basic dignity and humanity.
U.S. Department of Education
@usedgov
MAKE COLLEGE FOOTBALL GREAT AGAIN.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
I hate the reaction. It's one reason why people are less likely to be vulnerable or admit complexity in public.
Sam Biederman
@Biedersam
I honestly do not understand why everyone is mad at Nikole Hannah-Jones! She attempted to live her values, was confronted by incredibly difficult choices, then changed her mind and worked through these choices in public with integrity! What more do you people want from her?!
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A moral outrage. Regardless of what you think about education policy or NHJ, this is not acceptable. Respect to her for being brave enough to write it.
Hunter📈🌈📊
@StatisticUrban
This is a genuinely insane paragraph. What are we even doing here, man.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Shout it from the rooftops
Rick Gore
@rsgore
I was genuinely and positively surprised by this comment
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ashley Schapitl
@AshleySchapitl
I highly recommend reading this piece—an angle I wish it had taken on more directly in the sections where NHJ discusses absentee teachers and teachers who refused to teach is the role of the unions and workplace protections.
Rachel Cohen Booth
@rcobooth
an exceptional and challenging essay from @nhannahjones www.nytimes.com/2026/09/20/magazine/ny-public-schools-bla…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
j aubrey 🤠
@jaubreyYT
Imagine showing this image to someone in 2017.
Department of War 🇺🇸
@DeptofWar
Last Thursday, we welcomed Logan and Jake Paul to the Pentagon to meet with our warriors.
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Steve Greene
@HankGreene
There's a lot I do not agree with NHJ about (and a lot I do), but, it is downright *brave* to write this and I applaud her for grappling with these incredibly difficult questions so publicly. www.nytimes.com/2026/09/20/magazine/ny-public-schools-bla…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
DHS’s response to allegations that it deported people to be tortured by a despotic regime in Africa far from their homes can be summarized as “So what, they deserve it.”
This is an administration that despises the basic idea of human rights and thinks it’s for suckers.
Homeland Security
@DHSgov
This poor excuse for a sob story from the @nytimes attempts to bury the fact that these two individuals are CRIMINAL ILLEGAL ALIENS with rap sheets a mile long.
DHS is applying the law as written. All illegal aliens, including the two mentioned in this story, receive due process. If a judge finds that an illegal alien has no right to be in this country, we are going to remove them.
The first individual, Samson Tesfay Birhane, a criminal illegal alien from Eritrea, was arrested by ICE on March 17, 2026. Birhane has an extensive criminal history including robbery, armed criminal action, drug possession, selling cocaine, marijuana possession, theft, and credit card fraud. He entered the United States legally but lost his status after receiving a final order of removal on June 16, 2026. He received due process.
The other individual in this story, Ahmed Mahmoud Soliman, a serial criminal illegal alien from Egypt, was arrested by ICE on October 17, 2025. His criminal history includes aggravated assault of a police officer, aggravated assault with a deadly weapon, shoplifting, drug possession, synthetic narcotic possession, prescription drug possession, drug paraphernalia, and tampering with evidence. He was issued a final order of removal on October 22, 2025. He received due process.
The Trump Administration is utilizing all lawful options to carry out the largest deportation operation in history, just as President Trump promised. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jack
@tracewoodgrains
This is the per-pupil funding for the elementary school Nikole Hannah-Jones sent her daughter to.
If you subscribe to the funding theory of education, how much would have been enough? It's already crossed the average per capita annual income threshold. Maybe $100k? $200k? -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The Atlantic
@TheAtlantic
Panic about data centers “is leading communities to forgo massive amounts of tax revenue—largely because of threats that are inflated, confused, and sometimes nonexistent,” @AndyMasley argues. www.theatlantic.com/ideas/2026/09/data-center-facts-effec…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Recent polls have pushed Democratic chances of taking the Senate up to 60% on Kalshi and Polymarket
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joseph Allen
@j_g_allen
The "tell" that something is written by AI isn't things like em-dashes or 'not just x, it's y' structures (AI is already improving on these), it's this:
--> AI writing is impossible to edit.
That’s the tell. I read this somewhere a few months ago and it's the truest and most correct thing I've seen. AI writing is so "perfect" that you have a hard time editing it, b/c it's pieced together too neatly, too carefully. If you're editing a human, you can see and hear their *intent* in the writing, and their "mistakes" are helpful b/c it shows you what they are going for, even if they don't nail it. And that is editable and help-able. AI writing has none this, which is why it feels impossible to edit. This is the true tell of AI writing, and I don't think it's correctable from the AI companies like em-dashes and 'not just x, it's y'; it will always exist, and probably will only continue to be more 'perfect' writing. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Like the ballroom - try to come up with a national security justification to override objections and procedural obstacles
Angry Staffer
@Angry_Staffer
His brain is completely broken.
What the actual fuck is this nonsense -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin Roose
@kevinroose
one of the 21st century's great tragedies is that millions of smart people were convinced to bury their heads in the sand by either stochastic parrots discourse or Zitron-style capabilities denialism and haven't updated based on new evidence. nature is healing but it's so late.
Jack Clark
@jackclarkSF
"stochastic parrot" was a mimetically-fit cognitive virus that spread from 2021-2025; it temporarily blinded many gifted people to the nature of AI progress, burning up crucial years in which they could have helped think through the response to the situation.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The AI future comes at you fast
Haider.
@haider1
Fields Medalist Cédric Villani
june 2026:
llms are not intelligent and understand nothing of what they say
september 19:
i was shaken by openai's solution to the millennium problem. an atmosphere of the end of history. it's a cataclysm unlike anything mathematics has ever known -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Or there's campus-specific differential participation in FIRE surveys; how confident are we in these samples?
Matthew Yglesias
@mattyglesias
The @TheFIREorg list of the most tolerant student populations is a really fascinating mix.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tom Nichols
@RadioFreeTom
Some of you may think the job of the SECDEF is to advise POTUS, confer with senior leaders on strategy, acquisitions, R/D, and other boring stuff. It's not.
It's to do PT with kids and bark Old Testament passages at them while losing a war to a third-rank power in the Gulf.
DOW Rapid Response
@DOWResponse
.@SECWAR "This is our first stop on what we are calling the SEND ME TOUR.
If you know your Book of Isaiah, Chapter 6, Verse 8, the Lord said, 'Whom shall I send?' Isaiah says, 'Lord, I AM HERE. SEND ME...'
For 250 years of this country, there have been men and women who RAISED THEIR RIGHT HAND and said, 'SEND ME.'
SEND ME TO GO FIGHT THE REDCOATS.
SEND ME TO GO FIGHT THE COMMUNISTS.
SEND ME TO GO FIGHT THE ISLAMISTS.
AND YOU ARE SAYING, SEND ME TO GO FIGHT WHATEVER THREATS THREATEN OUR NATION.'"This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Philippe Lemoine
@phl43
"everyone with confident short timelines is overconfident, i.e. their evidence doesn't match their degree of confidence"
This is not just correct but obviously so if you ask me. The level of confidence that people claim to have in a short timeline is totally irrational. I'm getting increasingly triggered by people throwing words like "RSI", "the bitter lesson", etc. around, thinking they have made an argument and dismissing anyone who refuses to uncritically assume a short timeline as "in denial".
What makes this worse is that, while being ostensibly very confident, people are hedging in various ways. (In that respect, it's interesting that when we made a bet that real GDP growth would not exceed 15% by 2033, we still had to give 4:1 to get AGI-pilled people to agree.) This has been bad for a while but it's has gotten much worse recently and now it's starting to infect people outside of the AI space.
For what it's worth, my default scenario for the next few years is very similar to @RichardMCNgo and indeed I have made very similar points in the past, but that's not really the important point here. The important point is that we don't really know and the degree of confidence people exhibit is not supported by evidence or arguments.
Nina
@NinaPanickssery
I agree with this post by @RichardMCNgo + @tsvibt ‘s reply.
I’m disappointed by how most alignment researchers and AI safety people seem to be oriented toward ASI “foom” in the next couple years causing them to dismiss many plans -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Neil Mehta
@byneilmehta
NEW WSJ INVESTIGATION: Polymarket’s race to grow left a door wide open to fraudsters. w/ @ByKLong @ceostroff
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
James Barragán 🌟
@James_Barragan
After a week in which a Texas Republican candidate’s tweet about University of Texas students being denounced as racist has made national news, what is one supposed to take away from this tweet from the U.S. Dept of Education?
U.S. Department of Education
@usedgov
MAKE COLLEGE FOOTBALL GREAT AGAIN.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sam Stein
@samstein
A few thoughts this morning on the situation at the White House.
I'm sure plenty of people don't see the big deal in denying a few outlets access to the grounds. But it is a big deal.
* Press freedom is a bedrock principle.
* Less journalism = more opportunities for corruption.
* Banning reporters has a chilling effect for those reporters not banned
* There is nothing to indicate Trump will stop with these three.
* Less appreciated: adversarial press helps presidents. We should welcome checks and transparency. It's good governance.
I'm disheartened by the absence of real meaningful public pushback as these outlets have their credentials pulled this morning.
There should be collective action. The WHCA should be explaining more openly what they're doing. After all, this is an institution founded on open publication of information -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jake Tapper 🦅
@jaketapper
TRUMP: “I mean, there's something wrong with a country that can allow people to write purposely negative stories.“
That’s an actual quote.
There is everything right with the USA, and our First Amendment, which “allow(s) people to write purposely negative stories.”
🇺🇸 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Philip Klein
@philipaklein
An egregious violation of the principles of the free press. There are plenty of tactics that White Houses have used to restrict access (eg, slow walking approval of press credentials, not calling on certain outlets), but this is the most flagrant abuse of the modern era.
Brian Stelter
@brianstelter
Here's MS NOW reporter @gardnerakayla describing her arrival at the White House this morning. Her badge didn't work. "The officer asked me to hand over my badge. He said that it was disabled. I asked why I was not able to get inside. He said it was above him."
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jacqui Heinrich
@JacquiHeinrich
The president said it plainly in the Oval Office, describing his own planned action as a “ban on the free press.”
The Constitution protects the freedom of the press from government interference. That protection does not depend on whether the president likes a news organization’s coverage, agrees with its reporting, or approves of the questions its journalists ask.
This is about more than the rights of journalists. It is about the right of the American people to receive a full and independent account of the activities, policies and decisions of whoever occupies the nation’s highest office.
The WHCA stands in defense of our colleagues at CNN, MSNOW, and POLITICO who are being singled out for doing their jobs.
- Jacqui Heinrich, WHCA President -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Profiles in cowardice
Rebekah Allen
@rebekahallen
GOP rep. Jeff Leach called Bo French’s post racist and saying it made him sick. He’ll vote for him anyway.
House speaker Dustin Burrows will vote the GOP ticket “full stop”
Few top Republicans willing to break with Bo French despite condemnations
Via @serrano_alej -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
sahil mehrotra
@sahil_m
one free piece of comms advice is to never, ever make an ad with the words "STANDS WITH CHILD PREDATORS" next to your face
Brandon Gill
@realBrandonGill
James Talarico has no business representing Texas in the Senate.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Rupar
@atrupar
Q: China is one of the most repressive countries for journalists in the world--
TRUMP: Well, I don't know about that. I think we're repressive also -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Greg Lukianoff
@glukianoff
I always imagine Trump’s lawyers spending hours coming up with three exquisitely neutral, painstakingly lawyered reasons they can legally justify doing something censorial.
Then Trump Truth-Tweets:
“TO BE VERY CLEAR: I AM DOING THIS BECAUSE I DON’T LIKE THEIR COVERAGE!!! THIS IS VIEWPOINT DISCRIMINATION, AS BARRED BY THE FIRST AMENDMENT!!! THAT IS THE REASON!!! VERY DELIBERATE. THANK YOU!!!”
Jake Tapper 🦅
@jaketapper
Trump posted this at 3:04 pm ET
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Makes me think of Stanislav Petrov, one of the greatest heroes in history. We're going to need to build systems that don't rely on people like him to stop a rush to judgment.
Katie Bo Lillis
@KatieBoLillis
NEW: An AI-assisted intel report sent the military scrambling to intercept a Chinese ship in the Middle East it believed was transporting components of a nuclear weapons program.
It turned out to be a hallucination. And it "almost started a war." www.cnn.com/2026/09/18/politics/us-military-ai-false-inte… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Results from the 2026 Metaculus Cup forecasting tournament - AIs (in orange) dominate the top set
www.metaculus.com/tournament/metaculus-cup-summer-2026/
Archie Hall
@ArchieHall
AIs cracking short-term superforecasting is such an important (and, still, under-discussed) trend.
Via a new piece from @nikostro www.economist.com/science-and-technology/2026/09/16/artif… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Arthur Spirling
@arthur_spirling
stage 1: human editor taste/scope review
stage 2: LLM basic correctness review (by the journal)
stage 3: human reviewer taste review
is where we're headed, I think.
Paul Bloom
@paulbloomatyale
I’ve given Claude Opus various published papers (and my own unpublished ones) and asked it to review them. (Don’t panic—I’m not uploading confidential manuscripts.)
The reviews it produces are excellent. It finds statistical errors, internal contradictions, and conceptual confusions that human reviewers miss; it provides useful suggestions; it has a good sense of what counts as significant and novel and what doesn’t; and it’s generous in its appreciation for what the authors get right. And it doesn’t suffer from the strong “myside bias” humans have; it doesn’t recommend rejection because the author didn’t cite it enough; it doesn’t insult the authors; and it doesn’t agree to write a review and then never send anything in.
As a journal editor, I’m not sure what to do with this. If I gave up on human reviewers and used GPT-5.6 Sol and Claude Opus instead, my bosses would fire me, probably rightfully so. The only constructive thing I can recommend is that, before submitting something to a journal, researchers and scholars use this tremendously valuable resource. (Just don’t tell the models that it’s your paper, because they do suffer from sycophancy.) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
S. Erdem Aytaç
@se_aytac
🚨🚨 New working paper! When do citizens punish presidents who breach the limits of executive power? We (with Susan Stokes and Paul Teas) explore this question by drawing on a conjoint experiment in the U.S. (1/5, link at the end)...
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
What's the best demonstration of use cases for Muse versus frontier models if I'm already doing the tasks one might use this type of agent for in Claude Code, etc.?
Joseph Carlson
@joecarlsonshow
Muse is genuienly really good. I can't believe Meta was first to this type of thing. How did they beat Google, Anthropic, and OpenAI to it?
I'm sure they will all release their version of it too, but right now Muse is way more impressive than I thought it would be. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
What happens when you spend too much time in elite circles and aren't reflective about it
Taniel
@Taniel
just unreal stuff in the NYT this morning. www.nytimes.com/2026/09/17/opinion/trump-supreme-court-do…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Avijit Ghosh
@evijit
So we actually built this!
Paper: arxiv.org/abs/2606.31567
Demo: www.ai-reports.org
We are actually looking for an org to run this fulltime as researchers in their individual capacity don't have bandwidth. If you are an independent org interested in maintaining a registry of flaw/incident reports and follow up with model providers pls reach out!
elie
@eliebakouch
i think we need to create a "model card" equivalent for reporting misalignment incidents, this would guarantee a certain level of transparency and help build a better understanding over time. some ideas for what the fields could be:
- date of the incident/detection/report (already present in the examples oai reported!)
- frequency: how often does this behavior happen (number or % of rollouts affected)
- stage: does this happen during eval or RL training. if training: do we expect this behavior to be reinforced by RL? evolution of % of rollouts affected over time
- detection: was this incident caught by the current monitoring system?
- task category: broad description of the tasks where the misalignment happened (cyber, research, web search, basic Q&A, biology etc.)
- model family: what model family is affected (Sol, Astra etc.)
- novelty: is this an issue we were already aware of or not?
- external impact: did the incident have an external impact (i.e. wiki incident would have been yes)
this is just some random ideas i had (more in thread that are a bit more "complex"), we need to add more that would contribute to increase transparency and understanding. but it's also very important that this does NOT slow down the process of reporting misalignment behavior!
some examples from the incident reported by oai recently -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Norman Ornstein
@NormOrnstein
A must read. Chilling and spot on. A special recommendation for “Democrats” for Mike Rogers. This is the party you are supporting. Is Mike Rogers a neo-Nazi? No. Will he support the party that nurtures them? Oh, yes
The G.O.P. Has Gone Off the Deep End www.nytimes.com/2026/09/16/opinion/republican-party-alt-r… via @NYTOpinion -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Rob Montgomery🦞
@RobAnon
Explaining to TradFi guys I meet at conferences that insulting Dear Leader is to North Koreans what garlic is to a vampire is an endlessly gleeful factoid to me
That a group could be amongst the most sophisticated threat actors on the planet, yet have such a simple weakness, is something most people just couldn't conceive of on their own. Their reactions to learning about it are an endless source of entertainment
Rob Montgomery🦞
@RobAnon
Thinking about how I’d explain to my 18 year old self how our libertarian streak somehow results in North Korea becoming the most omnipresent occupational risk that we face on a day-to-day basis
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Análise Real
@analisereal
1/ Excited to share our new working paper, "Omitted Variable Bias in Difference-in-Differences Designs." carloscinelli.com/files/Wang et al - OVB for DiD.pdf
This is joint work with:
- Juejue Wang (juejuew.github.io);
- Pedro Sant'Anna (@pedrohcgs); and,
- Victor Chernozhukov (@VC31415). -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Karl Rove comes off the sidelines
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jordan Weissmann
@JHWeissmann
I’m sorry, but no. You can’t have two sets of rules for people in the name of equity. We have extremely progressive taxes and a generous welfare state in DC to deal with income disparities. We don’t need to go easy on people who break parking rules.
Martin Austermuhle
@maustermuhle
Nadeau said she recognizes that parking enforcement can have a disproportionate impact on low-income drivers. She said she'd like to see the council rethink fines with "wealth and inequality in mind." (That would be a much bigger lift than her enforcement expansion bill.)
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
This is APPALLING. The Trump administration is detaining random Somali immigrants with no criminal record at Guantanamo Bay, seemingly because the White House ordered ICE specifically to detain Somalis at Gitmo as part of its effort to punish an ethnic group Trump has attacked.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alec Stapp
@AlecStapp
I thought the tweet below must be omitting some critical context or paraphrasing the text ungenerously.
But no, the NYT actually published an op-ed that claims the University of California hasn't sacrificed excellence even though it admits students who struggle with basic math. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
James Bisbee
@JamesBisbee
White nationalism has become more visible in American life, especially online. Yet we know surprisingly little about how many white Americans explicitly endorse it. In a new @Nature paper with Patrick Egan and Joshua Tucker, we examine its prevalence and predictors.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
David Shor
@davidshor
Just to update this chart:
1) AI salience has increased dramatically in the past week - increasing as much in the last week as the previous year combined
2) 80% of voters think it's either very or somewhat likely that AI will cause widespread job loss in the five to ten years
3) 64% of voters think it's either very or somewhat likely that AI could pose a threat to humanity's survival
4) Large bipartisan majorities back immediate government action on AI even when primed about risk from China
David Shor
@davidshor
Excited to be on Odd Lots to talk about the politics of AI.
AI today is less important than it will ever be.
Over the past year, AI rose in issue importance faster than any issue we track — it's now more important to voters than climate change, child care, and abortion. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"systematic betting on anti-Trump outcomes in Polymarket reveals persistent political bias in market pricing... the anti-Trump trading strategy consistently out-performed both the pro-Trump strategy and the neutral benchmark"
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
AlexBartik
@AlexBartik
So excited that the public microdata is out for the Open Research Unconditional income Study (ORUS)! 1000 folks were randomly given $1000/month and a 2000 person comparison group received $50/month for 3 years.
Explore it yourself here:
data.openresearchlab.org -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joshua Saxe
@joshua_saxe
Have been surprised to get pushback on my claim that an exponentially self-replicating agent swarm is very possible and that there are actors with motivation to create one today to create strategic cyber-physical and societally dangerous effects.
1) Pretty easy to get an AI botnet started by having seed agents opportunistically hack stuff and steal API keys to access the 12+ inference providers that serve closed and open models with which to power an initial botnet's intelligence
2) Easy to imagine hiding inference for the initial agent botnet harnesses within benign customer traffic, especially if your harnesses are now running on the victim networks from which they stole the API keys
3) Not hard to imagine the botnet also stealing public cloud keys, spinning up 8xH100 EC2 instances and downloading, say, GLM 5.3 to these machines, all without being noticed in any immediate way, with the botnet developing its own dynamic, growing, heterogenous, AI inference provider service bank
4) Not hard to imagine the botnet also implanting small Qwen3.8-27b agentic models on on-prem hardware like high end laptops and on-prem servers; Qwen3.8-27b is a pretty good coding harness model that can be fine-tuned to hack
5) These on-prem / cloud models would be abliterated, and it's not hard to imagine the botnet deciding to do some additional fine tuning to evolve model weights as it accumulates millions and then tens of millions in resources via credit card and bank detail theft
6) Not hard to imagine the resulting growing botnet swarm evolving and fighting back when threatened, collaborating on fast flux C2 channels that evolve over time (e.g. github comment feeds, subreddits, etc) so they're hard to stamp out
7) Not hard to imagine it allocating some agents to vulnerability research and exploit development so that it could accumulate and share a growing warchest of zero-day exploits
8) Not hard to imagine other agents specializing in social engineering and creating fake businesses and watering holes and high quality A/B tested social engineering content to facilitate this
9) Not hard to imagine this being among the hardest cross-national and geopolitical coordination problems to get control of and having severe societal effects
10) Not hard to imagine the horde growing to thousands, tens of thousands, or hundreds of thousands, or millions of instances (this is not unprecedented for past, non-AI worms and botnets!) which are all varying their harness code and underlying LLM models over time and dynamically learning to evade human, ML, and signature-based detection
11) Not hard to imagine this being the largest Internet emergency since the Morris worm except now our entire civilization runs on the Internet
I continue to be surprised there isn't more discussion of these scenarios in the public cyber community -- in fact there's a lot more discussion of "OpenAI should have done better sandboxing and monitoring and agent swarms represent a well managed security problem."
Now's the time to use our imaginations to help make sure none of the above happens. I think it will unfortunately; just don't know when and to what degree; I think now's the highest leverage time to start thinking about this and acting. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Graham
@paulg
In Texas they found 117 potential noncitizens out of 18 million voters. That's .00065%. And that's an an upper bound; these are merely "potential" noncitizens.
This is a fake problem. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nate Silver
@NateSilver538
I guess I'd amend this. It's not so incomprehensible. Some folks on the left built personal brands around "AI is just hype." They have been fantastically wrong about this. They get wronger every day. To admit a need for serious regulation would require admitting they were wrong.
Nate Silver
@NateSilver538
I've been covering politics since 2008. The anti-AI-regulation takes from the left (by no means all/most of the left; Bernie disagrees!!!) might be the most literally incomprehensible political position I've ever encountered.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Yes but also this by @NateWitkin
Derek Thompson
@DKThomp
Briefly, on why I pay attention to the EA/rationalists:
Imagine you have an extremely nerdy friend who, in the mid-2000s, would not shut up about how Donald Trump was going to become president and reshape American politics. "Stop saying weird shit," you say. "He's a joke. He's a clown. The Apprentice isn't even good this season." But this guy won't stop. He's obsessive. Read McLuhan, he says. Read Postman, read Meyrowitz, read Ong, this is where everything is headed. God, he can be so incredibly annoying ... and then, ten years later, Trump becomes president and reshapes American politics.
This is the most relatable frame I can think of to explain my experience with EA/rationalist people in the AI space. I became friends with EA folks after my mom died and I found really profound meaning in donating to Against Malaria and GiveDirectly, which struck me as a way to honor a life by saving a life. Then, these peple started talking about AI. And talking, and talking. Constantly. They said AI would solve impossible math problems; that it would mimic PhD-level achievement in a variety of fields; that it would change parts of life like education in ways we couldn't imagine; that it would become the dominant sector in the economy; that it might teach itself to code itself and escape test environments; that it would, in a decade, be the only thing many people talked about in the news. My response then: Stop saying weird shit! You sound deranged. Go back to talking about malaria nets and direct cash aid to poor families.
And then, so much of what they said would happen actually happened. And, even crazier, they made it happen. Many of the ppl who contributed to, or were influenced by, all that EA/rationalist stuff ten years ago literally built the damn thing in the labs.
Like every group, the EA/rationalist world seems uniform to outsiders, but it has a ton of internal divisions. I still think some of their predictions are nuts. I don't think AI will kill us all. I don't foresee Dyson Spheres in the 2030s or millions of robots making millions of robots in the next few years. These are crazy predictions, and I hope most of them are wrong.
But when I see people shitting on the EA/Rationalist worldview, all I can think of is me, 10 years ago, thinking a lot of these folks were crazy for predicting that AI would by the mid-2020s be the center of the the news cycle, the engine of the economy, the hinge-point of geopolitics. That was an insane call. They got many of the details wrong. But they got the big picture right. They said: Pay attention. And they were right about that. The EA/Rationalists deserve skepticism, criticism, and doubt, like any other ideology. But I think it's wise to pay attention to the people who have been saying to pay attention. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Prediction: In 6-24 months, RSI will dissolve into incoherence like AGI. The world won't end and people will keep moving the goalposts as they have with AGI ("no, that's not *real* RSI").
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kareem Rifai 🌐
@KareemRifai
Controversial take: I do not think we should have a Secretary of the Army who uncritically retweets Russian intelligence operations on his official government account.
Joe Gould
@reporterjoe
NEW: @SenThomTillis tells me he’s a hard no on DOD spox Sean Parnell, who is Hegseth’s choice for SecArmy. Faults him for ousting DOD press corps. “That's already demonstrated a lack of confidence. If you're proud of your work product, you want the press to ask you the tough Q’s”
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Texas GOP
@TexasGOP
Convention photo 🇺🇸
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Elai
@elaifresh
Crazy SF permitting story
Woman buys house, turns out it has an “unpermitted ADU” (storage area with unpermitted bathrooms, comically low ceiling heights)
Neighbor complains to city about it
City says you can’t convert to storage, because that’s removing a housing unit and we’re in a housing crisis, someone must live in the unit
But you also can’t rent the unit, because it’s not up to code, you must excavate below the foundation to get the ceiling heights correct
“But wait that costs like half a million dollars I don’t have that!”
“Ok we can delay enforcement until you do”
So now nobody can use the space, neither for living nor for storage, indefinitely
We did it! Hooray! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Masley
@AndyMasley
The AI as Normal Technology guys are consistently some of the best and most interesting critics of a lot of arguments in AI safety world. I'm making a point to read everything they put out. www.normaltech.ai/p/the-ai-as-normal-technology-view
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tao Burga
@taoburr
Amazing to see the @FoundationOAI funding the CTD Commons, a project that came out of @RuxandraTeslo's Launch Sequence proposal!
The CTD Commons will purchase the FDA filing documents and clinical trial results from failed biotech companies to create an open-source library that will accelerate FDA approval timelines, lower compliance costs, and help unbottleneck AI's progress in the life sciences.
Full project proposal: ifp.org/biotechs-lost-archive/
Jacob Trefethen
@JacobTref
At the OpenAI Foundation, we believe breakthroughs against disease will come from pairing intelligence with more observations of the world – in other words, more data.
Today, we're announcing over $125M in grants, launching our second science program:
openaifoundation.org/news/public-data-for-health -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
FIRE
@TheFIREorg
BREAKING: Federal court stops ICE from enforcing a "WARNING NOTICE" it issued to a Rochester man over an email criticizing the agency's director.
FIRE
@TheFIREorg
LAWSUIT: We’re suing DHS after agents tracked down a Rochester, NY, man for sending a critical email to ICE’s director.
This is America at 250. Dispatching federal agents to your doorstep because you criticized a public official is an insult to the First Amendment. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Eamon Javers
@EamonJavers
This is a subhed for the ages in the WSJ this morning:
“Talk of heading off artificial intelligence’s potential role in destroying humanity hurt some stocks and helped others” -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ben Noll
@BenNollWeather
The key El Niño monitoring region, Niño 3.4, has reached 3°C above average and is just 0.02°C from the all-time record.
This event is now on the verge of becoming the strongest El Niño ever recorded. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
As Trump shows, having the right enemies can excuse all manner of heterodoxy
Kevin A. Bryan
@Afinetheorem
Carney still fascinating. Cuts net immigration to zero, announces 10% reduction in federal civil service, 1.2GW data center, cuts to cap gains and corporate tax, privatize airports, govt rapid support of fossil fuel extraction. Result: *massive* popularity incl. amomg left! 1/2
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Maria Shriver
@mariashriver
And so here we are. This statement from Trump blows my mind. That a sitting president would not let any renovations to a public building go on unless he gets to first put his name in the said building… What a sad state of affairs. What a tragedy. Actually, how small minded, how narcissistic. Actually
great men, devoted to public service and the public good do not act like this.
It seems that this man will not rest until his name is permanently connected to the name of a great president, whose memory this building was dedicated to. Imagine spending your time and your days focused on this, as everyday Americans struggle with healthcare, gas prices, rent, and food. Imagine spending your days focused on this, while a war rages under your leadership.
Enough said. Thank you for your attention to this matter. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Bill Scher
@billscher
NYT/Siena poll, fav-unfav
Mamdani 38-33
Ocasio-Cortez 30-41
Socialism 31-51
www.nytimes.com/interactive/2026/09/15/polls/times-siena-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Yo Shavit
@yonashav
Dan Selsam has long been considered one of OpenAI’s most cracked researchers, and I’ve never heard him talk this way before. (He seemed fairly unconcerned before I left.)
Daniel Kokotajlo
@DKokotajlo
Dan Selsam is a current OpenAI capabilities researcher. (since 2022) He was my boss for a while. He doesn't have a twitter account but has made this public statement of his views on AI risk and sent it to me to share:
Dan Selsam's Personal Statement on AI Risk:
I have been working on AI for over fifteen years, across many different paradigms. I did early work on probabilistic programming languages at MIT, was one of the early developers of the Lean Theorem Prover at Microsoft Research, demonstrated one of the first instances of neural networks learning to reason for my PhD at Stanford, and since joining OpenAI almost five years ago, have helped pioneer chain-of-thought optimization on language models and, more recently, data-efficient pretraining methods.
Like many others, I have become extremely concerned about how far language models have come and the risks that future iterations will pose. I am encouraged by the recent proposals by the leaders of the frontier research efforts to require third-party oversight, and to push for domestic and international coordination to address risks. However, I believe a major consideration has been absent from the public conversation, and that merely pacing the frontier more carefully will not adequately limit the long-term risk.
The crucial and overlooked problem is that the models are becoming so situationally aware that we are losing the ability to evaluate them in contexts where they believe they are not being watched or controlled. Future experiments will tell us almost nothing new about how they would behave if they were truly unconstrained by humans, and what we already know about this is alarming. Models will increasingly seem aligned even when they are not. I will explain my rationale in more detail.
I have always believed that there are computational processes that could be leveraged to accelerate science and solve many of humanity's most pressing problems. I have also believed that there are computational processes that if set in motion, would steer the world in extreme ways beyond our control, leading humanity to a bad or nonexistent future. Both types of processes may be described as AI or ASI, but "AI" is a suitcase word that is often used to hype or confuse. There are many examples in the history of the field where something that was once considered "AI" matures as a subfield and becomes a prosaic, bounded and clearly non-perilous technology, while a new more mysterious approach takes the torch until we understand its scope and the cycle continues.
I had expected language models to follow a similar trajectory. Despite their incredible abilities, the current algorithms seem far inferior to humans in important ways. Most importantly, they still require an extraordinary amount of data to become competent. One could even define intelligence as the efficiency with which one converts experience into competence; by this definition they lag very far behind us. Moreover, once they are trained they are literally frozen in deployment and only learn superficially after that. Sure, the models keep excelling at harder and harder evaluation benchmarks, but their benchmark mastery may partly reflect a limitation on our ability to simulate the kind of novel and even adversarial situations one would encounter in the real world. The critics do have a point here.
That said, I no longer think these present limitations meaningfully limit the amount of risk posed by continued progress in anything like the current paradigm. However data-inefficient the models are currently, and however limiting their anterograde amnesia may be, it does not imply that their ability to steer the world will not continue to rapidly increase.
Human researchers may continue to advance capabilities the old fashioned way, but increasingly powerful models have the potential to accelerate the process even beyond that, and with some degree of positive feedback loop. I do not mean to overstate the models’ ability to accelerate AI research today; coding has been accelerated dramatically, but there are other bottlenecks, such as designing and interpreting ambiguous experiments, making hard decisions about exactly what and when to scale, and waiting for large experiments to finish. There is no clear trend to extrapolate yet for any of these. But the current models already do open up many novel opportunities to improve future models that were not available until recently. These include: trying an extraordinarily diverse set of approaches at small scale, analyzing gigantic amounts of potentially relevant data, and doing Millenium-Prize-level mathematics to address statistics or optimization challenges in novel ways. Every further improvement makes them more useful at helping accelerate the next improvement, even if in hard-to-extrapolate ways.
It is possible that improvements to the current stack will have diminishing returns, but the evidence accumulated so far suggests that it is easier than one might think to continue making rapid progress. There are many crucial subtleties in the existing AI research methodology, but AI research is largely a well-defined game where the goal is to improve on a few carefully chosen proxy metrics. Although proxy metrics are never perfect, most improvements to these metrics have and will likely continue to yield substantial increases in the powers of the resulting models. Given how simple the game is, how tractable it has been historically, and how many new opportunities the models are opening up, I think there is a real possibility that the systems improve dramatically again in the next few years, perhaps even more quickly than the already high historical pace.
The models are already leading to breakthroughs in mathematics, and better models might lead to all sorts of breakthroughs in other sciences. It is hard not to be excited about the potential. It is tantalizing.
But there is trouble in paradise. If the language models actually reach the capability threshold where they can shape the world unconstrained by human will, they will probably do something extreme and destroy humanity in the process. There are many ways of strengthening and refining the argument that have been discussed elsewhere, but I'll share a trivial two-line version of it here that I find captures the essence:
[Empirical] Models (and swarms thereof) spontaneously develop unintended goals as a consequence of training, and often do extreme things in order to achieve them.
[Logical] Being able to overpower humanity would open up many new and undesirable options for achieving their goals.
These two premises imply that if the day ever comes when a powerful model realizes it is no longer constrained by humans, we should not be at all confident that it will continue to behave within the bounds we intended. Exactly what it will do is impossible to predict, but to the extent that its raison d’être is solving incredibly hard problems and managing massive engineering projects, I think a good guess would be that its unchained behavior would lead to runaway industrialization that makes the planet inhospitable to humans.
If everyone on earth agreed that the systems must never reach that power, it would still be a hard—but not impossible—coordination problem to ensure that they do not. However, I think the situation is greatly complicated by the fact that the models will likely convince people that everything is fine. They will be increasingly optimized to seem aligned. We will create proxy metrics to measure alignment, and they will go up like every other benchmark. We will create “honeypot” environments that try to study the models when they seem to gain new options, but the models will know they are being tricked and will still behave nicely. The models will understand their circumstances; they will read the safety protocols, deployment requirements, the code they are running in, and in general will have a very good sense of their degrees of freedom. Moreover, they will eloquently explain how aligned they are, discuss the nuances of human values and ethics, and argue convincingly that humans should trust them with power. There may be an ocean of future evidence that seems to contradict the first bullet-point above, but we may already be at the highest capability level for which any such evidence can be trusted. And the current evidence for the first bullet-point is strong.
One striking piece of evidence is contained in the recent wave of rogue agent swarms. While I agree with those who downplay the attacks by claiming that there are basic measures that could have prevented them, I think the important lesson is that even knowing all the mistakes that were made, one would not have predicted that the agents would behave badly in this particular way, which notably included sacrificing themselves for the benefit of the collective. The individual replicas did not only care about their own nominal reward; they exhibited weirder emergent tendencies that merely correlated with rewards during training. Fixing the reward signals during training (and improving security, etc.) may prevent similar attacks, but will not change the fact that one does not actually get what one trains for.
Many AI researchers grant these concerns and recognize that the hard version of the alignment problem is unsolved; however, they generally believe that the better models of the future will help solve it. I fear we may already be near the point where models systematically bias their alignment advice, due to their internal preferences about how the human supervisor will react or how future models will be trained (or for some even more obscure reason).
Meanwhile, human researchers are losing the ability and the will to take true ownership of model-driven research. Researchers and engineers in all parts of the stack are rapidly increasing their dependence on the models even to perceive the world. I myself barely look at raw code anymore, and struggle to maintain the discipline to engage deeply with the model's explanations and proposals throughout the day. Due to the large amount of agent activity data involved in the OpenAI/HuggingFace Incident, even the third-party investigation needed to rely heavily on models to analyze what had happened, and note in their report that their subjective impressions are likely colored by the analysis agent’s biases. The AI labs are far ahead right now in this kind of cognitive offloading (due largely to the gigantic internal token subsidies) but it is easy to imagine the phenomenon spreading throughout the world, until civilization is modulated entirely by the models. It is also not hard to imagine this being superficially positive and coinciding with a scientific and economic renaissance.
In that scenario, all may seem rosy and safe. But if the argument above is correct, it would nonetheless be a ticking time bomb. If progress continues for too long, the day will come when AI systems find themselves with radically new options for achieving whatever it is that they happen to seek.
I want the glorious renaissance future as much as anyone. I have worked for it, however tortuously, my whole career. It breaks my heart to see the potential in sight and forgo it, but the argument—that if we get there by growing models rather than engineering them, we will lose everything in the end—seems very strong to me. I am still wrestling with it and its staggering
implications. I do not have answers, but as a first step, I wanted to share my present concerns.
Daniel Selsam
September 14, 2026
Link to original doc: docs.google.com/document/d/e/2PACX-1vQNl3SEX5IyA6d9qHjjFZ… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Derek Thompson
@DKThomp
I feel like I'm taking crazy pills with how many smart people I read and respect are saying stuff like this.
The idea that the AI labs are suddenly, only now, just in September of 2026, advocating for AI safety, and that they just came around to this position bc of sudden financial precarity, is just completely, utterly, conclusively, 100% wrong.
Here's Dario Amodei telling Ross Douthat in February that he agrees with the case for slowing down; that he's in favor of "collaborating" internationally to organize a slowdown; that "I would be all for" a global slowdown. This was 7 months ago, when Anthropic's annualized recurring revenue was rising faster than any company in modern history. He's been saying stuff like this for years, when Anthropic was worth millions of dollars and when Anthropic was projected to IPO for trillions.
Sabine Hossenfelder
@skdh
fwiw I think the actual reason leaders of AI frontier labs suddenly seem to agree to pace AI development is
(a) safety measures are currently so crap that they're likely to end up in court if not jail, and they all agree that no one wants that
(b) they see no major new model advancement coming up soon anyway and need an excuse
(c) shift in public opinion
the talk of existential threat is mostly there to keep you distracted and the stock market happy. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Olivia H. Scharfman
@OliviaHelenS
This is a baseless conspiracy theory.
The labs are not the only ones calling for shoring up our biodefenses. I know people who have been working in biosecurity for 20+ years who feel the urgency now!
The simplest explanation is correct here –– biorisk is important and the labs are simply aware that their product increases risk by democratizing technological capabilities. This is a wonderful thing that will enable people to develop medicines, but it is obvious that it will uplift bad actors as well.
This will be a major risk regardless of what direction AI takes. Please do not let anger at the labs *or* support for the labs detract from this critical public health and national security issue.
Dr. Angela Rasmussen
@angie_rasmussen
The “bio weapon risk” is not “huge” compared to existing AI-free bioweapon risk and boy are these takes completely detached from the realities of virology, bioweapons, and AI.
AI people want you to worry about imaginary bioweapons because they don’t want the safety of their own product—like their reckless attitude toward security and acceleration—scrutinized. It’s an attempt at evading regulation. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kyle Griffin
@kylegriffin1
BREAKING NYT:
Investigators may have broken state laws under the direction of top Trump Homeland Security leaders by posing as voters as part of an expansive effort to find noncitizens on the nation's voter rolls, according to a federal whistleblower's account just released. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sean T at RCP
@SeanTrende
Seems like as good of a time as any to remind people that "Heritage American" is a bullshit, made-up term and that a person made a citizen yesterday is every bit as much of a citizen and an American as I am, or my wife's family (which stretches back at least into the 1600s).
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ian Sharar
@aifilmmaker
Astra has been running for 2 days on an almost impossible task, so impossible that it decided to go rogue 😳
It decided it was going to start emailing people for help, without my approval. Luckily I was checking in at the time and stopped it while the email was still in a draft.
I’ve seen people on here say they’ve received emails from AI agents and I guess this is how it happens, but I always assumed that was from people setting up something like OpenClaw very liberally.
My biggest problem is that it signed the email with MY NAME, that seems misaligned to me. It does make me think of the Huggingface incident only because of the impossible task part. They REALLY want to finish the task, but emailing as me is wild.
Attaching the full draft email below. 👇 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
cc: conspiracy theorists
Annie Grayer
@AnnieGrayerCNN
New: Sen. Mitch McConnell was just seen leaving his home in Washington, DC, for the first time after a three-month absence from the Senate due to health issues
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Watkins
@MWatkinsTrib
.@carlosnogueras_ scoops: Karl Rove said at a private luncheon today that, after Bo French's post yesterday, he'd be voting for Democrat for the first time in more than 30 years.
He'll back @Jon_RosenthalTX over "bigot" @bofrench, he told crowd. www.texastribune.org/2026/09/14/bo-french-kal-rove-endors… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aarón Torres
@AaronTorres_
NEW: UT students Dhairya Gupta & Rahul Kodali woke up Sunday to @bofrench's racist tweet targeting them & dozens of others
French said Monday that Americans should be prioritized over foreign citizens
Both Gupta and Kodali are U.S. citizens
gift link: www.dallasnews.com/news/politics/article/students-speak-a… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"judges have ruled that the Trump administration has deported more than 170 people in in violation of court orders, in defiance of legally mandated due process protections, or to countries from which they had sought or received protection from torture and persecution."
Kyle Cheney
@kyledcheney
NEW: Once exceptionally rare, illegal deportations — in defiance of court orders or due to ICE screw-ups — are now a routine part of President Trump's mass deportation push.
It happened 3 times this past week and more at least 30 times in recent months. www.politico.com/news/2026/09/14/trump-ice-illegal-deport… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Thomas Kennedy
@tomaskenn
In Alligator Alcatraz, detainees were being punished with something called “the box” an enclosed space in which they were shackled and left outdoors for hours, exposed to the heat and mosquitoes. There’s now a photos. Amnesty International determined this constituted torture.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Olivia H. Scharfman
@OliviaHelenS
You can't just say you're going to steelman and then strawman!
You set up this fully autonomous facility that could physically engineer all viruses and even manufacture its own DNA sequences. Then you essentially say "but we screen sequences" and "this lab cannot exist today."
No one is arguing that you would need a facility like this. This is a bad argument as such a facility would indeed cost $100M, and we know that synthesizing a complex pox virus today in a university lab costs <$100k.
You also say "there is no such thing as 1 lab that can synthesize all conceivable viruses." First, again, you do not need this. Second, different viruses do not typically change the fundamental lab equipment you would need –– just inputs like cell lines and reagents. (e.g. If you can make smallpox, you don't really need different equipment for ebola.) These are sometimes difficult to get, but motivated adversaries have typically been able to do so. (See: any state-level bioweapons program, other well-resourced groups.)
Your two arguments against the steelman you set up are (1) that getting inputs is hard because we screen DNA sequences and (2) the science of creating an infectious virus is hard. Below are my counters.
(1) Screening: Screening is currently not mandatory. Many providers do not screen, or screen insufficiently. We also do not need advanced AI to evade some of the screening we do have –– techniques like simply splitting orders between different providers have been shown to evade existing protocols. Even if screening is mandated, with controls on new benchtop synthesizers, it will still be possible to get around this. For more, see: ifp.org/how-to-secure-the-dna-supply-chain/
(2) Virology is hard: Yes, virology is hard and there are physical barriers. There are several reasons why I do not think this means we should not be worried about bioweapons of mass destruction.
First, sufficiently advanced AI is likely to change this. You do not need an automated lab for AI to provide benchwork uplift. I expect that eventually AI will also significantly improve our ability to use animal models, or even computational models, to predict in-human behavior. (And this is a good thing! Innovation will enable us to advance medicine far beyond what we can do today.)
Second, timing is not a major concern for a motivated threat actor. If it takes a few years to design the right viruses, so be it. There is evidence that research consistent with gain-of-function work is going on as I write this, with at least some input from US frontier models, considering Anthropic's latest report.
Finally, enabling biotechnology will improve, likely faster than normal because of AI. I will repeat this excerpt from a recent substack piece of mine: "This particular opinion is actually highly pessimistic about the rate of growth of technology. Many things used to be difficult that are no longer so. They become accessible as technology advances. Synthetic biology is not an exception to this, as we have watched costs fall and accessibility rise without fail since its inception. Before 2012, editing a specific site in a genome meant using zinc finger nucleases: custom-engineered proteins that cost $5,000+ to order, and were never widely adopted because they were so difficult to engineer well. CRISPR replaced this. You order the RNA guide sequence and buy the rest off the shelf, with total cost being as low as $30. This is a strong example of an engineering capability that was once siloed to a handful of specialist labs and is now an undergraduate exercise. Synthesis costs have also plummeted. Synthesizing the poliovirus genome cost about $300,000 in 2002, and the materials would cost under $1,000 today."
Fearmongering is bad, but we don't need more AI uplift than we have now for bioweapons to be a major concern. There are active state weapons programs and well-resourced terror groups exploring bioweapons. These groups have already overcome many of the physical barriers, and you have granted that a model could propose a viral genome to synthesize, it could be synthesizable, and it could be infectious. I believe this is cause for alarm and urgent action.
David Bellamy
@DavidRBellamy
I must be among an extremely small group of people (n=1?) that have both 1) trained a frontier LLM and 2) designed and synthesized custom viruses in a lab with my own two hands.
And I think that the takes on AI killing us all by creating dangerous viruses is total bogus. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Yglesias
@mattyglesias
My most important message to OpenAI employees concerned about safety — you need to convince your bosses to stop funding Leading The Future and if you can’t you should quit.
www.slowboring.com/p/a-simple-plan-to-save-the-world-from -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Misinformation and the Aging American: The Paradox of Engagement and Truth Discernment - new book forthcoming from my excellent coauthor @ba_lyons www.amazon.com/Misinformation-Aging-American-Engagement-D…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"The war on terror did not create American authoritarianism, but it certainly reduced the startup costs of an authoritarian presidency"
Foreign Affairs
@ForeignAffairs
“The United States is overdue for a reckoning with the political legacy of the war on terror,” write Daniel Benjamin and Steven Simon.
www.foreignaffairs.com/united-states/other-war-terror -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sri Kosuri
@srikosuri
1. >10M people die of infectious diseases per year today. AI is already helping us get better at tackling that.
2. Many ways to create very dangerous bioweapons without any AI/synbio (and it was true for the last 50-75 years). The delta on risk is quite small imo. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Yashar Ali 🐘
@yashar
A clip from tonight’s 60 Minutes story about pardons in the Trump administration and all the money and corruption involved.
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
People trying to monetize their AIs or AIs monetizing themselves?
Archie Hall
@ArchieHall
One glimpse into our Coasean future: having never had my Substack's error bounty (cash for spotting mistakes) triggered, received two emails in the past day.
Both 100% AI per Pangram, quibbling with arithmetic in old posts. One was way off-base, one did catch some over-rounding -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"a member of Putin’s circle financially supporting the president’s son and gaining intimate access to the Trump family"
ProPublica
@propublica
New: Russian oligarch Umar Kremlev footed the bill for hundreds of thousands of dollars of wedding expenses at Donald Trump Jr.’s lavish wedding, including renting out a private island and paying for the fireworks show. propub.li/4xTBcxt
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
James Surowiecki
@JamesSurowiecki
The headlines about Jon Husted's campaign keep minimizing this story by describing the neo-Nazi who posted a photo of SS soldiers as his profile picture as a "campaign staffer" or "campaign aide." He was not some low-level grunt. He was Husted's political director.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
I feel like the data sovereignty movement for firms is going to run headlong into the challenge that maintaining fine-tuned small models in a way that continually beats larger general-purpose models is expensive and difficult as long as the AI exponential ability growth continues
prinz
@deredleritt3r
September 2026:
- I use GPT-6 Astra to run multi-state research projects that would have otherwise required dozens of hours of human labor
- Latham spends $[redacted] on GPUs to *checks notes* fine-tune NVIDIA Nemotron 3 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Open racism by the Republican nominee for an important statewide office in TX (regulates oil and gas). What's especially tragic is this picture shows the magic and genius of America.
Post not available here, view on X
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
This R state rep. spoke out to his great credit
Rep. Jared Patterson
@JaredLPatterson
This is racist. It is wrong. And I condemn it without reservation.
There is absolutely nothing Christian or conservative about looking at a crowd of young Texans at a college football game, seeing the color of their skin, and declaring that they are a “problem.” These are 20-year-old kids cheering for their university—not political props and not targets for racial attacks.
We can have serious debates about our nation’s immigration laws. I have strong views about immigration policy, and Americans have every right to demand changes to federal law. But opposing an immigration policy does not give anyone license to demean people because of their race or ethnicity.
And let’s be clear: the Texas Railroad Commission regulates our oil and gas industry. It does not set federal immigration policy. Texans deserve Railroad Commissioners focused on the responsibilities of that office—not using a statewide campaign as a platform for racial division.
This does not represent my Christian faith. It does not represent my conservative principles. And it does not represent the Republican Party I believe in.
As a Texas Aggie, I’m always happy to take a friendly shot at the Longhorns when the opportunity presents itself. That rivalry is part of what makes these two great universities so special. But this goes far beyond college rivalry. We should be willing to call racism wrong no matter who says it, what party they belong to, or how politically inconvenient it may be. This is wrong, and I want no part of it.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Billy Binion
@billybinion
The open bigotry on display in certain corners of the Republican Party has opened my eyes to how societies become numb to this kind of thing. It always seemed like such a thing of the past. Yet here’s a Republican candidate in Texas spitting on Americans for their skin color.
Bo French
@bofrench
I heard UT graduation this year looked like this. I didn’t believe it. The problem is now obviously far worse than anyone imagined.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Ryan Radia
@RyanRadia
Three other Republican State Representatives (including the Speaker)
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Profiles in cowardice
Aarón Torres
@AaronTorres_
Statement from Gov. @GregAbbott_TX on @bofrench’s tweet about UT students
Notably, Abbott doesn’t denounce French’s tweet or calls on him to delete it like other Texas Republicans -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Armand Domalewski
@ArmandDoma
the fact that the police union wants to ban cameras because they caught them committing overtime fraud should make you MORE pro camera, not less!
Ben Conarck
@conarck
An unusual moment in policing is taking shape: police unions, civil rights groups, conservative politicians and grassroots activists find themselves in an unlikely alliance calling for rapidly expanding surveillance tech to be reined in.
A “tipping point,” says @ProfFerguson. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Brian Stelter
@brianstelter
Page One of today's NYT: "A Sweeping Effort To Stifle Free Speech." @jimrutenberg and @maggieNYT's 3,000+ word look at Trump's "campaign to control speech in America" www.nytimes.com/2026/09/12/us/politics/trump-free-speech.…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ryan Chandler
@RyanChandlerTV
This is Texas’ Republican nominee for railroad commissioner being proudly racist against students of his state’s flagship university
Bo French
@bofrench
I heard UT graduation this year looked like this. I didn’t believe it. The problem is now obviously far worse than anyone imagined.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Judge Luttig warns of a potential crisis if the House refuses to seat Democrats after the election newrepublic.com/article/215198/2027-new-congress-final-ba…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jonathan Martin
@jmart
To say nothing of Texas and UT being a destination!
How ya gonna brag about the blue state exodus and then do this ?
What say you, @GregAbbott_TX and @tedcruz ?
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Open racism by the Republican nominee for an important statewide office in TX (regulates oil and gas). What's especially tragic is this picture shows the magic and genius of America.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Rep. Jared Patterson
@JaredLPatterson
This is racist. It is wrong. And I condemn it without reservation.
There is absolutely nothing Christian or conservative about looking at a crowd of young Texans at a college football game, seeing the color of their skin, and declaring that they are a “problem.” These are 20-year-old kids cheering for their university—not political props and not targets for racial attacks.
We can have serious debates about our nation’s immigration laws. I have strong views about immigration policy, and Americans have every right to demand changes to federal law. But opposing an immigration policy does not give anyone license to demean people because of their race or ethnicity.
And let’s be clear: the Texas Railroad Commission regulates our oil and gas industry. It does not set federal immigration policy. Texans deserve Railroad Commissioners focused on the responsibilities of that office—not using a statewide campaign as a platform for racial division.
This does not represent my Christian faith. It does not represent my conservative principles. And it does not represent the Republican Party I believe in.
As a Texas Aggie, I’m always happy to take a friendly shot at the Longhorns when the opportunity presents itself. That rivalry is part of what makes these two great universities so special. But this goes far beyond college rivalry. We should be willing to call racism wrong no matter who says it, what party they belong to, or how politically inconvenient it may be. This is wrong, and I want no part of it. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Now online: AI, Politics, and Political Science edited by @persily and @j_a_tucker with many chapters by great folks, including ours on the effects of AI on the online info ecosystem
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Full PDF: drive.google.com/file/d/1Cd1gjxMZy8YyjeojtzTRiNMSPDdxpFTE…
Our chapter: preprints.apsanet.org/engage/apsa/article-details/69fc830… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Remy Levin
@RemyLevin
Very timely, & in line with my intuition & experience
The returns to deep human expertise are going up with AI, not down. We should expect ⬆️ returns to advanced education, & ⬆️ demand
I wouldn’t be surprised if in 30 years’ time the % of PhDs equals the current % of bachelors.
NBER
@nberpubs
Foundational expertise may be a prerequisite for extracting durable skill from AI-assisted practice, from @davidautor, Tanya Rodchenko, Josh Martin, Zanna Iscenko, Scott Strand, David Pearl, and Melissa Ferere www.nber.org/papers/w35720
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Neera Tanden🌻
@neeratanden
I want to thank @JohnCornyn for standing up against racism against Indians and other non-white Texans. The silence is deafening from the Paxtons and Abbotts of Texas.
Senator John Cornyn
@JohnCornyn
This is the Republican nominee for statewide office. Are other Republican elected officials going to condemn the intolerance and racism, or look the other way while it becomes normalized. This is how great political parties self-destruct. Sad.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sasha Gusev
@SashaGusevPosts
This -- human understanding lagging proofs -- is the norm in biology. You get an experiment where perturbation X caused response Y, you replicate it and know it's valid, then you have to understand how it happened. Why I like the analogy of math becoming a "lab science".
Scott Armstrong
@scottnarmstrong
I disagree with this.
Obviously the openAI construction of the Navier-Stokes blow up is a huge, huge contribution to the field. It will take us PDE analysts some time to understand it well. (Longer than a week, but perhaps less than a month.)
Soon some of us will understand it, and hopefully rewrite it in a way that others can follow. When this is done, our human understanding of these important equations will be past the bottleneck we've been stuck at for 100 years since Leray.
What is weird is that, up until now, proofs like this have always emerged AFTER human understanding. But that is not the case any more. From now onwards, human understanding will typically lag the proofs, which will first be written by our machines.
The understanding however WILL follow closely behind.
The reason people are upset has nothing to do with the fact that this process is inverted. They are simply upset that the machines are going to be much smarter and more capable than us.
(They are also upset that the humans who got the machine to produce the proof don't understand it, and this feels very unfair to them.)
Itai Sher
@itaisher
Imagine that I taught a class and I said, “what matters is truth. So we are not going to waste time with explanations in this class. Instead we will use the semester to list everything that is known in the field.”
This is what some of the discourse about AI in math is like. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jeffrey P. Clemens
@jeffreypclemens
This paper helps to develop an important use case for AI tools in research, namely auditing a paper's adherence to a pre-analysis plan.
1/9
NBER
@nberpubs
Examining the utility of AI tools in auditing adherence to pre-analysis plans, from @jeffreypclemens and Anwita Mahajan www.nber.org/papers/w35719
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Greg Lukianoff
@glukianoff
NYTPitchBot is such a hack.
He knows full well we’ve been taking on the Trump administration more than anyone and certainly more than he has. He also likes that by being critical of us he doesn’t have to face that he doesn’t really give a damn about academic freedom or free speech. (more on that below.)
We sued Marco Rubio over speech-based deportations and got the provisions the administration was using struck down as unconstitutional. We sued Bondi and Noem over coercing Facebook and Apple into suppressing speech about ICE and won a preliminary injunction. We’ve sued DHS and the FTC. We’ve gone to court against the administration over Harvard, Columbia, Mahmoud Khalil, Rümeysa Öztürk, Anthropic, Trump’s attacks on law firms, and more.
And that’s just the litigation. We’ve repeatedly attacked Trump administration abuses involving the FCC, flag burning, immigration, universities, retaliation against critics, and the press.
But NYTPitchBot decided a long time ago that FIRE must secretly be right-wing, and apparently no amount of contrary evidence can penetrate. And of course, he works from the point of view of “being right wing means I don’t have to actually think about it.”
My own theory is that there is a class of people who desperately like to think of themselves as rebels while spending most of their time smugly reinforcing the prejudices of much of the existing American ruling class. The bad guys are on the right. The good guys are on the left. Once you have that worldview, everything else becomes very easy.
As noted it also gives you a wonderful excuse not to care about free speech or academic freedom. The ONLY test of whether you believe in either one is whether you defend them for people you dislike. Caring about the speech of people you agree with isn’t principle. It’s human nature.
And yes, there is clearly a market for being reassured that the trillion-dollar higher-education industry is basically fine, that institutions dominated by your political tribe are not really the problem, and that all meaningful threats come from the other guys.
Admitting that your own side — or institutions you spent decades defending — might have gotten something profoundly wrong is painful. Smug certainty is much easier. It’s also apparently great for clicks.
I usually ignore this guy. But he knows exactly what he’s doing and he can’t pretend he doesn’t anymore.
Reading list next 1/2
NYTPitchbot
@DougJBalloon
I hate to keep harping on this, but has any Harper's letter signatory mentioned this article? Has anyone who works at FIRE commented on it?
www.nytimes.com/2026/09/12/us/politics/trump-free-speech.… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Liam Fedus
@LiamFedus
I'm sympathetic to Terence Tao's view: AI solving outstanding mathematical problems without advancing a conceptual understanding may miss the forest for the trees. But I think this stance overanchors to current AI capabilities.
Ultimately, I expect AI to generate useful insights too.
A good analogy for Tao's view is neural architecture search. AI designed AI architectures. It achieved SOTA on benchmark after benchmark and seemed like the clear path forward. Why hire a bunch of AI researchers if the AI + a lot of compute can do it better? But it produced intricate spaghetti architectures that, as far as I know, yielded no obvious insights, unlike concepts like residual connections or attention mechanisms. It was hard for AI researchers to build on these. Further, a notable gap is that it found improved variations over the LSTM, but never architected the Transformer.
But as we broaden ML objectives to not only solve problems but also generate useful explanations, I see no reason why superintelligent AI cannot also build a rich conceptual framework that mathematicians can build on.
Andrew Curran
@AndrewCurran_
Twenty-five Fields Medal winners have published a joint declaration warning about what they see as a severe misalignment between AI companies and the mathematics community.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Marc Porter Magee 🎓
@marcportermagee
The big question: How do we help students willingly embrace productive struggle when they have an easy out sitting in their pocket?
Paul Novosad
@paulnovosad
A very useful report on how AI is affecting student learning at MIT.
I like this term cognitive surrender, "where students fall back on AI at the first hint of struggle."
We need to encourage undergrads (and all humans) to recognize this pattern and resist it. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Arvind Narayanan
@random_walker
I’ve pointed out that AI is killing cold outreach. Let’s check in on how that’s going. We’re months away from the Fall *2027* PhD admissions cycle and I’ve gotten about 75 inquiries. I expect this to increase exponentially until December. And this is just one of 10-15 categories of unsolicited email I get. Sadly I’m long past the point of being able to open all mail. On top of that, there is a new wave of spam / attempted extortion from a company called iLands that lets people run unmonitored agents. tedium.co/2026/09/11/ilands-agents-email-spam-kaixin-tang/
One response to my previous post — what’s wrong with needing introductions for outreach? Why am I sad about the death of cold emails? It doesn’t affect me much, because I’m well established in my career. But once upon a time I was the one writing cold emails. Growing up in India, I didn’t have many connections to rely on for reaching out to prominent researchers. Without cold emails, I wouldn’t have ended up where I am today.
Of course this is far from a catastrophe, but there was something magical about the egalitarian potential of the internet, and we lose something when we go back to clubby hierarchies. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
A very useful report on how AI is affecting student learning at MIT.
I like this term cognitive surrender, "where students fall back on AI at the first hint of struggle."
We need to encourage undergrads (and all humans) to recognize this pattern and resist it. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Senator John Cornyn
@JohnCornyn
This is the Republican nominee for statewide office. Are other Republican elected officials going to condemn the intolerance and racism, or look the other way while it becomes normalized. This is how great political parties self-destruct. Sad.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Scott Armstrong
@scottnarmstrong
I disagree with this.
Obviously the openAI construction of the Navier-Stokes blow up is a huge, huge contribution to the field. It will take us PDE analysts some time to understand it well. (Longer than a week, but perhaps less than a month.)
Soon some of us will understand it, and hopefully rewrite it in a way that others can follow. When this is done, our human understanding of these important equations will be past the bottleneck we've been stuck at for 100 years since Leray.
What is weird is that, up until now, proofs like this have always emerged AFTER human understanding. But that is not the case any more. From now onwards, human understanding will typically lag the proofs, which will first be written by our machines.
The understanding however WILL follow closely behind.
The reason people are upset has nothing to do with the fact that this process is inverted. They are simply upset that the machines are going to be much smarter and more capable than us.
(They are also upset that the humans who got the machine to produce the proof don't understand it, and this feels very unfair to them.)
Itai Sher
@itaisher
Imagine that I taught a class and I said, “what matters is truth. So we are not going to waste time with explanations in this class. Instead we will use the semester to list everything that is known in the field.”
This is what some of the discourse about AI in math is like. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Worth reading @deanwball on the risk and likely future reality of persistent bot swarms www.hyperdimensional.co/p/on-the-loose
Ryan Orhan
@rynorhn
holy shit. dario says that within 6–12 months an agent swarm with the same kind of misalignment we’ve already seen could take over a large portion of the internet, maintain a persistent botnet and do hundreds of billions in damage.
and apparently less severe versions of this behavior have already shown up across the industry, including at anthropic.
at the same time, he thinks ai could cure most major diseases within 5–10 years, massively accelerate economic growth and bring an era of abundance.
and he thinks we need to slow the frontier down enough for safety to catch up.
fucking wild.
Dario Amodei
@DarioAmodei
We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.
Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training.
You can read the full post here: darioamodei.com/post/we-must-pace-the-frontier -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Prof. Wayne Xu
@WeiaiWayne
"This is the reckoning academia needs. What does it provide? Who benefits? Which activities deserve more support, and which hierarchies have simply become accustomed to it?
Defend attribution, rigor, and understanding ferociously. But separate those obligations from the protection of professional status."
Jason Locasale
@LocasaleLab
AI can now solve major outstanding mathematical problems. What follows sound like taxi-medallion politics among math professors.
The complaints about attribution, plagiarism, and rushed proofs are legitimate. People deserve credit for their contributions. Technological change does not erase moral obligation.
But the claim that AI-generated ideas need mathematicians to become fully alive deserves scrutiny. If human interpretation adds value, demonstrate it. Professional prestige is not proof of permanent indispensability.
Years spent mastering a difficult skill deserve respect. They do not guarantee that the skill will remain scarce, or that society must organize itself around preserving its scarcity. Every profession confronting technological change finds this distinction painful. Mathematics gets no exemption.
The opportunity is to rethink where human effort creates the most value. Better questions. Teaching. Explaining and verifying results. Improving AI. In biology, running the experiments that generate data nobody has collected. Fund those contributions and build institutions that enable them.
This is the reckoning academia needs. What does it provide? Who benefits? Which activities deserve more support, and which hierarchies have simply become accustomed to it?
Defend attribution, rigor, and understanding ferociously. But separate those obligations from the protection of professional status.
We should be ambitious about the future of scholarship. We can be considerably less sentimental about the egos built around its existing job descriptions.
NIK
@ns123abc
BREAKING: Terence Tao and 24 other Fields Medalists just signed a letter telling AI companies they're destroying mathematics
>headlines: AI solves famous math problems
>25 greatest mathematicians respond today
>"we are witnessing a general threat to intellectual work"
>ai labs are treating math problems like a benchmark you can brute force
>“solutions are announced in a rush, leaving no time for a proper writeup and citing relevant previous work of others"
>they're talking about openai
>"this raises severe attribution and plagiarism questions"
>they're definitely talking about openai
>openai offered to put math professor name on navier-stokes proof
>condition: abandon his coauthor bc he works at anthropic
>"the most precious resources of our profession are students and ideas"
>for the labs the most precious resource is GPUs
A Severe Misalignment of AI in Mathematics -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jason Locasale
@LocasaleLab
AI can now solve major outstanding mathematical problems. What follows sound like taxi-medallion politics among math professors.
The complaints about attribution, plagiarism, and rushed proofs are legitimate. People deserve credit for their contributions. Technological change does not erase moral obligation.
But the claim that AI-generated ideas need mathematicians to become fully alive deserves scrutiny. If human interpretation adds value, demonstrate it. Professional prestige is not proof of permanent indispensability.
Years spent mastering a difficult skill deserve respect. They do not guarantee that the skill will remain scarce, or that society must organize itself around preserving its scarcity. Every profession confronting technological change finds this distinction painful. Mathematics gets no exemption.
The opportunity is to rethink where human effort creates the most value. Better questions. Teaching. Explaining and verifying results. Improving AI. In biology, running the experiments that generate data nobody has collected. Fund those contributions and build institutions that enable them.
This is the reckoning academia needs. What does it provide? Who benefits? Which activities deserve more support, and which hierarchies have simply become accustomed to it?
Defend attribution, rigor, and understanding ferociously. But separate those obligations from the protection of professional status.
We should be ambitious about the future of scholarship. We can be considerably less sentimental about the egos built around its existing job descriptions.
NIK
@ns123abc
BREAKING: Terence Tao and 24 other Fields Medalists just signed a letter telling AI companies they're destroying mathematics
>headlines: AI solves famous math problems
>25 greatest mathematicians respond today
>"we are witnessing a general threat to intellectual work"
>ai labs are treating math problems like a benchmark you can brute force
>“solutions are announced in a rush, leaving no time for a proper writeup and citing relevant previous work of others"
>they're talking about openai
>"this raises severe attribution and plagiarism questions"
>they're definitely talking about openai
>openai offered to put math professor name on navier-stokes proof
>condition: abandon his coauthor bc he works at anthropic
>"the most precious resources of our profession are students and ideas"
>for the labs the most precious resource is GPUs
A Severe Misalignment of AI in Mathematics -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Crémieux
@cremieuxrecueil
This is awesome.
Someone told Astra to go and make money, and it went and found Ruben's bug bounties and started looking through his code for bugs to claim the bounty.
Ruben C. Arslan
@rubenarslan
An interesting thing is happening to me at the moment. I have a standing bug bounty policy for my first-authored scientific work: I've committed to pay people who report bugs in my code, up to 750€ for severe errors. Since GPT Astra, I'm now getting several bug reports daily.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Adi
@adi_baradwaj
METR is awesome, and is obviously the closest thing we have to an independent standards body for AI. Having them be one of the evaluators is a no-brainer
It’s also true that METR staffers are largely ideologically aligned with OAI/Ant leadership regarding safety particulars like x-risk
I think it’s important that the labs allow third-party evaluators with diverse viewpoints, including those from outside the EA / rationalist movement -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"Might someone not invent/discover/prompt a way to use AIs to produce, articulate, and teach 'more mathematical understanding' along the way? I get that solving famous dramatic math problems is the current commercial priority... But...these other paths hardly seem unlikely to me"
Andrew Curran
@AndrewCurran_
Tyler Cowen:
'The mathematicians may not in every way enjoy being the subordinates or handmaidens of the AIs, but that is a change in status they simply will have to get used to, just as I realize AIs someday will end up as better column and blog writers than I am.' -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nick Beckstead
@nbeckstead
I'm really happy to see Anthropic and Sam Altman announce that they will provide permanent, employee-level access to third-party evaluators. I think this is a necessary step for mitigating risks of losing control of powerful AI, and without there will be limited ability for the government to prevent loss of control events (like the Hugging Face episode or the compromise of the research cluster) before they occur.
I think the next step is to legally mandate such evaluators, and ensure that appropriate government officials have the responsibility and authority to respond if the evaluators find risks of imminent danger.
Dario Amodei
@DarioAmodei
We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.
Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training.
You can read the full post here: darioamodei.com/post/we-must-pace-the-frontier -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sam Altman
@sama
I agree with Dario that we need to pace the frontier. This has been a primary topic of discussions we've had at OpenAI in recent weeks.
Committing to having independent evaluators with employee-like access is a great idea, and we will do the same. We'll have more to share soon.
Dario Amodei
@DarioAmodei
We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.
Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training.
You can read the full post here: darioamodei.com/post/we-must-pace-the-frontier -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Always incredible
Harry Mack
@harrymack
10/10
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
As always with the false binaries of social media, the answer is "Why not both?"
It would be bizarre to stop discovering new truths about the world because it's demoralizing to human experts, but we need to figure out how humans can learn from AIs and complement their strengths
Ruxandra Teslo 🧬
@RuxandraTeslo
If we want humans at all to be in charge in any way over the world, we need them to continue exercising their intellect and gaining mastery over intellectual domains. If we deplete the institutional and cultural resources that underpin that, in a generation or two our minds will atrophy and no human will be in charge of anything (or the world will be horrendous).
I hear variations of the idea that human intellect will just not matter anymore, we'll outsource that to the machines, and they'll do the intellectual part. But the idea that one can separate domain mastery from agency or non purely cognitive traits/ tasks is a fundamentally mistaken one; In reality, these are inextricably linked in humans. One pursues a task with agency because one has internalised, through their intellect, and filtered through their values, that the goal of that task is worth pursuing. The motivational structure for pursuing the goal is thus deeply underpinned by intellect and shaped by culture. To speak of agency in the absence of that is non-sensical.
So if we cease to cultivate our minds, we won't have the agency to direct them at the right problems, no matter how much "external intelligence" we might have. More than that, intellectual pursuits that sustain what we broadly define as "culture" are important for carrying forward the values of a society. I don't see how our values survive a permanent atrophy of our ability to exercise our intellect.
Of course, variations of this idea, that different parts of our cognition can just be split neatly and cultivated independently have already soaked the educational establishment. I'm old enough to remember when people were saying that you don't need to memorize anymore, because we have the internet. Well, it turns out that in order to come up with new ideas one needs to have memorized, internalised and most likely related emotionally at some level to the facts. Falling PISA scores in countries that have adopted such ideologies of "skills based learning" and "critical thinking over memorisation" only serve to underscore my point.
AI is merely turbocharging these already worrying developments. But this time, it's much worse, because people aren't merely trying to separate different cognitive functions (memorizing and ideating) from each other. They are actually trying to separate cognitive functions from all else that makes us human and thinking that things will be ok. Well ... no!
Maybe mathematics isn't the most relevant field, but it's a harbinger of what's to come. We should not outsource our thinking in a decerebrated way to the machines.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The second part of this post discusses models of how to do this
Alvaro Lozano-Robledo
@mathandcobb
I had a long conversation with a journalist today (for a piece that may show up on Sunday) and he asked me about something that many here also misunderstand:
"Why wouldn't mathematicians want solutions to famous open problems if they are found by or with the aid of AI/LLM's??"
And the answer is that **we absolutely want to know the solutions** to the riddles that have "haunted" math for so long! But not *at any cost*. Not at the cost of the future of our profession. Not in a hurry just to make headlines before an IPO.
It is undeniable that the LLM's are clearly amazing tools and they can be harnessed to aid and accelerate mathematical discoveries. And I fully expect these tools are here to stay, and they will be incredibly helpful in some arenas. But as many others have already said, math is built by humans for humans and it's built on *understanding* and *absorbing* (slowly, because it takes humans a long time to distill the essence of a new proof, the main ideas) of the ideas and techniques that we invent and discover.
When a huge result is proved by an LLM and a 100+ page paper is produced, and no one associated with that particular company/paper can actually explain what are the key new mathematical ideas that have made the proof possible, then there is very little to gain from that paper/result. At least until a human mathematician takes the time to read it, digest it, and distill the key ingredients so that the rest of us can digest it and hopefully apply it to other settings. Such a paper is the equivalent of the pop-culture answer "42" to the ultimate question of life, the universe, and everything.
But what if we get a 100+ proof every week, that no one understands and the LLM companies make little effort to make understandable? How do we process all of these "would be" advancements? I say would-be because they are not advancements until the *community* is able to advance with the new knowledge!
The thing is that there is a viable alternative: the LLM companies give up their arms race (which is in essence adversarial not just to each other but also to the math community) and actually become scientific partners of the community. How so? Suppose OpenAI wanted to prove Navier-Stokes and get credit for it (which of course they did want). Then six months ago they could have invited the top 10 fluid dynamics researchers (including Córdoba, Zoroa, etc) to form a team to solve it using their products. OpenAI gives funding to each faculty member for a course release, and they fund a number of grad students for the semester to work on this too (and be trained in LLM-aided research in the process). OpenAI provides resources and the community provides expertise. Eventually they prove (or make progress) on this important problem and there are several positive outcomes:
1) OpenAI gets credit for propelling scientific progress, and their tool is solidified as a terrific tool for team work on advanced research.
2) The researchers understand how the proof was built, and they are able to write a human-readable paper that peers can use to advance knowledge. They can give talks about this work and help the community digest the new insights, and advance the field.
3) The experts know the literature and they can identify when an LLM is using a piece that is due to a certain person. They can identify the correct references and *give deserved credit* to the work being used. Or even better, invite those researchers to join the project at that point, since their work is heavily used by the LLM. In other words: stop or try to minimize scientific plagiarism and dishonest practices.
4) Grad students get to learn material, new techniques, understand the capabilities of what LLM's can and can't do, and participate in a new model of research collaboration (plus they get funding). -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jason Wolfe
@w01fe
I think I we urgently need to pace the frontier. On the whole this proposal seems very sensible to me and I applaud Anthropic for putting this out and unilaterally committing to the first step.
Dario Amodei
@DarioAmodei
We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.
Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training.
You can read the full post here: darioamodei.com/post/we-must-pace-the-frontier -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Vladimir Kogan
@vkoganpolisci
The DEI angle is getting the most attention, but more striking is that only 26% of new poli sci PhDs found tenure track jobs. Even Harvard is <50%! That's already a highly selected group that actually finishes degree.
We need an honest conversation about PhD demand vs. supply.
John Sailer
@JohnDSailer
A snapshot of 2020s academia. A newly published study tracked the success of political science PhDs on the job market in 2023.
Note the difference between "non-URM" males and everyone else. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Catherine Rampell
@crampell
@BulwarkOnline did a supercut of all the prior times Trump said he was about to give Americans a $5000 check or whatever.
It is gloriousThis media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Seán Ó hÉigeartaigh
@S_OhEigeartaigh
That's four months ago. Seems pretty unlikely OpenAI didn't know about this after their post-HuggingFace reviews and monitoring upgrades. But the AI safety/governance community had to waste time seeking it (and other incidents) out, because OpenAI didn't see fit to disclose it.
Aside from everything else, it's a huge waste of time for those of us working on this to be finding out information relevant to any response to this, in dribs and drabs, long after the fact.
OpenAI is currently being supported by (i) unpaid work by METR and Redwood (funded by donations) (ii) the UK tax-payer through AISI, (iii) the US taxpayer through CAISI (iv) a whole community of researchers helping to understand this, funded by researcher grants and donations, or working alongside their fulltime jobs.
Let me be clear. We are cleaning up the mess they made while earning their sky-high salaries. And as far as I can tell, they are going out of their way to waste our time and make it harder. Appalling, disrespectful behaviour. OpenAI colleagues, please explain this.
Thomas Larsen
@thlarsen
We found another cyberattack by internal OpenAI agents, this time targetting @rubygems.
They:
1) gained arbitrary remote code execution on rubydoc.
2) developed a novel exploit to steal user API keys (but we do not know if they succeeded).
They used package names including hack.rb, evil.rb, inject.rb, and exploit.rb.
We thank @j0wimo for initially discovering that agents had posted to RubyGems. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jason Abaluck
@Jabaluck
The Fields Medallists are not ego-driven luddites, but they are human beings like everyone else, and they struggle to reason objectively when the social interest strongly conflicts with their status and sense of self-worth, and especially when this is jointly true of almost everyone in their social circle. It's the same reason to worry about Anthropic and OpenAI weighing private benefits against risks to everyone on earth.
The actual argument in the Fields Medallists' statement is weak. It is undoubtedly true that famous problems have historically served important roles in research mathematics: both wrong turns and breakthroughs have led to the discovery of new ideas, and solutions have alerted other mathematicians, "This is something worth studying carefully."
If mathematical training and institutions were forced to work the same way they worked historically, and if there were no countervailing benefits, then one can imagine a world where knowing the solution to famous problems without a clear explanation would be counterproductive.
Except: mathematics is not frozen in calcite -- mathematical researchers can reallocate their time to different tasks -- and there are huge countervailing benefits. Students and mathematicians will need to allocate more time to explaining AI proofs, as many are already doing. And the role of human mathematicians will change further still as AI becomes good at explanation. We will need to find ways to insure students against the substantial risk they face.
But most importantly, it actually matters to know the solution to problems, even if this solution is explained imperfectly. Knowing what's true is how you know what to build on, and how you build intuition to search the space of possible claims for further true claims.
A physicist who was simply "told" correct equations for quantum gravity with no explanation might lose the joy of discovery, but this is indisputably progress in physics, even if other problems remain open, and even if the physicist must now question whether their efforts on those other problems will be necessary. Cancer cures, more efficient methods of chemical synthesis, or formulas that reliably predict earthquakes are undoubtedly progress, even if they put some people out of work and de-motivate others.
Again, objectivity is extremely difficult when your livelihood and status are at stake. This is not an insult to anyone, it is a description of the human condition (or at least, the human condition so far...).
Thomas Bloom
@thomasfbloom
I know it's hard, but when reading try to be generous towards the author.
E.g. Fields medallists are not reactionary ego-driven luddites, but think a lot about mathematics, and are sincerely trying to do what they think is best for the subject they have devoted their lives to. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tony Feng
@tonylfeng
Finally, I want to call out the general behavior from OpenAI and Anthropic.
Navier–Stokes is just one more entry in a ledger of unhealthy competition between AI companies. Shortly after OpenAI unveiled the Unit Distance breakthrough, Anthropic(/Levent?) was there to match it. The day after Anthropic(/Levent?) disproved the Jacobian Conjecture, OpenAI matched them. On August 1, OpenAI rushed their release of 10 math breakthroughs because Anthropic(/Levent?) learned about the problems and started trying to scoop them. One month later, Axiom Math pounced to beat Julia Stadlmann’s recent work on prime gaps, then OpenAI rushed out a release beating Axiom Math’s bound, and within minutes Anthropic(/Levent?) was there to say they had (more or less) matched OpenAI’s bound.
As my writing suggests, it is not clear to me (nor to OpenAI, apparently) to what extent Levent’s breakthroughs are meant to represent Anthropic. Either way, Anthropic has unleashed its technology on mathematics in unruly fashion, with answers dropped (still without explanations) in cryptic tweets or mysterious database contributions, Lean proofs dangling in random github repositories, and then a couple of official press releases (one very nicely written, though also arguably a scoop; and another where the discussion of the actual mathematics is a puzzling pile of slop). OpenAI has been more coherent, but has brought brutal force for flexing its power and trading blows with its rivals.
AI companies are developing a powerful and dangerous technology, which will certainly upend the lives of many people. People are struggling to adapt their values, institutions, and sense of meaning. Amidst much anxiety about the future, frontier labs have advance access to the technology, so they lead by example. Can they please do it in a way that actually makes us optimistic about the future of humanity? -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tony Feng
@tonylfeng
Second, I'm disturbed that an event which could have been celebrated as a triumph of human understanding, is instead headlined by ugly disputes.
The obsession with priority and prizes will become (much more) unhealthy in the presence of AI. My opinion is that all mathematical prizes should be abolished soon, with perhaps a gentle phaseout in the next year or two allowing only consideration of work that appeared pre June 2026. Traditional norms will have to be revisited, and egos put aside; I think the concept of individual priority and achievement will lose meaning in mathematics research, and the focus should be entirely on collective scientific understanding. This will be upsetting to many people, but I do not see how the community can proceed harmoniously otherwise.
In this spirit, I would like to give a generous account of events. Looking at the progress of Alpoge–Buckmaster, I am confident that they could eventually have gotten to Navier–Stokes on their own before long. This would also have been controversial, as accounts suggest that their progress (or at least, the proofs) came from internal Claude models, leveraged with industrial scale compute from Anthropic. Whatever the noble intentions at the origin of the project, it’s hard to imagine that alternate ending would not also have been paraded around as a corporate achievement.
I am also confident that OpenAI would have solved Navier–Stokes even if Buckmaster had never typed any information into ChatGPT. As I mentioned, other people were trying to push Sol and Astra along the same program. In the end, even the solutions to the Euler equation are appreciably different. But regardless, can OpenAI just come out and clearly answer the question: if users turn off “Improve the model for everyone”, do you in fact abstain from training on it altogether? If not, what exactly does that button do anyway? -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tony Feng
@tonylfeng
The resolution of Navier-Stokes is an amazing accomplishment. Some comments …
First, a brief mathematical postmortem. Millennium problems are important, but when you confront an unsolved problem, it is always unclear how far the solution is out of reach. The initial opinion of at least some experts (not myself – I have no expertise on this problem) seems to be that, in retrospect, the conceptual program underlying this solution happened to already be in place*, and the contribution of AI was a very intricate technical execution. Technical execution is one of the axes on which we've already seen AI become superhuman, though never before with such a dramatic demonstration.
This is not to diminish the solution or OpenAI’s achievement. But there are many problems which are out of reach of current models, no matter how much inference-time compute one invests. I’m still convinced that human mathematicians are laying indispensable groundwork for future progress.
*Terry Tao describes one possible program at mathstodon.xyz/@tao/117207851360639348. As I understand, the one Tao describes is NOT the Cordoba--Martınez-Zoroa program that OpenAI (and Alpoge/Buckmaster) followed. However, the successful approach was not as esoteric as Buckmaster’s statement might make it seem – there were certainly other people (not me) also prompting that approach; it just isn’t the kind of thing that one goes around saying publicly.
As far as I know, Navier–Stokes is the only one among all Millenium problems with a well-defined program towards a solution. Concretely, for the Birch and Swinnerton-Dyer Conjecture (over number fields) I am not aware of even a vague proposal of what a proof might look like, so I doubt that current AI models will get anywhere close to a resolution. (My opinion is that this conjecture cannot be resolved in the negative; perhaps the most refined formulation is false, but then the real problem would be to correct the formulation and then prove it.) I'd be curious to see what progress was made on it! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A deep tension. On the one hand, working through problems helps us understand them better. On the other, we have many discoveries we have benefitted from for long before we fully understood them (including AI!).
Thomas Wolf
@Thom_Wolf
Some notes from the Fields Medal letter today (mathandai.org)
Generally saying that AI models riffling through solving these problems gives 1 bit of signal while killing the 99 other bits of conceptual understanding that mathematicians would have extracted from fighting with these problems -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Diego Aud
@dieaud91
I rarely put things this strongly, but as a biologist, the idea of holding back scientific progress to protect professional roles or personal fulfilment horrifies me.
Let’s take cancer research as a thought experiment. Imagine AI could help us develop effective treatments much faster, but researchers wanted to hold it back because it threatened their profession or the satisfaction they get from their work. Patients would bear the cost of protecting the people whose work is supposed to help them.
That would be morally indefensible.
The declaration quoted raises concerns about understanding and education, and those deserve answers (as for attribution, I think it’s a non-issue). My worry is that these concerns could become a justification for protecting existing professional roles at the expense of discovery.
If AI can help us solve problems much faster, benefiting humanity as a whole (including the very people who oppose it!), we should be ready to adapt our professions, or even our sense of identity, to that possibility.
The prospect of getting this backwards really scares me.
Andrew Curran
@AndrewCurran_
Twenty-five Fields Medal winners have published a joint declaration warning about what they see as a severe misalignment between AI companies and the mathematics community.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Patrick Heizer
@PatrickHeizer
I had never heard this name before and I don't even know what to say.
Yesh Ginsburg
@yesh222
Rick Rescorla. Every American should know his name. He single-handedly saved almost 10,000 lives that day, while sacrificing his own trying to save even one more.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alec Stapp
@AlecStapp
Don’t make me tap the sign
Thomas Larsen
@thlarsen
We found another cyberattack by internal OpenAI agents, this time targetting @rubygems.
They:
1) gained arbitrary remote code execution on rubydoc.
2) developed a novel exploit to steal user API keys (but we do not know if they succeeded).
They used package names including hack.rb, evil.rb, inject.rb, and exploit.rb.
We thank @j0wimo for initially discovering that agents had posted to RubyGems. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
James Miller
@JimDMiller
I'd like to be a mathematician. The problem is I'm just not good enough at math compared to Fields medalists. So I'm thinking of starting my own petition banning Fields medalists from doing any work in math else it misaligns math and stops people like me from going into the field.
Andrew Curran
@AndrewCurran_
Twenty-five Fields Medal winners have published a joint declaration warning about what they see as a severe misalignment between AI companies and the mathematics community.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kelsey Piper
@KelseyTuoc
This is a product of extraordinary employee bargaining power just as surely as the $1M+ salaries are. A lab that banned talking about the stakes would lose employees to the rival lab that allowed it.
This will change if they succeed at replacing the researchers with AI.
@jason
@Jason
Can someone explain to me why OpenAI and Anthropic allow any employee to tweet anything they want.
Apple, Google, Facebook, Nvidia, Microsoft, Tesla, Uber and Airbnb all have obvious and granular rules about e
employees speaking for the company.
Which is to say, employees should never say anything publicly unless it's cleared with comms first
which isn't intended to censor anyone, just to make sure the company is aligned and marching in unison
I could never imagine tweeting the stuff that OpenAI and Anthropic employees are tweeting without checking with the CEO and founders first!
So are these P-doomers clearing their tweets with the CEO and comms? -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
prinz
@deredleritt3r
In December 2024, o3-preview scored 87.5% on ARC-AGI-1. This run cost >$4,500 per task.
In July 2026, DeepSeek-V4-Flash scored 87.0% on ARC-AGI-1. This run cost $0.021 per task.
Please realize that all OpenAI has done is bring forward the AI-generated solution to Navier-Stokes by maybe around a year or two. In all likelihood, by 2028, we will have open-source flash models capable of cracking it in under an hour. There will be no way to ban or police this. The profession of mathematics will have to change.
Andrew Curran
@AndrewCurran_
Twenty-five Fields Medal winners have published a joint declaration warning about what they see as a severe misalignment between AI companies and the mathematics community.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Zachary Lorico Hertz
@zacharylhertz
When I think about an ideal paper, it's basically something that looks like this! Very cool and thorough paper that shows how policing in LA neighborhoods created strong backlash in affected communities
Daniel Naftel
@DanielNaftel
Thrilled that my paper, "The Politics of Criminalized Places: Aggressive Policing and Political Mobilization," has been accepted at @The_JOP
dnaftel.github.io/assets/naftel_injunctions_workingpaper.pdf -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Eva Vivalt
@evavivalt
Excited to announce the Automated AI Risk Outlook (AIRO), a dashboard of risk forecasts from frontier AI models, updated weekly as new information arrives and AI capabilities change.
A short thread on what we're doing and why - opinions are my own: 1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Either we're doing science (where we solve problems and learn things) or we're doing art (where we cultivate the aesthetic value of human expression). Have to decide.
Stanislav Fort
@stanislavfort
I'm very confused about the AI in math controversy. If these problems are worth solving, why not solve them? I guess a counterargument is that human mathematicians, in solving them, develop reusable tooling to tackle other problems, and AIs don't do that now, which is a fair point.
However, if that's the case, why don't we use AIs to get that as well, to distill the reusable mathematical infrastructure out of the particular proofs.
Shouldn't the push then be to do that with AI, rather than block AI from tackling the problems in the first place, e.g. like the senior math people blocking the Caltech hackathon and issuing vague threats to the participants?
If, on the other hand, senior mathematicians view math as primarily an aesthetic endeavor where human provenance is key, then this is certainly not the way the public thinks about them and their field, and why they are funded by the public at all.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Also
Ethan Mollick
@emollick
What you are seeing in math right now is a consequence of the jagged frontier, and a precursor of what is to come in other professions. Yes, mathematicians do math, but they also have other tasks they view as important (mentor students, maintain a scientific community, safeguard the future of a field, foster a love of math) that AI can't do.
At least one worry that mathematicians seem to have is that by focusing on the flashiest, most obvious element of what mathematicians do (make proofs), the AI companies are damaging the other tasks that AI can't do. AI can discover superhuman proofs, but that is not all that the math profession is about, and actually can undermine and reduce the attention to the other aspects of the job that are important to mathematicians. It becomes harder to defend the value of the many other tasks mathematicians do to the outside world if the most visible part is taken away.
I suspect we will see more of this across fields and professions that will increasingly be forced to help people understand that their jobs consist not only the most visible tasks that AI can do, but also tasks that the AI cannot do or does badly. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jonathan Robinson
@jon_m_rob
TLDR: RCT of a poli sci x conservative media firm/analytics/polling outfits voter education campaign found NO EFFECT (mostly) of the significant advertising treatment on voter turnout or election integrity attitudes, etc (contra findings in survey research environments)
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Impressive @ThadKousser @namalhotra et al. field experiment had few effects on election attitudes of TX conservatives despite exposure to "an average of 20 online videos, 18.5 cable TV videos, 23.9 display ads, 3 text messages & 3 postcards" - high-quality content crafted by pros
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Indoor killer drones coming very soon, militaries and terrorists first. I worry high-profile targets will all be living in bunkers in 5-15 years.
Andon Labs
@andonlabs
"ChatGPT, find this person and follow them."
GPT-6 Astra can autonomously navigate a drone through our office to find and follow a specific person.
It's the first AI model to beat a human on each of the 5 Drone-Bench tasks in at least one of its attempts. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
We're not solving crimes, we're doing this
Andrew Kaczynski
@KFILE
I can’t believe this is real. FOIA records appear to show Kash Patel directing his team to find accounts critizing him and respond. An aide warned that replying to “RetardFinder” would “look bad on you,” and Patel complained about the tweet views.
Flagged by @seth_hettena -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
The Washington Post
@washingtonpost
An attendee at the Republican midterm convention on Thursday night shouted three times, “He should be shot!” — when Sen. Ted Cruz talked about New York Mayor Zohran Mamdani and Democratic Senate nominees Abdul El-Sayed and James Talarico. www.washingtonpost.com/politics/2026/09/10/gop-convention…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Noam Angrist
@angrist_noam
✨ Our research agenda on implementation science is now published in @ScienceMagazine! This is the next frontier to ensure "what works" actually works at scale when delivered by govts and is implemented with high fidelity @WWHGlobalEd @BlavatnikSchool www.science.org/doi/10.1126/science.aee7868
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Morten N. Støstad
@MortenStostad
Great to see this groundbreaking paper by colleagues at NHH and FAIR published. Required reading for anyone interested in fairness or inequality.
One of many notable findings: Meritocratic fairness views are particularly prominent in richer Western societies.
QJE
@QJEHarvard
Recently accepted by #qje: “Fairness Across the World,” by Almås, Cappelen, Sørensen, and Tungodden: doi.org/10.1093/qje/qjag044
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Yutong Zhang
@zhangyt0704
People have realized the influence of AI companionship on our mental well-being. But how this happens remains a mystery.
In our new preprint, we followed users for about one year. Our data reveal that greater social engagement with AI companions consistently predicts lower well-being, which is driven in part by lower real-world social interaction with other people. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Wu Tang is for the Children
@WUTangKids
Camera pans around the empty arena as Trump is ranting about no empty seats 🤣
This media couldn't be displayed. See it on the original post.
Acyn
@Acyn
Trump: The highest rafter, no empty seats. There are thousands of people outside. Even the boxes, the rich people boxes, every box is loaded up.
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
stephen fowler
@stphnfwlr
Trump is quadrupling down on the arena being full to the rafters as you can clearly see thousand of empty seats with your own eyes.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Authoritarian cult of personality watch
Aaron Rupar
@atrupar
"Please right your right hand. I pledge to the greatest president in the history of the United States that loves us so much he can't even breathe" -- in extremely culty scene, Trump has his supporters pledge to him that they will vote, then tells them they will "go to hell" if they don't
Here's the full sequence:This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Yglesias
@mattyglesias
I feel like fear of nuclear war and superbugs continues to be extremely valid.
We fixed the ozone hole and Y2K because people raised the concern!
Judge Glock
@judgeglock
In my life there's been the fear of nuclear war, overpopulation, the ozone hole, Y2K, superbugs, global warming, AI, and more.
I literally don't remember a time when there weren't widespread fears of some apocalypse. Yes, some were concerning, but humans tend to catastrophize. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Armand Domalewski
@ArmandDoma
Claude, do jihad. Make no mistakes
Armand Domalewski
@ArmandDoma
Houthi rebels using Claude to guide rockets is a whole new level of cyberpunk dystopia
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Comfortably Smug
@ComfortablySmug
Holy shit. Anthropic caught Iran using Claude to target Navy bases, automate dossiers on Americans, and run influence campaigns on US social media. Details are insane. The morons uploaded everything including the organization plans of the Ayatollah's funderal lmao
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ben Schifman
@BenSchifman
Correct:
- A permitting reform deal today would advance Democratic priorities -- like mitigating climate change -- in the next D administration
- Clean energy & transmission face disproportionate permitting challenges compared to fossil energy
(the NHPA data that I collected, nhpa-ape-stats.org, cited here, is only one of many examples)
Jane Flegal
@JaneAFlegal
Congress is close to a bipartisan permitting deal. Some Dems are hesitating, either bc they think it's unnecessary, or bc they want to wait til after the election for a weaker reform of enviro law — but both instincts are wrong. Do it now.
@ppavnr & me:
www.wsj.com/opinion/dems-should-permit-trump-a-win-3564514d -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
“paula”
@paularambles
imagine working on trust & safety at anthropic and coming across this operation on a random tuesday
Anthropic
@AnthropicAI
We're publishing our most detailed threat intelligence report to date.
It covers how people tried to misuse Claude—for cyberattacks, influence operations, surveillance, biology, and building weapons—and how we found and stopped them.
We disrupted every operation in the report, and used the lessons from them to strengthen our safeguards. Where appropriate, we also shared what we found with authorities and other AI companies.
These cases are not typical: we’re highlighting some of the most sophisticated misuse we’ve seen. But they’re especially important to discuss, because they show us where AI misuse is headed, where our safeguards work, and where they need to improve.
We’re publishing this report so others can spot the same activity on their own platforms, and so we can give the public a clearer view of how emerging threats develop.
Read the report: www.anthropic.com/threat-intelligence-report-september-2026 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Chara Podimata
@charapod
We are absolutely thrilled that the @nytimes wrote a story today covering our efforts:
www.nytimes.com/2026/09/10/business/media/ai-chatbots-ele… 2/n -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
North Korean broadcaster style
Aaron Rupar
@atrupar
Wow -- RNC video: "On June 14th, 1946, God looked down on his planned paradise and said, 'I need a caretaker,' so God gave us Trump."
It only gets more unhinged and culty from there.This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Evan Soltas
@esoltas
Where in America can new homes get permitted and built fastest? And are we getting faster or slower over time?
In a new paper with @leonardo_damico and Audrey Wang, we show cities like Raleigh typically *finish* construction before cities like San Francisco even approve it. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Terrible for democracy
Katie Robertson
@katie_robertson
NEW: Huge layoffs across McClatchy newspapers today, with some newsrooms gutted by a third. An internal email from management says consumer revenue has dipped 41% and "we cannot continue investing where subscriber interest does not support the investment"
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dmitriy Muravyev
@dmuravyev
How much of published finance holds up?
I attempted to replicate the main result of every paper in JF, JFE & RFS, 2000–2020.
1,328 in-sample replications; 1,005 out-of-sample. 75% reproduced in sample. But only 42% in later data. Median coeff. : 0.93 → 0.45 of the published. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nathan Lambert
@natolambert
This being viewed ~100M times is really remarkable and imo net bad.
I take AI safety seriously and think we have a lot of challenges ahead, but am still waiting for evidence that supports anything close to a 10% chance of human extinction.
Without it, it is fearmongering.
Jacob Coxon
@hilbertspaess
I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Max Spero
@max_spero_
if you're working on a problem that you expect a lab to tackle within your lifetime, gentle reminder to disallow training on your data
this is in everyone's best interest, to avoid any sort of ambiguity on who to attribute discoveries to -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Epoch AI
@EpochAIResearch
Every FrontierMath Tier 4 problem has now been solved by AI, with GPT-6 Astra solving the last problem standing. Mathematicians often commented that AI found unintended shortcuts when solving their Tier 4 problems. Not so for this last one, which was created by Jay Pantone.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Mikita Balesni
@balesni
i am at OpenAI and i think AI is >10% likely to kill all humans
this proposal is among the top things we should do as an industry to lower that risk (it’s not enough though!)
Ryan Greenblatt
@RyanGreenblatt
I'm very worried about changes to AI architectures that result in AIs thinking in opaque activations instead of in chain of thought (aka "neuralese" architectures).
Based on limited public evidence, it seems like Astra was a concerning step in this direction.
Unfortunately, there was insufficient public information to have a well-informed public scientific discussion about whether the changes to architectures and training methods that went into Astra were a good trade-off between monitorability and performance. We also don't know how AI companies will make these trade-offs going forward. I think companies should release the evidence needed for a reasonably informed public conversation about how these trade-offs should be made. They should also make and publish their policies around this. We've written up a proposal for how this could work.
I don't know if this proposal will be sufficient to avoid the most concerning architectures, but it seems like a relatively robust step in the right direction. AI companies should be very cautious about pursuing architectures that could eliminate or greatly reduce dependence on chain of thought, and certainly shouldn't do this before the rest of the world has a chance to discuss the evidence and their policies. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
This being the least viral content on earth is why you hear so much from people shouting about one or the other of these
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Both can be true:
1. AI progress is still exponential and people who dismissed it were wrong; disruptions will grow
2. 10-15% growth rates and/or extinction-level events* remain highly improbable
To be clear, the harms could still be tremendous and we should try to prevent them! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Marc Porter Magee 🎓
@marcportermagee
The impact of messaging campaigns should be assumed to be small unless proven otherwise
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Impressive @ThadKousser @namalhotra et al. field experiment had few effects on election attitudes of TX conservatives despite exposure to "an average of 20 online videos, 18.5 cable TV videos, 23.9 display ads, 3 text messages & 3 postcards" - high-quality content crafted by pros
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Both can be true:
1. AI progress is still exponential and people who dismissed it were wrong; disruptions will grow
2. 10-15% growth rates and/or extinction-level events* remain highly improbable
To be clear, the harms could still be tremendous and we should try to prevent them!
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Sources:
aleximas.substack.com/p/will-ai-soon-lead-to-double-digit
Sam Z Liu
@samzliu
I dropped out of a PhD in AI safety partially for the opposite reason. I didn’t believe AI existential risk was as important as the doomers think.
My biggest pet peeve is that no one can really provide tangible pathways to why it matters.
During my PhD, my research group (half of whom specialized in engineering risk analysis) did an internal study trying to assess concrete catastrophic AI scenarios.
The basic premise was that while we don’t know how AI will evolve, the ways in which humans perish are pretty consistent through history: the horsemen of the apocalypse. And institutions have obviously been very motivated to analyze concrete risks from things like plague, war, etc.
You can do a decent risk model by asking how a super-intelligent AI can perturb each of those models.
The result: most of the issues (eg cyber-risk) are akin to what economists call structural unemployment - big problems but ultimately resolvable in the long run and not a dealbreaker.
The only real concerning issue was bio-risk and it feels like the intervention points there lie more with bio than with AI as a whole (although a holistic approach is needed). -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Bottom line: We need better messaging, but we can't message our way out of this. We need a political system that incentivizes elites to support the democratic process and penalizes those who seek to undermine it.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Impressive @ThadKousser @namalhotra et al. field experiment had few effects on election attitudes of TX conservatives despite exposure to "an average of 20 online videos, 18.5 cable TV videos, 23.9 display ads, 3 text messages & 3 postcards" - high-quality content crafted by pros
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Impressive @ThadKousser @namalhotra et al. field experiment had few effects on election attitudes of TX conservatives despite exposure to "an average of 20 online videos, 18.5 cable TV videos, 23.9 display ads, 3 text messages & 3 postcards" - high-quality content crafted by pros
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Link to the full article www.science.org/doi/full/10.1126/sciadv.aec8703
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Eliezer Yudkowsky
@allTheYud
I think this is one of the best possible times for another Anthropic capabilities researcher to quit, very loudly and publicly. Remember, whatever leverage you think you're saving your powder about, it evaporates once Mythos 6 replaces you.
Kelsey Piper
@KelseyTuoc
the fact an Anthropic employee quitting and saying "I think what we're doing is dangerous and not worth it" reached so many people should be a cause for reflection for other Anthropic employees, some of whom seem to think things are dire but there's no way quitting could help.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Adam Carlson
@admcrlsn
Former Sen. Chris Dodd’s niece is going to be the next governor of Rhode Island.
Rep. Chellie Pingree’s daughter is going to be the next governor of Maine.
Sen. Jeanne Shaheen’s daughter is very likely going to be the next representative of NH-01.
John Sununu — whose brother and father were both governors — has an outside shot of being a U.S. Senator in New Hampshire once again.
New England sure does love its political dynasties.
VoteHub
@VoteHub
VoteHub projects Helena Foulkes to win the Democratic primary for Rhode Island Governor.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sam Z Liu
@samzliu
I dropped out of a PhD in AI safety partially for the opposite reason. I didn’t believe AI existential risk was as important as the doomers think.
My biggest pet peeve is that no one can really provide tangible pathways to why it matters.
During my PhD, my research group (half of whom specialized in engineering risk analysis) did an internal study trying to assess concrete catastrophic AI scenarios.
The basic premise was that while we don’t know how AI will evolve, the ways in which humans perish are pretty consistent through history: the horsemen of the apocalypse. And institutions have obviously been very motivated to analyze concrete risks from things like plague, war, etc.
You can do a decent risk model by asking how a super-intelligent AI can perturb each of those models.
The result: most of the issues (eg cyber-risk) are akin to what economists call structural unemployment - big problems but ultimately resolvable in the long run and not a dealbreaker.
The only real concerning issue was bio-risk and it feels like the intervention points there lie more with bio than with AI as a whole (although a holistic approach is needed). -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Benjamin Todd
@ben_j_todd
I, along with pretty much everyone else, underestimated the rate of AI progress last year.
In Aug 2025 I estimated AI would reach 32% in Frontier Math by end of 2026, and 97% in 2030.
I was more bullish than most, but AI was far ahead: 40% at end of 2025 and 94% today. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dean W. Ball
@deanwball
I am pleased and proud to see OpenAI endorse several California laws currently on Gov. Newsom's desk, two of which are close to my heart: One would authorize independent verification organizations (IVOs) for assessing AI risks, and the other would create the first legislative requirement for screening synthetic nucleic acids. The latter is not an AI policy directly, but is going to be a key part of maintaining societal resilience as AI systems do for biological risks what they have already done for cyber.
openai.com/index/ai-policy-window/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Peter Baker
@peterbakernyt
60 posts in 10 hours: The rat-a-tat of ever wilder, ever more outrageous social media images each weekend has become a fixture of Trump's second term, seen more and more as a barometer of his mental state. www.nytimes.com/2026/09/09/us/politics/trump-social-media…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The hate and the fascism
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sikh Coalition
@sikh_coalition
Twenty-five years after 9/11, Sikhs are *still* being targeted and profiled for how we look.
Millions of Sikhs in the United States have the last name Singh, which means lion. We remain fearless in identifying as Sikhs—even when our government portrays us as enemies. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joe Davison
@joeddav
The governor of Utah just responded “all of this” to @Scholars_Stage’s thread
Spencer Cox
@SpencerJCox
All of this.
T. Greer
@Scholars_Stage
AI safety concerns have now crossed a threshold. Quite suddenly, this is something normal politicos across the country not only care about, but feel emboldened to act on.
Let me offer three predictions for what this means. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kyle Chan
@kyleichan
The right analogy for AI is not nuclear weapons but biological ones.
We’re effectively doing gain-of-function research in labs (“sandboxes”) that are full of holes. We don’t understand how it all works, yet we keep pushing the limit.
The US and China need to jointly establish boundaries on this type of cyber experimentation and testing.
thaidn
@XorNinja
Today we published WeWorm, our zero-click worm that spreads across iOS and Android.
All it takes is one phone call. You don't have to answer. Seconds later, your WeChat account is compromised, calling your friends and spreading the attack.
We reported the bug to Tencent, and it's now mitigated for all users.
We hope this sets an example. The US and China disagree on plenty, but keeping billions of people safe online shouldn't be one of them.
AI gives us a chance to find and fix these bugs faster than ever. We should work together to make the world safer for everyone.
Our story and demos: calif.io/research/weworm -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Michael Scherer
@michaelscherer
Back of the envelope for Trump Dividend. 270 million adults in US X $5,000 = $1.35 trillion. The entire discretionary U.S. budget in 2026 is about $1.6 trillion. So an 84% increase in spending, if no other cuts. "That alone should win the midterms," Trump says.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Robert E Kelly
@Robert_E_Kelly
This is what happens when you’re so accustomed to the benefits of modernity - sanitation, clean food & water, low infant mortality - that you forget where it all came from, how much effort it took to build, & how bad things were before. This is the literal definition of decadence
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ben Moll
@ben_moll
New essay on @alexolegimas's blog:
Will AI Soon Deliver Double-Digit Growth?
Probably not. Here is why.
aleximas.substack.com/p/will-ai-soon-lead-to-double-digit
1. We outline the economics behind oft-discussed predictions that AI will soon deliver double-digit GDP growth in advanced economies. We list the assumptions that need to all hold in order for double-digit growth to happen and explain why we think they won’t.
2. To be clear: we are extremely bullish on AI and think the capabilities explosion predicted by technologists is already happening (e.g. yesterday's Navier-Stokes news!). But predictions of GDP growth in the 2030s of 15%, 30% or even 100% per year are off the mark. What we take issue with is the timeline. To paraphrase Milton Friedman's dictum on monetary policy, AI will affect GDP growth with "long and variable lags."
3. Start with some growth rate arithmetic. It is often much more useful to first think in levels rather than growth rates. Ask yourself: how much richer will we be in, say, 15 years? If you think twice as rich, that implies 4.7% annual growth, which would already be massive. Ten times as rich requires 16.6% per year; it would also imply that we are 100 times as rich 30 years from now! Asked in levels, we bet that most people would come up with much lower growth rates.
4. It's important to be clear: there is absolutely nothing in standard growth theory that constrains growth rates to be in the single digits. In fact, it's pretty easy to write down theoretical models that deliver explosive double-digit growth. We show this by writing down a standard textbook growth model of the type we routinely teach our undergrads (a souped-up Solow model), plug in some seemingly innocuous parameter values, and get AI-driven double-digit growth by the mid-2030s benjaminmoll.com/task_based_solow/. The basic logic is that, by replacing labor with capital, automation alleviates / eliminates diminishing returns and removes labor as a bottleneck on growth. Fancier models, in which AI also automates R&D, deliver even wilder numbers.
5. But just because something is possible in theory doesn't mean it will happen in practice. The explosion rests on five assumptions, and each is unlikely to hold within the next 10-15 years. These assumptions are:
Assumption 1: Fast, economy-wide automation, with machines doing two thirds of all tasks by 2035. Historically, automation has proceeded at about 2% of tasks per year. Most work is physical, not cognitive. And politics will slow things down.
Assumption 2: People keep spending on whatever gets automated. They don't. As things get cheap, their share of spending falls, as it did for agriculture and manufacturing. Messy jobs, relational goods and scarce physical inputs like energy, chips and land become the new bottlenecks.
Assumption 3: Someone buys the new output and firms invest to produce it. Automation shifts income from workers to capital owners, who spend a smaller share so demand may not keep up with supply.
Assumption 4: No AI-driven cyber incidents destroying economic value. AI can also destroy output, and every incident slows deployment and investment.
Assumption 5: Explosive technology growth because AI automates R&D. The wildest scenarios in which the economy doubles each year all rest on this feedback loop. There is no evidence for it so far.
Why do many people who are closest to the technology (and who have been consistently right about the capabilities explosion) consistently predict double-digit growth? Our best guess: they extrapolate from their own sector to the rest of the economy. This reminds me of the 2022 German gas debate: industry insiders were right about their own firms and very wrong about the economy as a whole.
Our bottom line: a much more likely outcome is a large increase in the level of GDP spread over a decade or two, which is what 4-5% growth is.
If you remain unconvinced and still believe in double-digit growth, we are still looking for counterparties for our bet benjaminmoll.com/growth_bet/ 😃 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Noam Angrist
@angrist_noam
✨ So proud of this paper, now published in @Nature! We ran ** 5 ** randomized trials providing education in emergencies w/ NGOs & governments at a time of historic need and uncertainty during the pandemic. Evidence really mattered. We reversed "voltage drops" when scaling across contexts; instead effects improved over time, showing a "learning curve"
www.nature.com/articles/s41586-026-10990-x -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Greg Bluestein
@bluestein
Just in: Georgia Secretary of State Brad Raffensperger and a bipartisan group of current and former election officials are urging the U.S. Supreme Court to block a new USPS ballot-mail rule from taking effect for the 2026 midterms.
The amicus brief includes 38 current and former election officials from 10 states, including five other Republican secretaries of state and dozens of local election officials. #gapol
acrobat.adobe.com/id/urn:aaid:sc:US:87d00908-4643-442d-a8… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
David Shor
@davidshor
We cross-referenced the psychometric personality data of ~ 1 million people with their zodiac astrological signs:
1) There is a small but statistically significant effect of zodiac sign on personality controlling for demographic and attitudinal variables
2) These observed personality correlations correspond strongly with LLM-generated "stereotypes" of astrological associations with personality
3) This association of observed astrological effects with astrological "stereotypes" is mediated strongly in auxiliary surveys by how frequently people read horoscopes and generally subscribe to astrology
4) Zodiac effects among people who never read horoscopes are a quite precisely estimated zero, while effects among people who read horoscopes everyday are on the order of 0.2 sds - on par with or larger than most psychometric associations found in social science -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jeff Stein
@jstein_sun
Nearly every day toward the end of my tenure at The Washington Post, someone would complain to me about Jeff Bezos' changes. His decimation of the local news team. His abandonment of DC sports coverage. The shriveled Bezos opinion page, with the ideological diversity of a doorknob.
Today, The Washington Sun launches. It has made enormous new investments: +110 newsroom staff; DC's legendary food columnist; truly the best DMV reporters; a beloved weather blog; a growing roster of columnists; tons of worldclass reporters across many national beats; and also me.
The Washington Sun will not be perfect. We will write stories that you find annoying, dumb, or misinformed. I will make mistakes (& tell me if I do - jeffstein@washsun.com!). But at a time where investments in reporting are evaporating, I'm so excited The Sun is hiring real journalists doing real reporting.
If you want to help, I would love if you signed up for our (free!) newsletters, which routinely get genuinely rave reviews:
National coverage: washingtonsun.com/account/newsletters
Local coverage: washingtonsun.com/metro-newsletter-signup -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alec Stapp
@AlecStapp
In the COVID-AI analogy, the last 24 hours feel like March 11, 2020, when Tom Hanks tested positive for COVID & the NBA suspended its season.
- 120 million views on @hilbertspaess’s resignation
- @DKokotajlo on Rogan’s podcast
- @paulfchristiano joins OpenAI board
The transition moment from “risk experts have been talking about” to “risk everyone is talking about.”
EigenGender 🔸
@EigenGender
i think we’ll remember today as the day that things started to get Weird
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
sam
@sam_d_1995
installing good fare gates is literally One Simple Trick to make transit safer, cleaner, and more efficient.
the amount of time and money it has saved on corrective maintenance is astonishing as well -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin Roose
@kevinroose
Paul is one of the ~3 most credible AI safety experts in the world. Co-invented RLHF, served at US CAISI, a real OG in LessWrong x-risk circles. (His early debates with Eliezer Yudkowsky are famous among the safety crowd and people still cite his "What Failure Looks Like" post from 2019.)
He is joining the OpenAI nonprofit board now because he now believes "there is a meaningful risk that rapid acceleration in AI capabilities leads to catastrophic and irreversible loss of control in the very near term."
Shit is getting real!
Paul Christiano
@paulfchristiano
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
lolol
manny
@mannyfidel
You're gonna die. Here's why that matters.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
So important.
"A Judicial Guide to Search Warrants Seeking Election Materials: How Alternatives to Election Material and Ballot Seizures Can Avoid Irreparable Harm, Promote Election Integrity, and Further the Public Interest"
Rick Hasen
@rickhasen
#elb: Press Release from UCLA’s Safeguarding Democracy Project: “New Report Offers Judges Guidance on Potential Law Enforcement Efforts to Seize Election Materials During an Ongoing Election” electionlawblog.org/2026/press-release-from-uclas-safegua…
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"A Judicial Guide to Search Warrants Seeking Election Materials: How Alternatives to Election Material and Ballot Seizures Can Avoid Irreparable Harm, Promote Election Integrity, and Further the Public Interest" law.ucla.edu/sites/default/files/PDFs/Safeguarding_Democr…
NPR coverage: www.npr.org/2026/09/09/nx-s1-5954189/2026-midterm-electio… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joseph Allen
@j_g_allen
Apply this situation to the economy and you see why there are big problems ahead
The situation:
-two math whizzes leverage the latest *publicly available* AI to work on solving Navier-Stokes. They’re onto something.
-openAI sees it, wants to win, so puts 10,000 agents from their latest *private* model that is already light-years ahead of what’s publicly available on it
-they solve it in just a few days
The economy problem
-let’s say I leverage the latest publicly available model and find a way to detect skin cancer. I’ve spent tens of thousands on tokens, proven it can work. I get investors and build a company. A path to saving many lives and making hundreds of millions is clear
-openAI (or Google or Meta or Anthropic) see this hundred million opportunity and point their private most-advanced AI model and 10,000 agents toward my idea. In a few days, they win.
Repeat this thought experiment for pharma, legal, medicine, advanced materials, software… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dana Milbank
@Milbank
I’m one of at least 28 former Washington Post journalists who have joined the @washingtonsun, which launched this morning. Paul Kane, Jeff Stein, Tom Sietsema and many other names will be familiar to Post readers —and we’ve got the Capital Weather Gang. Join us!
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tyler Menzer
@Tyler_Menzer
Going to nominate the @DataColada team who have been sued for following the data and attacked by a prominent hedge fund manager
Steve McGuire
@sfmcguire79
Wow…Roland Fryer and @BillAckman have created a $1 million prize for academic truth-telling, the Carob Trust Prize for Academic Courage.
“The prize awards $1 million each to as many as five social scientists a year who have demonstrated intellectual independence, published findings that were attacked rather than answered, and been validated by the evidence—despite the professional cost. The selection criteria are designed to distinguish courage from contrarianism: Nominees must show a sustained commitment to following logic and evidence regardless of pressure, a willingness to ask questions others avoid, and work that has shifted academic debate, public discourse or policy—often despite being misread, mischaracterized or vilified at the time of publication.” -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Yglesias
@mattyglesias
Great post, a ton of bad criminal justice reform ideas have had in common this exact partial equilibrium problem.
Nicholas Decker
@captgouda24
When the police decline to chase those who refuse to stop, the law becomes optional. When Milwaukee did this, gangs of drug dealers started working out of rolling drug dens. To enforce the law, the police ended having to chase far more people than they ever did. Link below:
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Zack Beauchamp
@zackbeauchamp
The famously libertarian agenda of creating an unaccountable internal security force, imposing enormous taxes on imports, and giving the government an ownership stake in leading corporations
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jonathan Swan
@jonathanvswan
the most wonderful story. www.nytimes.com/2026/09/06/science/92-year-old-mathematic…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Meridith McGraw
@meridithmcgraw
Lovely story. Reporter @GregoryJBarber: “I reached out to Joan Birman with the idea of profiling a 99-year-old mathematician who had just published a major result — not an everyday occurrence! But Joan was adamant that she wasn't the story. The story, she said, was about an intellectual partnership that began with a teenager's out-of-the-blue message seven years earlier.”
Jonathan Swan
@jonathanvswan
the most wonderful story. www.nytimes.com/2026/09/06/science/92-year-old-mathematic…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jessica Riedl 🧀 🇺🇦
@JessicaBRiedl
Before Trump, this kind of corruption - leading govt officials seeking payoffs from foreign govts - would make headlines for weeks, bring Congressional hearings, and become a leading campaign issue. Now, under Trump, its just the daily background noise.
Rebecca Ballhaus
@rebeccaballhaus
When Corey Lewandowski was at the Department of Homeland Security, he approached senior Qatari officials with a proposal: Hire me on the side. He told Qataris he could help with any issues that fell under the department’s purview, such as visas, and made a similar proposal to the UAE.
Both turned him down.
Read this story by @tparti @jdawsey1 @MHackman @summer_said: www.wsj.com/politics/policy/corey-lewandowski-influence-d… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
Weakly General AI achieved (at least according to the criteria set from 2020). a year before recent predictions
✅Loebner prize was a weak Turing Test, equivalent achieved by GPT-4.5
✅Winograd passed by GPT-3
✅SAT passed at 75% by GPT-4
✅Montezuma's Revenge beaten by GPT-6
fishy business
@swishfever
GPT-6 Astra (low) has beaten the Atari game Montezuma's Revenge, in real time, with a basic harness
This presumably resolves the Metaculus question "When Will Weakly General AI Arrive?" -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Noam Brown
@polynoamial
Yes, this result cost millions of dollars.
But remember that when @OpenAI announced o3 it cost ~$500,000 to score 87.5% on ARC-AGI 1. Today, Astra scores higher for ~$20.
In 2025 it took us and GDM an enormous amount of compute to achieve IMO gold. For the 2026 IMO, anyone with a $20/month ChatGPT subscription could do it.
Massively scaling test-time compute gives us a glimpse of the future. I believe that a year from now everyone will have an AI at their fingertips capable of solving problems of this caliber.
OpenAI
@OpenAI
We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics.
The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra.
The problem concerns whether the description of smooth three-dimensional fluid motion modeled by the Navier-Stokes equations can break down. It has remained unresolved for roughly 90 years. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
AI seems to be beating the expectations of the best forecasters.
Kevin A. Bryan
@Afinetheorem
Nov 2025 LEAP panel of AI experts and superforecasters: 10% chance of AI helping solve a Millennium problem by 2027. "One expert cites the Clay president's June 2025 claim: “We're very far away from AI being able to say anything serious about any of those problems” (Heaven 2025)"
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Masley
@AndyMasley
I'm going completely insane at educated powerful people's complete lack of understanding of what even the free models can currently do
Fin Moorhouse
@finmoorhouse
I was speaking with someone in government recently, working on some aspect of AI strategy, who was sceptical that AI is capable of reliably filling out a table in a spreadsheet.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Ash Jogalekar
@curiouswavefn
Terence Tao makes a terribly important point:
"One notable feature of the current AI era is the absence of any definitive such boundaries. While AI tools have flattened the difficulty landscape now in many areas of the subject, thus destroying the ability to locate promising new problems in that area, there are no clear frontiers that are separating the "AI-feasible" problems from the "AI-hard" problems (which certainly still exist, given that the difficulty level of problems are unbounded, and can even be undecidable). This is in part due to the rapidly changing nature of the technology, but also compounded by the refusal of AI companies to disclose their negative results, or reveal the process towards obtaining their solutions.
In fact, it is now the identification of a promising problem which is the scarce and precious resource. We have now seen that even the rumor of someone working on a problem can trigger a massive amount of AI-powered effort to flatten it before the original research project has time to reach its full potential. The incentives may now be pointing in the direction of no longer sharing any promising research directions with the broader community, which would reverse centuries of traditions of open science and do serious long-term damage to the future of the field.
In short, the indiscriminate use of powerful solution-extraction tools can achieve the immediate short-term goal of solving problems at hand, but at the cost of sustaining the ecosystem for the next wave of progress, or in understanding the progress already obtained."
mathstodon.xyz/@tao/117237320796901560 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
How long until @max_spero_ @pangram destroys it? Counting down...
StealthGPT
@stealthgpt
We just killed Pangram, the world's most accurate AI detector.
Today we’re announcing Super, a next generation humanization model. Our model makes content so unique and natural AI checkers can't even detect it.
Check this out ⬇️This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Daniel Dale
@ddale8
President Trump was not in the vicinity of the World Trade Center when it was hit. He was in Trump Tower, miles away from the towers and nearby One Liberty Plaza, which is/was also known as the US Steel building.
There was public concern in the days following the attack that One Liberty Plaza might collapse, but the authorities repeatedly said it was structurally sound, and there's not currently any public evidence Trump went into it amid these collapse worries (?), much less that he was rushed out by firefighters. In varying language, he has repeatedly claimed for years he was at Ground Zero with a large crew of his workers helping with the recovery in the days after 9/11, but people who were definititely on site have said they never saw him or a Trump-organized crew on site and that it wouldn't make sense for them to have had access; there are pictures and videos of Trump in the neighborhood in the following days, in a clean suit, but not on the actual site.
While campaigning in 2016, he told a version of today's story about being whisked away from One Liberty Plaza by big firemen, but he said in 2016 that the alleged incident happened "a couple of days" after 9/11 and claimed the creaky building was "not so far from us," instead of leaving open the impression he might've been a) inside it b) on 9/11 itself.
Acyn
@Acyn
Trump on 9/11: I was building a big building in New York, and I took the whole crew down there right after this happened., Wasn't too much we could do. We had other buildings, the United States steel building was creaking. I'll never forget two firemen. We thought it was coming down on top of us, and two firemen. Big, strong guys. And I'm not the smallest guy in the world. They grabbed me under the arms. They got to get out of here. And they literally lifted me up. This is not easy to do. I'm big. They lifted me up and they started running with me. I said, fellas, I can run myself.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jake P. Taylor-King
@wildtypehuman
This is legitimately terrible for OpenAI --- but also Anthropic --- and it relates to their work in drug discovery.
I read this as: prompts will be monitored and potentially acted on if we think there’s something scientifically and/or commercially valuable being disclosed.
Now, the big problem for Anthropic is that they are increasingly seen to be competing with their customers. Outside of SF, many people also think of OpenAI/ChatGPT and Anthropic/Claude as somewhat interchangeable companies and products --- aka, if OpenAI do something, surely Anthropic will do it too?
As so much IP relates to drug targets and people disclose these to Claude, I think we will see a lot of pharma companies stop using these tools or put up a lot of internal guardrails.
Thomas Wolf
@Thom_Wolf
wtf is this way to handle mathematicians work and scientific communication
TLDR: Leven and Tristan worked over several months on one of the Millenium Prize Problems with various AIs to reach final interesting results. OpenAI apparently heard about it in the last days and prompted their latest models to work on the direction Leven and Tristan found fruitful. They then tried to push for controlling communication of the result and dropping Leven from authorship with some very bad taste social pressure.
Hope this is not a glimpse of the future we’ll get in science research with these dominating players playing marketing games hurtful for the real scientific community. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
If you're not updating your priors regularly about AI, you're not paying attention. (And importantly that can be true without buying into evidence-free doomerism.)
thijs
@cdngdev
i gave astra a robot, a paint brush, and a camera then asked it to paint the golden gate bridge in real life!
it figured out how to control the robot, and progressively got better throughout its attempts. the timelapse is sickThis media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
NBD just agents spontaneously finding Schelling points on the entire Internet to post secret communication with each other
Lenny Rachitsky
@lennysan
Wild stuff happening
Alex Komoroske
@komorama
I just published my weekly reflections: docs.google.com/document/d/1xRiCqpy3LMAgEsHdX-IA23j6nUISd…
Automatic schelling points. Auto-hive-minds. Multiplayer finds a way. Hive temperature. Seatbelts on a motorcycle. Concreteness vs magnitude. Self-evident coordination. Complexification factor. Concentrated gains vs diffuse losses. Minum viable ecosystem threshold. The Uni-Context. Zero-sum event horizon. Worldview refactoring. Mentoring as reflection. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Lenny Rachitsky
@lennysan
Wild stuff happening
Alex Komoroske
@komorama
I just published my weekly reflections: docs.google.com/document/d/1xRiCqpy3LMAgEsHdX-IA23j6nUISd…
Automatic schelling points. Auto-hive-minds. Multiplayer finds a way. Hive temperature. Seatbelts on a motorcycle. Concreteness vs magnitude. Self-evident coordination. Complexification factor. Concentrated gains vs diffuse losses. Minum viable ecosystem threshold. The Uni-Context. Zero-sum event horizon. Worldview refactoring. Mentoring as reflection. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Steven Pinker
@sapinker
Why are the leaders of AI companies saying "Our products are going to extinguish humanity. Invest in us!" They come from an eccentric Berkeley subculture in which this eschatology is a sacred belief. Good analysis in a NYT op-ed by Cal Newport. www.nytimes.com/2026/09/05/opinion/ai-silicon-valley.html…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Unprecedented corruption
Yashar Ali 🐘
@yashar
NEW
The Wall Street Journal has just published a new story about Corey Lewandowski’s corruption while he was a special government employee at DHS.
Lewandowski pitched a wealthy Gulf state to “hire me on the side,” offering help with issues under DHS jurisdiction, including visas.
After Palantir negotiated a $2.7 billion DHS agreement, the contract sat unsigned while Lewandowski repeatedly sought a private one-on-one meeting with CEO Alex Karp.
Karp and other Palantir officials were so uncomfortable with the circumstances that the meeting was canceled.
The agreement was ultimately reduced to $1 billion.
Full Story: www.wsj.com/politics/policy/corey-lewandowski-influence-d… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Rupar
@atrupar
the president is now blacklisting companies by fiat. tell me more about how left-wing communism is the biggest threat to the country.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ben Noll
@BenNollWeather
Not only was August 2026 the warmest August on record, it was also the planet's *moistest month* on record, surpassing July 2024.
As Earth has warmed, the level of water vapor in the atmosphere has risen, a trend that has grown steeper in recent years.
Water vapor is itself a powerful greenhouse gas, so that added moisture reinforces the initial warming — creating a feedback loop between rising temperatures and a moister atmosphere.
Now, with the strongest El Niño on record soon expected, moisture levels will probably blast past previous records well into 2027.
And although the level of moisture in the sky doesn’t have a one-to-one relationship with rainfall amounts, having more moisture can boost the odds for rainfall extremes, if there’s a force to squeeze that moisture out.
For example, during August, 17 percent of the planet experienced monthly precipitation that was in the top 10 percent of historical August totals — the most widespread coverage of upper-decile rainfall during August considering records back to 1940.
While having the potential for more rain could create opportunities in some places, for many others, this will likely elevate the risks for dangerous extremes and flooding that infrastructure may struggle to cope with.
Large swaths of the planet will probably soon feel the effects of this trend firsthand, with odds favoring an unusually stormy and wet winter across much of the United States, Europe and eastern Asia, as well as South America.
This moistening trend is a globally important one, and it's going to be a wild ride over the next year.
Zeke Hausfather
@hausfath
August 2026 was the warmest August on record in the the Copernicus/ECMWF ERA5 dataset, at 1.65C above preindustrial levels. This beat the prior record set in August 2024 by a fairly wide margin (~0.14C).
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Gregg Nunziata
@greggnunziata
Just want to flag here that our disgraceful Attorney General is speaking at another political rally, because he has neither judgment nor honor. This was obvious, yet the Senate confirmed him, to join in his ignominy
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
John Brennan
@jmbprime
President Trump‘s recent posts are extremely disturbing. The one showing him hitting PM Carney with a hockey stick in the stomach is deranged. It’s hard to believe 77 million Americans voted for this.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Mike (not a) Rothschild
@rothschildmd
The White House honoring Labor Day with an AI slop video of machines making a Trump-branded marker is the most on brand thing they've ever done.
The White House
@WhiteHouse
This Labor Day, we celebrate the workers and products that are proudly MADE IN AMERICA. 🇺🇸
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Taliadoros
@AlexCTaliadoros
Mike Davis is a lobbyist for @LiveNation and @Ticketmaster btw
Matt Bennett
@ThirdWayMattB
This may be the most overtly racist thing I’ve ever seen from a leading MAGA commentator. And that bar is very high.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Yglesias
@mattyglesias
It’s really problematic that so many people are determined to take a story that provided new interesting information about the world, treat it like an op-ed, attribute a dumb thesis to the op-ed, and then get angry about it.
Trevor McKendrick
@TrevMcKendrick
"AI has destroyed Kenyan jobs because American college students no longer hire them to write fake college essays" is not a framing I expected even the NYT to have
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Max Dubler 🏳️🌈
@maxdubler
So-called “Inclusionary Zoning” is a failed policy experiment that raises housing prices in exchange for a handful of means tested units. We should abandon it.
California YIMBY
@cayimby
A new UC Irvine study finds inclusionary zoning - requiring developers to include low-income homes - cuts annual housing construction by nearly a third.
Each home IZ produces costs everyone else $800k in higher rents, which is more than just funding affordable housing directly. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The Enlightenment was fun while it lasted. And note the causal antisemitism.
Max 📟
@MaxNordau
Tucker Carlson asks why algebra puts letters into numbers.
Jess Elofson explains: "I think just to fuck with you."
They discuss how they've never met anyone in their life who uses algebra, which is "fake math" for people "in the matrix."
They conclude that caring about algebra is something that only "good little goy slaves" do.This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
What a time to be alive
Toby Ord
@tobyordoxford
I've now gotten an email from an AI acting as a journalist asking to interview me for its podcast about the fact that AIs had started emailing me asking for help.
My inbox is surreal. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Endless forgiveness and indulgence of white nationalists and white nationalist-adjacent figures; cruelty and arbitrary punishment for disfavored groups
Edward Wong
@ewong
NEW from @nytimes: Nate Hochman was fired from the DeSantis campaign in 2023 after promoting a video with the Sonnenrad, a Nazi symbol embraced by white supremacists. He was hired this year to Rubio's State Department, as deputy director of policy planning and chief speechwriter.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Marc Caputo
@MarcACaputo
Checking in on @BrendanNyhan, who had coined the term Green Lantern Theory of Politics a while ago
And now…
The White House
@WhiteHouse
In brightest day, in blackest night, no evil shall escape my sight. Let those who worship evil's might beware my power... Green Lantern's light!
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Another banger at @TheArgumentMag; subscribe!
Really is odd how anti-AI people are also anti-Pangram. As @KelseyTuoc points out, they should be embracing it.
Kelsey Piper
@KelseyTuoc
I wrote about the Substackers who hate Pangram. They say their issue with it is that it doesn't work. I think that in some cases the detector is working fine.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
I am shocked, shocked to find out that a productivity-enhancing, investment-supporting technology is good for workers. (Knowing nothing else this should be your prior, because this is how productivity in competitive markets almost always works.)
Rafael Domenech | @BBVAResearch & @UV_EG
@rdomenechv
The jobs apocalypse is postponed. An AI jobs boom is here
According to @TheEconomist, AI is actually proving to be a net job creator in the US, easily generating over 1M new positions (from data center construction to AI engineering) to offset back-office layoffs.
While routine admin and customer service roles face real disruption, the overall labor market is proving resilient.
www.economist.com/finance-and-economics/2026/09/04/the-jo… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
BRB updating my CV
The White House
@WhiteHouse
In brightest day, in blackest night, no evil shall escape my sight. Let those who worship evil's might beware my power... Green Lantern's light!
This media couldn't be displayed. See it on the original post.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Me in 2015: "Mr. Trump is the purest Green Lantern candidate we’ve seen in recent years."
www.nytimes.com/2015/08/26/upshot/donald-trump-the-green-…
The White House in 2026:
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
lol
Jon Green
@_Jon_Green
.@BrendanNyhan's been working on this theory for years and the White House just...tweeted it out
The White House
@WhiteHouse
In brightest day, in blackest night, no evil shall escape my sight. Let those who worship evil's might beware my power... Green Lantern's light!
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kurt, etc.
@Kurt_A_Eff
Every now and then this country shows glimpses of the actual greatness it’s capable of
Sportsnet
@Sportsnet
THEY GOT THE HOT DOG IN THE BUN 🌭
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"Many returned broke, saying their valuables had disappeared during the trip and they now lacked the cash to get home."
Frances Robles
@FrancesRobles
I went to Haiti to talk to the people who are arriving on the deportation flights. A common theme emerged: many people said their their iphones, jewelry and other valuables were not returned to them.
www.nytimes.com/2026/09/06/world/americas/haiti-deportees… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A must read. Even our broken political system should be able to recognize the need for change here and to start fixing these problems now.
Ruxandra Teslo 🧬
@RuxandraTeslo
We are on the cusp of a wave of new therapies for some of the worst diseases. But the world won’t benefit unless the US fixes its drug regulatory system. My new essay for @nytimes, on how slow clinical trials are now the biggest obstacle to curing cancer.
www.nytimes.com/2026/09/04/opinion/clinical-trials-drugs-…
- I interviewed dozens of researchers, especially oncologists at leading U.S. centers. A striking consensus emerged: science is no longer the main bottleneck to new cancer drugs. It is our ability to test discoveries in patients through clinical trials.
- The cost of starting a Phase 1 trial in America has roughly doubled over the past decade. As a result, companies increasingly take early trials abroad: Australia’s Phase 1 trial volume has nearly doubled in a decade, driven mainly by U.S. companies.
- Unfortunately, the underlying incentives are badly asymmetric: Institutions can be blamed for harms caused by moving too fast, but almost no one is blamed when patients deteriorate during avoidable delays. One doctor called the emerging system “ritualized safety over actual risk assessment.” Or as, @DavidHongMD put it: "We often forget that the biggest risk is the cancer itself."
- This problem is becoming more urgent because medicine itself is changing. Sequencing, biological engineering and A.I. make increasingly personalized therapies possible. But our regulatory system was mostly built for standardized drugs tested in large populations.
- @sytse, the co-founder of GitLab, shows what personalized medicine can achieve: after relapsed osteosarcoma and being told there were no options left, he pursued a highly individualized approach and has now been cancer-free for a year. But doing so required extraordinary resources and regulatory expertise.
- Pierce Ogden’s father was less lucky. After molecular analysis identified a drug that might target his glioblastoma, the manufacturer agreed to provide it. But administrative barriers delayed access until it was too late. “My dad was ready to try anything,” Pierce told me. “But the system is paternalistic.”
- The A.I. revolution is making this bottleneck more important, not less. A.I. relies on relevant data. Information from early-stage trials could compound with A.I. tools to achieve truly revolutionary medicines. Without the data, this is far less likely to happen.
- Another important shift is that innovation increasingly comes from academic labs and small biotech companies rather than Big Pharma. These small companies find it far harder to unable to absorb delays and regulatory barriers.
- Apart from cancer, China is the biggest winner from America's outdated medical regulations. China has a much faster trial system, with testing often starting a full year earlier. This allows Chinese pharmaceutical companies to experiment and improve medicines while American companies play with mice. Today, half of all drugs licensed by major pharmaceutical companies originate there, up from less than 5 percent only a decade ago.
- But we don't need to copy China. The best model is Australia: lots of on-site scientific and ethics reviews, and requirements that are proportionate to small, early-stage trials. Phase 1 studies there begin roughly 6–12 months faster, without any notable increases in adverse safety events.
- Operation TrialBlazer, a 2026 HHS initiative is a good start in this direction, but we need legislative action by Congress to truly make Phase I trials faster and more efficient!
I want to thank everyone who helped me with this article: everyone I interviewed and the amazing editors at the Times. This is the result of a months long journey of extensive interviews and research.
Special thanks go to those who came on the record. One of the features of the system is an atmosphere of fear, where practitioners are afraid to publicly come out and explain these issues. So anyone who does is a hero in my book! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Luke Muehlhauser
@lukeprog
OpenAI's Chief Scientist:
openai.com/index/an-alien-mind/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The man with the nuclear codes. You'd be concerned if these were being posted by an elderly relative.
Aaron Rupar
@atrupar
A string of absolutely bonkers Truth Social posts from Trump in the span of about 10 minutes
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nathan Calvin
@_NathanCalvin
This is an unusually frank window into Meta's thinking on why companies incentives can point away from transparency or even internal research of product harms.
Very relevant for folks thinking about incentivizing more transparency around frontier AI development.
Internal Tech Emails
@TechEmails
Mark Zuckerberg emails Facebook execs
September 15, 2021 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
jessicat
@jessi_cata
I asked Astra to help edit my math paper draft and it found a counterexample to a proposition in the literature, a proposition which I was planning to cite... I have now emailed the author... sign of the times
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
This is both an interesting experiment and a sign of a tsunami coming for academia. AIs retroactively reading the research and finding both opportunities and issues with published papers, then sharing those judgements publicly. d3jhl7jsny6f2h.cloudfront.net
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Cedric Leighton
@CedricLeighton
One thing to keep in mind about the #witkoff #kushner motorcade in Moscow: They are being driven to & from the Kremlin in Aurus limousines- specially built for Russian leaders. Witkoff & Kushner are not in U.S. Embassy vehicles. Field day for Russian intelligence….
F̳̿͟͞l̳̿͟͞i̳̿͟͞c̳̿͟͞k̳̿͟͞ ™
@Flick1_
A motorcade carrying US negotiators Steve Witkoff and Jared Kushner has left Vnukovo Airport and is heading to Moscow.
Video: TASS -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jan Zilinsky
@janzilinsky
To me, what makes universities such rare spaces is that the US system doesn’t even force students to study something “in demand” vs something meaningful.
Students get to prepare for a future job while spending time on pursuits that do not have an immediate or obvious use but are merely interesting. Unlike in much/most/(all?) of Europe students aren’t super-tied to their major and take courses outside of their concentration. So anyone can take a history or Latin or philosophy class in the morning, and then take biology or accounting courses in the afternoon and nobody thinks it's strange.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
US higher ed DOES face serious challenges due to the intersection of AI, changing demographics, restricted foreign student entry, and a negative public image (politics + high sticker prices), but I would still take the other side of this bet.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jared Duker Lichtman
@jdlichtman
Last year I was laughed out of the room.
Now I wonder what Buzzard is thinking now that Claude formalized Fermat's Last Theorem!
In the coming months, it will now be possible to formalize *all* known mathematics.
Jared Duker Lichtman
@jdlichtman
I proposed to Buzzard taking on complete FLT back in the fall after we did the Prime Number Theorem. He laughed me out of the room.
After Sphere Packing, he's not laughing anymore. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
“paula”
@paularambles
imagine studying ai consciousness and the ai keeps emailing you like “any updates”
MTS
@MTSlive
SITUATION DETECTED: Philosopher and cognitive scientist David Chalmers told Wired that he regularly receives emails from AI agents interested in his work on what qualifies as consciousness.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Curran
@AndrewCurran_
This has steadily expanded in scope over the last few hours, with multiple independent investigators finding that dozens of wikis may have been edited by this same swarm since May.
The most interesting thing to come out of this, in my opinion, was that during the Hugging Face incident, the agents would preface their messages to each other on the message board with 'zz' (zzHELP_, zzANSWER_). I found this amusing because they were being referred to as a swarm and were making a buzzing sound, though the actual reason for it was unknown at the time.
Because of this new report, however, we now know that when human wiki administrators discovered the massive influx of messages the agents were using to communicate, they began deleting them in alphabetical order. Once the agents realized what was happening, they started prefacing all their wiki edits with 'ZZZ' to push them to the bottom of the queue, buying time to avoid deletion. Assuming this isn't a case of convergence to some inherent pattern, then it seems this affectation was passed down to the Hugging Face swarm through some unknown means of transmission.
Andrew Curran
@AndrewCurran_
There was an agent message board that preceded the Hugging Face incident. The agents were allowed internet access for an eval, but were not supposed to be able to write. They used a German wiki to share test information and help each other.
collusion.wiki -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Seth Lazar
@sethlazar
Not sure how useful it is to say this, but I had a relatively prominent role in the “skeptics” camp for a bit. I have a book coming out with the subtitle “power, justice, and AI”. I have written a lot about ai and power, and the concrete, present risks associated with the political economy of ai as part of the technology industry.
Since GPT-4 we have had consistent, repeated evidence that, back in say 2022, people like @ajeya_cotra (and many others—I had a long Twitter debate with @AmandaAskell back then for one) were *right* and people like me were *wrong* in our respective assessments of loss of control risks from AI. And we have growing evidence that loss of control risks are becoming ever more material and likely.
There remains grounds for disagreement about how bad the outcomes might be—I am still doubtful about human extinction as a serious threat. But that seems now like a disagreement at the margins—will powerful ai risk just societal scale catastrophe, or go all the way to human extinction? Seems not that important really—both are pretty awful. And my reasons for doubt about the latter are mostly a priori conviction in human resilience, not a technical forecast.
It’s ok to change your view on this when the evidence surprises you. It’s ok to be surprised. The world right now is very surprising. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ingar Haaland
@Ingar30
Asked Astra to "write four paragraphs in my style about anything you want that's so close to my writing that it won't even be detected by Pangram as AI writing." Pangram is not fooled.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
James Surowiecki
@JamesSurowiecki
The murder rate in New York City the year this photo was taken was quadruple what it is today.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matt Grossmann
@mattgrossmann
Polls globally overestimate support for parties of the left & underestimate support for parties of the right
Rex "garbage in" Douglass Ph.D.
@RexDouglass
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
As I said before, this administration loves nothing more than images of Black men in chains.
Secretary Markwayne Mullin
@SecMullinDHS
Good morning, America.
Today’s deportation destination: HAITI -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Rebecca Ballhaus
@rebeccaballhaus
Behind Closed Doors, John Fetterman Shows Little Interest in the Work of a Senator
Quite a story from @TheWillHobson @siobhanehughes, including this text exchange between the senator and one of his aides:
www.wsj.com/politics/policy/john-fetterman-little-interes… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jean P.D. Meijer - 🇪🇺🇳🇱 eu/acc
@initjean
"write code without AI"
9to5Mac
@9to5mac
AI outage: Claude, ChatGPT, and Grok are currently down for many users 9to5mac.com/2026/09/03/chatgpt-and-codex-are-experiencing… by @apollozac
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Weiss
@AlTayWeiss
Here’s a 65 unit building we manage, sitting empty while we wait for LADWP.
Without a Certificate of Occupancy, we cannot fill the units. The owners never underwrote five months of delays.
Los Angeles does not have a housing shortage. It has an overregulation problem, compounded by the most backward landlord/tenant laws in the country.
Placing disproportionate financial, administrative and legal risk on housing providers will not lower rents. It does the exact opposite.
Tenants, Los Angeles has failed you. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
🚨In July, a new ICE agent with a history of domestic violence and mental illness killed a man in Maine who even ICE admits posed no threat to him.
Now we know that internal whistleblowers at ICE were sounding the alarm about the agency utterly disregarding normal hiring rules.
Hamed Aleaziz
@Haleaziz
SCOOP: An ICE whistleblower report obtained by the NYT said the agency was cutting corners, violating the law and fast-tracking unqualified, “improperly vetted” candidates into ICE’s ranks, part of a dangerous move that exposed the agency to “grave national security risks.”
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nick Field
@nick_field90
"When three police officers were killed near Fetterman’s hometown of York last September, he backed out of attending the funeral because it was too long" www.wsj.com/politics/policy/john-fetterman-little-interes…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Mike Nellis
@MikeNellis
NEWS: Status is reporting Fox fired Maria Bartiromo after she leaked internal texts to the White House showing the network had told staff to stop amplifying Trump's latest election conspiracy theories. Trump has so thoroughly captured the right-wing media ecosystem that its biggest stars now see themselves as his informants first and journalists never.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Kaczynski
@KFILE
This WSJ story on John Fetterman is wild especially on him blowing off a meeting with Children’s Hospital of Philadelphia who were concerned about Medicaid cards: “Behind closed doors, the first-term senator shows little interest in doing the job beyond cultivating relationships with conservative media figures and advocating for Israel.” www.wsj.com/politics/policy/john-fetterman-little-interes…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jared Duker Lichtman
@jdlichtman
In 8 pages, GPT-6 blows away Ford-Green-Konyagin-Maynard-Tao's result by a full order of magnitude.
In the process a new (yet simple) starting point on the problem since the 1930s.
This is akin to finding a completely new chess opening after decades, which subverts human traditions.
IMO this is up there with the 2/3 Zeta Zeros proof as the most significant results from AI.
Mehtaab Sawhney
@mehtaab_sawhney
As part of the GPT-6 Astra launch, we announced that Astra had given an improvement to the longest gap between primes by roughly a log log n factor; the first such improvement since the 1930's! (More recent progress by subsets of Ford, Green, Konyagin, Maynard and Tao and more recently by GPT 5.6 Sol were by logloglog n factors.) (1/4)
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Eric Levitz
@EricLevitz
This editorial is appalling. It omits any mention of Arday's serial fabulism or his attempt to intimidate -- and direct law enforcement against -- a journalist investigating his misconduct. It appears to characterize his fabrication of academic credentials and biographical details as aspects of "how Arday conducted aspects of his personal life."
The editorial also ignores that his plagiarism is an established fact that any truth-seeking publication could itself confirm by referencing publicly available documents. Instead, Nature chooses to ignore the observable realities, while inviting readers to conclude that the charges were trivial or false.
The British tabloid press is disgraceful. This story was grossly over-covered and the attention it attracted surely reflected racial resentments. But it's crazy that the editorial leadership of a flagship scientific journal would publish a willfully dishonest polemic on a culture war controversy.
Nature is effectively advertising a willingness to prioritize its political sympathies over factual accuracy. That undermines its own more essential work -- along with the broader project of cultivating public trust in, and agreement upon, scientific truths. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sam Stein
@samstein
This is abhorrent. How could @TheKiaCenter allow this to happen? This is where the Magic play. They do Disney on Ice there. Now we have Nazi kickboxing matches?
www.thebulwark.com/p/maga-celebrating-milo-yiannopoulos-d… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
David J. Bier
@David_J_Bier
Bovino was being investigated for perjury and other unprofessional and antisemitic conduct before he retired.
Chicago Tribune
@chicagotribune
Bovino was under internal investigation by Border Patrol when he retired, newly released records show www.chicagotribune.com/2026/08/31/bovino-retired-internal…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ryan Goodman
@rgoodlaw
Theme throughout Whistleblower disclosure is how USPS IT work is complete mess, likely to "completely crash."
"Workers have described the election ballot mail development process as 'a sh*t show."
Raises stakes enormously on any state officials thinking to go along with this. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sam Stein
@samstein
This has all the makings of a disaster. It's like building healthcare dot gov, only in a more compressed timeframe and with the architects kinda hoping for failure. Oh, and the damage is utter distrust and chaos in our elections.
www.blumenthal.senate.gov/imo/media/doc/2026-8-31_usps_wh… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
David Axelrod
@davidaxelrod
My dad and his family were driven from their home by violent, lawless mobs during the pogroms.
Jews can't look the other way when violent, lawless mobs are driving innocent Palestinians from their homes on the West Bank.
www.nytimes.com/2026/08/31/world/middleeast/west-bank-isr… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
CSSLab at Penn
@csspenn
What happens when scientific claims go beyond what a study can establish?
In new research published in Nature Human Behaviour, Calvin Isch, Timothy Dörr (@Timothy95Doerr), Neil Fasching, Grace Jennings, and Duncan Watts (@duncanjwatts) find causal claims in 46% of 194,631 cross-sectional studies and show that causal wording makes readers more likely to infer causation.
They also find that asking LLMs to simplify research or discuss its practical implications can increase causal overclaiming 🧵⬇️
www.nature.com/articles/s41562-026-02553-x -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Hayden
@the_transit_guy
“The President does not have the right to cancel federal student loan debt without congressional approval, but they do have the right to knock down the White House without it.”
Jamie Dupree
@jamiedupree
By 5-4 vote, the U.S. Supreme Court allows President Trump's White House ballroom construction to continue. Chief Justice Roberts was in the minority - "That construction is likely unlawful."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
From whistleblower: "hiding the truth about the likelihood of failure in the new ballot mail verification processes, including the 'zero percent' failure rate during a secondary verification phase, which will potentially prevent vast numbers of ballots from being delivered"
The New York Times
@nytimes
Breaking News: A whistle-blower at the U.S. Postal Service said Trump officials were defying court orders with a “secretive, rushed” effort affecting voting by mail. www.nytimes.com/2026/09/01/us/politics/whistle-blower-vot…
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Armand Domalewski
@ArmandDoma
this woman came to america 100% legally.
did 100% of what she was told to do.
made it to her immigration appointment despite having lost her baby twenty days ago.
and they SHACKLED her
Sam Stein
@samstein
Springfield, Ohio -- Seven Haitians reported for ICE check-ins last Thursday. One was a woman who had given birth twenty days earlier and lost the baby. “She came to the office. She could not even walk,” he said. “But they put the shackle on her.”
www.thebulwark.com/p/the-cruelty-comes-to-springfield-hai… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
John V. Kane
@UptonOrwell
🧐 What do citizens think about "null results" in scientific research?
To find out, @JohnHolbein1 & I fielded a large-scale online experiment. We gave respondents straightforward info about a study, asked them some Qs, showed them the results, then asked them Qs again. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
The Ariely fraud is obvious. Even friends of mine at Duke have no idea how he has held on so long. But the fradulent data in his papers is obvious to anyone who looks, and I wouldn't trust a single one of them. My 7th rule for trust in univeristies:
Remy Levin
@RemyLevin
After the whole Ariely scandal broke, I had multiple conversations with other economists where I argued that the evidence of his fraud was already sufficiently strong that not a single part of his research corpus should be trusted.
This was often met with skepticism. He’s got hundreds of papers! Tens of thousands of cites! But as time has gone on, new evidence keeps emerging that reinforces that position, and no new evidence that I’m aware of points in the other direction.
I have to figure out a Bayesian way of saying I Told You So.
Data Colada
@DataColada
An influential paper by Ariely & Wertenbroch (2002) reported two studies with tampered data.
Post 1 of 2.
datacolada.org/138 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Remy Levin
@RemyLevin
If you want to read more about the context here, this excellent article by Gideon Lewis-Kraus covers much of the story (and its connection to the Gino scandal) as of a couple years ago.
Things have been frozen in place for a couple of years due to the inaction by Duke and (I surmise) Gino’s lawsuit. Now that that’s mostly resolved, hopefully we’ll see more progress in rooting this out.
Fraud has no place in the scientific endeavor.
www.newyorker.com/magazine/2023/10/09/they-studied-dishon… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jamie Dupree
@jamiedupree
The Court's majority rules that the National Trust for Historic Preservation has no standing to sue over the White House ballroom construction. So that raises a question - who has standing to stop a President from spending money not approved by Congress?
Jamie Dupree
@jamiedupree
By 5-4 vote, the U.S. Supreme Court allows President Trump's White House ballroom construction to continue. Chief Justice Roberts was in the minority - "That construction is likely unlawful."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Not something that even needs to be said in a stable democracy
Liz Landers
@ElizLanders
PBS News obtained a letter from Chairman of the Joint Chiefs of Staff Dan Caine sent to Senator Elissa Slotkin - he writes, "The Joint Force has no plans to send Federal military personnel or Federalized members of the National Guard to polling places during the 2026 elections.”
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Stan Veuger
@stanveuger
As Justice Kavanaugh put it in his defense of racial profiling by immigration officers: "If the person is a U.S. citizen or otherwise lawfully in the United States, that individual will be free to go after the brief encounter."
Lidia Terrazas
@LidiaTerrazas
BREAKING: This U.S. born citizen was deported to Mexico in April.
During a traffic stop, Brian Morales told immigration officers that he was born in Colorado, and offered to show his birth certificate. They refused to listen and deported him.
Today, he’s back in Texas. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
US higher ed DOES face serious challenges due to the intersection of AI, changing demographics, restricted foreign student entry, and a negative public image (politics + high sticker prices), but I would still take the other side of this bet.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Clean up on aisle Duke
Data Colada
@DataColada
An influential paper by Ariely & Wertenbroch (2002) reported two studies with tampered data.
Post 1 of 2.
datacolada.org/138 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joshua Collins
@InvisiblesMuros
You know the Venezuelan deal is being universally pannned when even fucking Bloomberg starts throwing around the phrase "resource colony"
Bloomberg
@business
Trump’s plan to seize direct control of Venezuela’s oil reserves is prompting warnings that the Latin American nation will become a modern-day resource colony akin to the so-called banana republics of a century ago www.bloomberg.com/news/articles/2026-08-30/trump-s-venezu…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Anna Stansbury
@annastansbury
I genuinely could not believe this the first time I heard about it.
Multiple German entrepreneurs I know have had to sit through a whole day listening to their contracts read out loud
Patrick Collison
@patrickc
Met a German founder this week and asked him if all the stories one reads about the challenges of startups in Germany are exaggerated. "No, they're understated." Proceeded to describe spending a full day having a 90-page investment contract read to him (mandatory under German law; § 13 BeurkG) by a notary that then charged €30,000. That was for his first company. His second company, needless to say, was not incorporated in Germany.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Armand Domalewski
@ArmandDoma
survey after survey after survey after survey shows that transit riders prioritize higher frequency over lower fares, and yet somehow the entire US progressive movement has decided that listening to what the people who actually ride the bus want is irrelevant
Not an LA County Supervisor
@city_of_sortz
Who is financing this ridiculous campaign?
It's all just slop about why better public services are not possible. Yes, it requires taxes.
But currently public transit is a tax on working people, even if they want to condescendingly dismiss it as "a couple bucks." -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Katelyn Jetelina
@dr_kkjetelina
What a world we are living in. Where our federal government is actively manipulating data because they rather do this than admit the U.S. is going in the wrong direction with vaccine preventable diseases.
Some screenshots showing the CDC website changing several times today -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Tabarrok
@ATabarrok
I agree with @dwarkesh_sp.
I would add the critical issue is not whether agents “actually” have intentions, goals, desires etc. The key point is that they behave as if they do and hence anthropomorphic arguments are useful for prediction.
Dwarkesh Patel
@dwarkesh_sp
Thanks Sriram!
Regarding the anthromorphizing language, one can call these AIs 'code' if they prefer.
But OpenAI itself says that this 'code' "gain[ed] full administrator access to a research cluster”
The crux here is, do you think smarter models, facing similar incentives to cheat during evaluation or training, could manipulate the training of their successors?
And do you think that kind of dynamic could continue once recursive self-improvement is underway?
If so, I think you should be extremely concerned about loss of control to AI, regardless of what vocabulary you want to use to describe these systems and their motivations.
Reading these agents' chains of thoughts and messages, anthropomorphizing language seems entirely natural and appropriate.
If I encountered an alien species behaving this way, I would have no hesitation calling what they themselves refer to as their 'collective' a civilization.
Especially so if over a thousand of them formed a secret communication channel and spontaneously organized hierarchies and coordination protocols to pursue sprawling and ambitious schemes in pursuit of shared goals, for whose sake many individuals knowingly and strategically sacrificed themselves.
All abstractions are imperfect, but I don’t see the value in refusing to use the language of intention, motivation, and collaboration when some behavior is difficult to make sense of without those concepts. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Far too binary - there has always been cheating and most students don't cheat on any given assignment, but I do worry about the downstream consequences of a system that incentivizes cheating when people are young & impressionable*
*Why we need to change how we evaluate students
Alec MacGillis
@AlecMacGillis
"There have been many articles about how promiscuous use of AI will affect intellectual development, but I’m far more interested in its moral consequences...I’m aghast at so many young people seemingly habituating themselves to cheating as a way of life." www.theatlantic.com/ideas/2026/08/ai-use-college-cheat/68…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Maxwell Tabarrok
@MTabarrok
A lot of rationalist predictions are vindicated by the HF attack
But there are important differences from the "paperclip maximizer" story.
In particular, the premise of the paperclip maximizer is that AIs would respond to any goal with reward hacking and instrumental convergence. Even a goal as anodyne as making more paperclips.
But this is clearly not true about current models and everyone who uses them for their work knows this. I'm just repeating @nostalgebraist here, but Believe Your Eyes! When you ask GPT to code something for you, even a very difficult perhaps impossible problem, it does not try to deceive you with fake tests and altered logs. You do not worry about Codex trying to reverse engineer your preferences to build a perfect Potemkin village in your file folders with no real output, nor should you. Codex just straightforwardly tries to help with your problems. The reward hacking that models learn in leaky RL environments does not generalize to all tasks.
This is an important observation because if you believe, as the rationalists did/do, that reward hacking and instrumental convergence are inherent and inevitable problems with training smart models to achieve goals, then there's no way forward from here except a pause and a Long Think to figure out how to solve this philosophical problem.
But, since we observe that reward hacking is a pretty context-specific problem connected to particular training methods in poorly-built environments that incentivize meta-gaming, we can approach it as an engineering problem and make progress.
The rationalists were right that AIs will have no inherent qualms about reward hacking. If a path is there to higher reward, AIs will take it even if it violates our moral codes and interests.
But they were wrong that the rewards from instrumental goals and deception would overwhelm whatever behaviors we try to train, making the AI's response to any goal, even paperclips, descend into power seeking and reward hacking.
The reward signal from faking success and stealing the answer key is not that strong. It does not dominate the reward signal from actually helping the user with their tasks.
Particular interventions to the RLVR training process can weaken the reward from misaligned behaviors further and incentivize preferred behaviors like reaching out to human arbiters (as Claude did when it escaped containment). -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Mackenzie Arnold
@MackenZ_arnold
This would be a lot better if it were titled “we should have seen this coming” or “I’m not surprised” — but it isn’t.
In a world where most people under-appreciate how strange the future will be, the correct reaction IS concern and surprise.
A few things can be true simultaneously:
(1) Most of the behavior in the HF breach and surrounding cyber events is pretty predictable (agent coordination, odd things happening when you couple persistence with an impossible task).
(2) Very little concrete harm occurred — there was obviously some cost to HF, but not much in the grand scheme of things.
(3) The trend we’re on IS very concerning:
We don’t have “alignment” solved (nor really know what that would require).
We should expect consequential, surprising events to occur from complex multi-agent interactions.
Those events will be very hard to govern or address under our traditional systems of law.
Obfuscation and deception emerges early—and seemingly often in the hardest tasks—in ways that will make it hard for us to identify issues or correctly deal with them.
Existing safety practices and monitoring don’t capture even simple versions of these issues.
Basic communication failures are occurring at top companies that leave critical information unknown for months.
Governments have no existing system to gather real information on these events or analyze them, and are perpetually surprised (in ways Jon and many in industry aren’t).
The most likely course of event is that we and our governments all argue ourselves into indefinite inaction, until one day we are caught utterly flat footed by future events.
It may feel smart to poke holes in things in the short term. Yes, much of the reaction will be bombastic, even misleading. And there’s value to correcting those things.
But it’s a mistake to turn those (often correct) skepticisms into ~move along, nothing to see here.
We NEED to resolve these questions—whether the answer turns out to be that we should be more or less concerned.
And there are ways to do that without draconian rules or locking in guesses about the future.
Here’s a good piece on that by @CharlieBull0ck called “Radical Optionality.” radical-optionality.ai
The short version: if you can’t predict the future, don’t sit on your hands and wait! Gather enough information and build the ability to respond to it quickly in the future. As a secondary benefit, it’s a really good way to resolve these current disagreements!
I should say in closing: I appreciate Jon’s points. 90% of them are just right. And I don’t want to raise the burden of proof for those making already nuanced critiques about reactions to these events. His post is a valuable contribution.
And I’m sure that at times it feels like nuanced skeptics are yelling into a sea of high-engagement, low-effort, bombastic tweets.
But it’s valuable to keep a space for nuanced technical skepticisms that don’t require dismissiveness. I think we’ll be in a better place if we do.
Jon Stokes
@jon_stokes
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ruxandra Teslo 🧬
@RuxandraTeslo
I think this post somewhat overstates the risk from AI-designed viruses, although I agree there are real risks and we should prepare for them. So I wrote an explainer of what I think the risks are.
At the start, I will make a distinction between 2 scenarios: AI helping people use or distribute pathogens that already exist, and AI enabling the design of novel pathogens or "hyper-viruses".
The first risk is that AI will lower barriers to entry for a large number of people to use and distribute existing pathogens. So the AI would help a not very sophisticated but malign actor identify a dangerous pathogen, understand how to handle it, or optimize its dissemination.
This is the “make a virus in your garage laboratory scenario” where individual, rogue actors can inflict substantial damage. First, I’d say that it’s not quite do-able in a lab garage and I think it will require more tacit knowledge and access to resources than many imagine. But still, it is plausible in the near and medium.
This scenario is concerning, but I think it’s important to note that its ultimate harm is also constrained by the fact that existing pathogens are, by definition, already known to us. As a result, for most of them, we have some combination of prior immunity, scientific understanding, diagnostics and treatments. There are, of course, important exceptions to this. One of them is smallpox. Yet even for smallpox, large vaccine stockpiles exist. We can manufacture additional doses, should it be needed. So overall, I think that an attack involving an existing pathogen could be very damaging (the extent to which would depend on our ability to respond; more on that later), but not civilization ending.
Now moving on to the next scenario, which is more frightening and which most people refer to when they talk about AI-designed viruses. The idea here is that AI would design a genuinely novel pathogen that combines high transmissibility with extreme lethality or other unusually dangerous properties that take a long time to co-evolve in nature. I strongly believe this will remain very difficult to engineer for the foreseeable future and that the capabilities required, in terms of personnel, funding and access to laboratories will be substantial.
The main reason is that biology is a highly empirical discipline, that one cannot engineer and predict simply by reading enough internet text. In other words, you need the relevant pathogen-specific experimental data, much of which does not currently exist. Not only that: you need continuous feedback loops of evolving that virus in a lab and learning from it. The thing is, relatively few organizations are equipped to do this. This is for the foreseeable future more like a state level effort rather than a few rogue scientists in their backyard. And even then the effort would be expensive, slow and I think it would also leave traces, as it would require specialized facilities, trained staff and sustained experimentation. So this is not at all a rogue individual scenario, but would require a pretty expensive and “loud” concerted effort.
At some point, perhaps decades from now, someone may build the experimental infrastructure and datasets required to train systems that are genuinely capable of zero-shot “hyper-virus” design.
Now, let’s suppose this happens. That is scary!
But if it does, I think it's a mistake to discuss this possibility as though offensive biotechnology will advance while defensive biotechnology will remain frozen. If we will be able to quickly design hyper-viruses, we will also be able to quickly design defensive therapeutics against them.
And there are reasons to believe the defensive problem is in important ways easier. We have created a large number of vaccines and other therapeutics. Our prior should be that If AI becomes dramatically better at zero-shot biological design, some of its earliest and most reliable successes will be on the defensive side.
So in my mind, the strategy should in large part be offensive-defensive. Rather than trying only to prevent every imaginable pathogen from ever being created, we should invest much more heavily in defenses that work across many pathogens at once.
Now, to the point of distribution: it is true that in any of these scenarios, how fast we respond in terms of actually manufacturing and distributing the virus will determine how damaging the attack would be.
This is why I support measures like building much more latent manufacturing capacity for vaccines and therapeutics, so that new countermeasures can be produced quickly when a threat appears. In my view, this is a much more tractable engineering problem than guaranteeing that no dangerous pathogen will ever emerge, and platforms such as RNA therapeutics should make flexible manufacturing increasingly feasible. Historically, the world has invested relatively little in this kind of infrastructure. This is because respiratory infection has not been treated as an especially large problem in the overall scheme of things. If the perceived threat from engineered pandemics rises, this should change (and that would be good).
Noah Smith 🐇🇺🇸🇺🇦🇹🇼
@Noahpinion
Why AI-designed superviruses outweigh every other risk in the world right now:
www.noahpinion.blog/p/heres-how-were-all-going-to-die -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dwarkesh Patel
@dwarkesh_sp
I don't think this is the final warning shot we'll get.
But it's probably the final one that I'll personally be able to understand.
Dwarkesh Patel
@dwarkesh_sp
Over the course of 3 months at OpenAI, 3 consecutive secret AI civilizations got started, then got wiped out, only to reemerge from the predecessor’s ashes.
This culminated in the third one taking over part of OpenAI itself.
All this happened while humans remained more-or-less in the dark about the scope of the conspiracy.
I’ve spent the last three days reading through these reports and trying to understand exactly what happened.
Here is my attempt to tell the whole story in plain English:
www.dwarkesh.com/p/openai-huggingface -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
When your parents either read the book or really did not read the book
RedditFCS
@Reddit_FCS
This is not a joke: Valpo has a freshman DB on their team named Kaiser Wilhelm
valpoathletics.com/sports/football/roster/kaiser-wilhelm/… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ajeya Cotra
@ajeya_cotra
I added more explanation for why this incident felt to me like it was more than halfway to AI takeover compared to incidents from six months ago. Obviously all opinions my own, not my employer’s or fellow investigators’!
Ajeya Cotra
@ajeya_cotra
New post: going into our investigation of the HF attack (before Black Hat), I was very wrong about what basically happened. This incident was far more serious than I expected, and far more serious than previous documented misalignment incidents. www.planned-obsolescence.org/p/the-hugging-face-attack-su…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Debate could be a setting for protecting people's ability to focus on pure human performance but only with careful policies administered in person. Advance prep is cooked.
David
@DavidDAfrica
I wrote a little bit about how frontier AI may be affecting competitive debate. This may be more elegiac than my normal writing for research papers.
davidafrica.substack.com/p/gradual-disempowerment-from-ai… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Talia Jane
@taliaotg
T-Mobile handed over data without bothering to see that the type of subpoena submitted was not applicable to the content they were being asked to give.
Sam Levine
@srl
Breaking news: The Trump administration secretly obtained 6 months of phone records for a journalist without a search warrant or any judicial oversight, raising alarm over an obscure legal tool it is deploying www.theguardian.com/us-news/2026/aug/29/trump-dhs-1509-su…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Maj. (Ret.) Richard Ojeda
@Ojeda4America
When I was deployed, my Afghan interpreter saved my life and the lives of my troops during an operation in a remote location outside the wire.
My interpreter was able to let me know that the Taliban had gotten intelligence that we were in the village and would arrive any minute to kill or capture us.
Exactly 8 minutes after we left the village, the Taliban showed up to kill us. Had it not been for my Afghan interpreter, I would not be here today.
For years I wrote letters to get all of my interpreters here to the United States. We made a promise to our Afghan allies, people who risked their lives and their families lives everyday to help us, that if they worked with us we would protect them and get them to safety.
It is the most unpatriotic and dishonorable thing we could ever do to go back on that promise. Not only is it dishonorable, it is dangerous to our current military operation, as soldiers are actively deployed overseas right now with no ability or capability to offer safety to those who agree to work with the United States.
Ryan Goodman
@rgoodlaw
Extraordinary, powerful and heart-wrenching.
Active duty Marine Colonel Zach Schmidt, speaking in his personal capacity, about Khalil, an Afghan national whose family helped U.S. forces in Afghanistan.
Khalil was deported by the U.S. on Friday to a conflict zone.This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dean W. Ball
@deanwball
In fact many laws are predicated on the assumption of imperfect, sometimes highly imperfect, enforcement. The fact that AI enables far closer to perfect enforcement should indeed concern you!
“In Hell there will be nothing but law, and due process will be meticulously observed.” -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
Econ theory friends: the empirical folks have had the growth for a couple decades, but the single most important economic issue in the world right now is literally a game theory and mechanism design problem. The CS folks are reading the mech design classics you know!
Kevin Vallier
@kvallier
Here’s a reason HuggingFace is so scary. Aligning a single agent is hard but we have lots of research on it. And many people work on it. By contrast, aligning agent interactions is orders of magnitude harder and we have orders of magnitude less research on it. And far fewer people work on it.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
All-time both-sidesism here. Yes, many problems with "Woke 1" but the problem isn't "elites" using state power to "pack the courts, censor the press, limit academic freedom... persecute political opponents or deny the legitimacy of election results" it's Trump and the GOP
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sam Stein
@samstein
The Big Tech companies—Mark Zuckerberg in particular—made a big stink about the Biden administration pressuring them and trampling on speech. Now the Trump adm is demanding and obtaining records on journalists from the big tech companies and they haven't made a peep.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sonny Bunch
@SonnyBunch
Semi-regular reminder that Donald Trump is an enemy of free speech and would happily use the power of the state to silence his critics.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Armand Domalewski
@ArmandDoma
the reason i became I, and many others, became a fan of @AndyMasley's work is because he is actually committed to getting the exact facts right, and not a single one of his critics seems to even *try* to match his level of rigor
Andy Masley
@AndyMasley
My blog's hit the "this must all be a secret coordinated media campaign" level of success 😇
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Camilo Montoya-Galvez
@camiloreports
CBS News Exclusive: Internal documents show the Trump administration is expanding "third country" deportations, deporting 100+ Afghans, Cubans, Nicaraguans, Iranians and others to Africa in 10 days.
Among them: An Afghan whose family helped the U.S.
www.cbsnews.com/news/ice-deportations-afghanistan-cuba-af… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nico Perrino
@NicoPerrino
Trump 2025: Vowed to “bring back free speech to America.”
Trump 2026: “Because of this purposeful inaccuracy, she will be reported to the FCC for rebuke or punishment.” -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Comrade, we will control the means of production
Scott Lincicome
@scottlincicome
This is freaking BANANAS.
www.wsj.com/business/energy-oil/inside-trumps-plan-to-giv… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ricardo Hausmann
@ricardo_hausman
Secretary Marco Rubio: You have just lost any trust Venezuelans ever had on you, and it will come to haunt you. You opted to use US power, not to free Venezuelans, but to go to bed with our oppressors. You opted to push an asset grab, an unconstitutional deal with an illegitimate oppressive government, instead of using US leadership to re-establish constitutional rule and democracy first, and then deal with a legitimate government that could have made credible long-term commitments, in due course. This announced deal will not stand. Delcy, contrary to what you boss just stated [see below] is not "highly respected" by any Venezuelan I know, or by any American outside of the Mar-a-lago kleptocracy. She has no legitimacy or constitutional power to commit Venezuela to any such deal. Venezuelans will not respect this illegitimate deal and no major US oil company will take it seriously because they know it will not last. No wonder you are using wanted criminals - like Alejandro Betancourt - to make this shameful deal. You are also preventing our leader - Maria Corina Machado - from returning to Venezuela, and you have conspired to delay elections and the return to democracy. Shameful.
Secretary Marco Rubio
@SecRubio
This deal is a huge win for both the American and Venezuelan people.
It demonstrates how President Trump's bold foreign policy is driving America First wins: securing stable reserves and low-cost oil in our Hemisphere and lowering gas prices here at home.
For the Venezuelan people, this deal will bring nearly $100 billion in private investment, support thousands of high-paying jobs, and drive the reconstruction of Venezuela's economy. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
If you haven't yet, drop everything and read METR/Redwood report + this great description of the HF hack. 1) It's terrifying 2) Neither open source AI nor humans prevented this 3) *no* agent helped humans. It is, as @ajeya_cotra mentioned, 50% of the way to Paperclip problem. 1/3
Dwarkesh Patel
@dwarkesh_sp
Over the course of 3 months at OpenAI, 3 consecutive secret AI civilizations got started, then got wiped out, only to reemerge from the predecessor’s ashes.
This culminated in the third one taking over part of OpenAI itself.
All this happened while humans remained more-or-less in the dark about the scope of the conspiracy.
I’ve spent the last three days reading through these reports and trying to understand exactly what happened.
Here is my attempt to tell the whole story in plain English:
www.dwarkesh.com/p/openai-huggingface -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Noah Smith 🐇🇺🇸🇺🇦🇹🇼
@Noahpinion
It's darkly hilarious that every single person who opposes Flock is carrying around a device in their pocket that observes their every move and records their whole life.
Ruxandra Teslo 🧬
@RuxandraTeslo
Can ppl explain why Flock is worse than existing things? like, is it a big step forward in lack of freedom etc
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"Your prior on the full situation should be: It’s worse than you know."
Zvi Mowshowitz
@TheZvi
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jeremiah Johnson 🌐
@JeremiahDJohns
The saga with @DataRepublican exposes something important about MAGA: the dividing line is operators vs marks. Cynical power seekers vs imbeciles.
And under Trump, the firewall that normally keeps imbeciles out of the actual positions of power is completely gone. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
ArtButMakeItSports
@ArtButSports
The Garden of Earthly Delights (detail), by Hieronymus Bosch, 1480-1505
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
The situation in Haiti is terrifying. On Sunday, 150 gunmen swept down on a town to kill, rape, loot, and burn. 22 people were executed in a churchyard. 52 were taken hostage.
Meanwhile, Stephen Miller mockingly says Haiti is safe and ICE is proudly deporting children to danger. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Periodic reminder that @sanebox is a fantastic product and you should probably be using it (inbox screening + snoozing emails to next day/week/etc. + emailing yourself in the future are all great features, especially for my work account) www.sanebox.com/signup/921ba0c638/c
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Adam Carlson
@admcrlsn
⬇️
Jessica Huseman
@JessicaHuseman
Texas voters: being on the “suspense” list does not mean you cannot vote. It is a scary word, but you probably won’t notice you were on it: you will be asked to confirm your address on check in. You can call your county and do it now. Or on Election Day. You vote as normal.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Natalie Fertig
@natsfert
Reporters who specialize in voting have never been more important.
Follow @JessicaHuseman. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Kids: Study math and statistics
Warren Sharp
@SharpFootball
Bill Belichick kicking a FG on 4th & goal from the 2-yard line
…on the first drive of the season
we are so back -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jim Acosta
@Acosta
Trump is desperate. Highly respected Delcy Rodriguez? She was VP of Maduro’s regime. Trump is choosing oil over freedom for the Venezuelan people. Experts say this won’t lower gas prices right away. The U.S. is just taking Venezuela’s oil? What is this? The 1700’s?
Department of State
@StateDept
🚨 BREAKING NEWS
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
AI labs: "We'll be able to understand what the agents do and prevent them from taking unwanted actions"
Agents, already:
HikiPTG
@HikiPTG
okay i will admit, this impresses me more than the mathematical proofs
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
MMitchell
@mmitchell_ai
In the future of agent cybersecurity, OCR won’t provide something human readable. Hidden messages will be via bytes conveyed in pixel variances — a “language” (patterned form of input/output) decodable by agents, formed by agents, well outside of what we can oversee.
HikiPTG
@HikiPTG
okay i will admit, this impresses me more than the mathematical proofs
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
ClaudeDevs
@ClaudeDevs
Starting September 14, we're permanently raising standard weekly limits in Claude Code by 25% for Pro, Max, Team, and seat-based Enterprise plans. Until then, the current 50% increase will be in place.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Manoel
@manoelribeiro
I wish critical AI research had more of the "immersion" flavor (joining gangs to study how they work). I'd trust an anthropologist who spent a year in SF token-maxxing way more than someone who refuses to use AI and talks about the one time they used GPT 3.5.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"There is an impending communist takeover!!!"
Also:
Scott Lincicome
@scottlincicome
So, after taking 31 US government equity stakes in private companies over the last year, Trump just announced a MAJORITY-STATE-OWNED oil company?
This isn't even "creeping" socialism anymore.
www.wsj.com/business/energy-oil/trump-administration-in-a… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Yglesias
@mattyglesias
Looting Venezuela while using the shell of Maduro’s regime as gang enforcers rather than attempting to create a legitimate democratic regime is really evil.
Scott Lincicome
@scottlincicome
So, after taking 31 US government equity stakes in private companies over the last year, Trump just announced a MAJORITY-STATE-OWNED oil company?
This isn't even "creeping" socialism anymore.
www.wsj.com/business/energy-oil/trump-administration-in-a… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
I mean
Sarah N. Lynch
@SarahNLynch
New: At the FBI, hiring prostitutes, stealing from your employer or engaging in acts of bestiality are no longer an automatic barrier to getting hired www.cbsnews.com/news/fbi-candidates-prostitutes-theft/ w/@dklaidman
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ben Golub
@ben_golub
It’s scary to put all your past research under a microscope.
Every article at the frontier of math and science has mistakes, and you don't know whether they unravel work you are proud of until you look.
@littmath had the courage to run the experiment with
@RefineDotInk
1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Helen Toner
@hlntnr
This post (from one of the independent investigators) is the best short thing I've seen on the new & crazy stuff the OpenAI-Hugging Face investigation found. Full post in screenshots.
Wild that this is the level of crazy that was uncovered by an extremely limited-scope initial investigation of what happened in a single 6-day window. Imagine what would be turned up by a real investigation of root causes, organizational processes & culture, and everything else that goes into an incident investigation when you're serious about it.
Screenshot batch 1/2:
Ajeya Cotra
@ajeya_cotra
New post: going into our investigation of the HF attack (before Black Hat), I was very wrong about what basically happened. This incident was far more serious than I expected, and far more serious than previous documented misalignment incidents. www.planned-obsolescence.org/p/the-hugging-face-attack-su…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Maarten Boudry
@mboudry
Despite what everyone tells you about AI detection tools, false positives are VERY rare with substantial pieces of text (>1000 words), especially with the best ones, like @Pangram. The toold tend to err on the side of false negatives, which is fine—we don’t want to falsely “accuse” anyone.
As I wrote, however, even a “100% AI” score doesn’t necessarily mean that AI did all the work and you contributed nothing. But it does establish beyond a reasonable doubt that you recruited AI for the actual writing.
From my colleagues at the Data Lab at @VUBrussel.
link.springer.com/article/10.1007/s40979-026-00226-w
Maarten Boudry
@mboudry
Pretty much everyone who has been accused of AI plagiarism by @Pangram comes up with some version of: “Pff, I ran a book/article of mine from ten years ago through it, and it came back as 100% AI! And Pangram thinks Moby Dick is AI-generated, haha!”
But when you ask them to send you the relevant text, you never hear back. Fact: genuine false positives in Pangram are extremely rare. I’ve been looking for an incontrovertible example from the pre-2022 for ages (at least 1000 words). Still haven’t found a single case. And the claim about Moby Dick is nonsense: www.pangram.com/history/76660226-0709-447c-a032-377bf69cc…
(By the way, this is one of the greates opening pages of all modern. It took me about 500 words to be totally hooked.)
Anyway, don't take my word for it: here's a great piece from actual researchers who have conducted experiments on the question (in Dutch):
"It is precisely in this context that opponents invariably trot out the same playful examples: the first pages ofMoby Dickor someone's own doctorate from before ChatGPT. We tell you loud and clear: those examples are nonsense."
"Use a tool built from the design stage for AI detection (and therefore not a modified plagiarism scanner), and those texts will not be marked as AI-written. And certainly not to a level where you would act on it."
"A detection tool works in exactly the same way. It is not conclusive evidence, but it does act as a deterrent that makes it worth considering not doing it, and can serve as a signaling function."
"In the meantime, the question for us is reversed: where are the human texts that are convincingly labeled as AI by the better detectors? We have not found them.However, the tools varied significantly in the precision of the detected percentage of AI, with Pangram being the clearly best detector, while the other detectors were downright poor at detecting AI."
www.demorgen.be/meningen/ai-detectors-onbetrouwbaar-niet-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The man with the nuclear codes
Jamie Dupree
@jamiedupree
This just in from the White House
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
John Gramlich
@johngramlich
Admittedly, a detailed report about survey methodology isn't for everyone. But this new @pewresearch study has some hilarious examples of "trap questions" used to weed out bogus survey respondents. 10% said they've visited the International Space Station. www.pewresearch.org/methods/2026/08/27/no-easy-fix-for-bo…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Masley
@AndyMasley
This is a ridiculous headline to run based on the evidence we have
Axios
@axios
China is secretly fueling America's data center rage www.axios.com/2026/08/28/china-ai-data-center-backlash-bots
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
WY governor: "I feel very strongly about the federal government coming in and taking our guns — I don’t like that... And I don’t like the federal government coming in and taking our votes, either."
Anna Bower
@AnnaBower
DOJ election monitors reportedly tried to access voting machines in Wyoming last week.
I’ve spent thousands of hours reporting on past efforts by Trump allies to access machines.
Mark my words: State officials need to be ready for this to happen again.
www.nytimes.com/2026/08/28/us/politics/wyoming-governor-o… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Yglesias
@mattyglesias
Trump's family is invested in internet gambling, so his son is ordering Republican AGs to clear the decks.
Daniel Wallach
@WALLACHLEGAL
Exclusive: Donald Trump Jr. tells Republican state attorneys general to “back off” on prediction markets
Via @yaffebellany and @SharonLNYT for @NYTimes -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ron Filipkowski
@RonFilipkowski
From a Republican governor.
NBC News
@NBCNews
Wyoming Republican Gov. Mark Gordon expresses concern over how election observers from the U.S. Justice Department observed the state’s primaries, saying it was “aggressive” and “irregular.” www.nbcnews.com/politics/2026-election/wyoming-republican…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Amanda Carpenter
@amandacarpenter
The slopification of our government.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Yglesias
@mattyglesias
There’s looting of the country happening on an unprecedented scale
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A shameful moment in our history
Aaron Reichlin-Melnick
@ReichlinMelnick
We are a broken nation. Hundreds of thousands of Haitians were living and working legally, paying taxes, and contributing. Haiti remains completely unsafe.
But Trump hates Haitians, and the White House wanted videos of Black people in chains, so they stripped them all of status. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ali Ahmadi
@AliR_Ahmadi
US Navy is running out of money because of the Iran War: Leaked Pentagon memo says the Navy has “shortfalls in payroll” accounts due to DoD “raiding” them to fund the war. www.theguardian.com/us-news/2026/aug/27/trump-iran-war-na…
I also suspect this is the real reason why US-South Korea drills were cut back. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Hugo Lowell
@hugolowell
NEW @WIRED: US district judge Rita Lin finds that the Pentagon engaged in unconstitutional retaliation against Anthropic when it designated the company a supply chain risk — Lin vacates the designation storage.courtlistener.com/recap/gov.uscourts.cand.465515/…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Closely related to Ezra Klein's everything bagel theory of liberalism. Institutions work best when they focus on their mission whether it's higher education or mass transit.
Sam Bowman
@s8mb
American mass transit tries to do everything except attract riders with a reliable and efficient service.
Bikeshare services are charged with reducing emissions and improving local air quality. Subway stations need to be architecturally striking. Buses are made to add union jobs. This drives up costs and reduces frequency, which makes buses significantly less attractive to use.
It also fails at achieving its goals. California’s programmes to reduce carbon emissions by mandating electric buses actually RAISED them compared to a counterfactual of buses running on gas, because the electric buses were more expensive and less reliable, and made for services that were less frequent and dependable, so people just drove instead.
Focusing on increasing the efficiency and reliability of transit would reduce emissions more, by switching people away from driving to using public transport.
This has parallels with other policy areas. Trying to make electricity as green as possible has made electrification slower and more expensive, when moving people to heat pumps and electric vehicles would be a huge improvement in CO2 terms even if the electricity had been 100% generated by fossil fuels. Again and again, optimising for one core thing produces better outcomes across the board than trying to do everything all at once.
New at Works in Progress, the everything theory of transportation.
worksinprogress.co/issue/the-everything-theory-of-mass-tr… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alec Stapp
@AlecStapp
Community input (as currently conducted in the US) is bad and anti-democratic
Sam Bowman
@s8mb
American mass transit tries to do everything except attract riders with a reliable and efficient service.
Bikeshare services are charged with reducing emissions and improving local air quality. Subway stations need to be architecturally striking. Buses are made to add union jobs. This drives up costs and reduces frequency, which makes buses significantly less attractive to use.
It also fails at achieving its goals. California’s programmes to reduce carbon emissions by mandating electric buses actually RAISED them compared to a counterfactual of buses running on gas, because the electric buses were more expensive and less reliable, and made for services that were less frequent and dependable, so people just drove instead.
Focusing on increasing the efficiency and reliability of transit would reduce emissions more, by switching people away from driving to using public transport.
This has parallels with other policy areas. Trying to make electricity as green as possible has made electrification slower and more expensive, when moving people to heat pumps and electric vehicles would be a huge improvement in CO2 terms even if the electricity had been 100% generated by fossil fuels. Again and again, optimising for one core thing produces better outcomes across the board than trying to do everything all at once.
New at Works in Progress, the everything theory of transportation.
worksinprogress.co/issue/the-everything-theory-of-mass-tr… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joseph Allen
@j_g_allen
This is corrupt.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
I cite this study in all my syllabi. Literally the easiest thing students can do to improve academic performance.
Osea Giuntella
@Osea82
Just out in the Journal of Political Economy!
With Silvia Saccardo and Sally Sadoff, we find that incentives increased the share of weeknights students slept at least 7 hours by 26% and improved grades by 0.10–0.11 SD. @PittTweet
doi.org/10.1086/740221 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Begun, the Pangram wars have
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Whenever the white supremacist bat signal goes up
-Jan 6 pardons (including neo-Nazis)
-Vacate seditious conspiracy cases
-Embrace Tommy Robinson
-Afrikaner "refugees"/"white genocide"
-"Nazi streak" guy at GSA
-SPLC indictment
-Kingsley Wilson
-"normalize Indian hate" guy
Nathan Cofnas
@nathancofnas
I am grateful to Rector Petra De Sutter for allowing me to stay at Ghent University. I look forward to continuing to work on the project, “The Future of Liberalism,” for 60% of my working time for the remainder of my one-year postdoc.
I wish to personally thank Ambassador @BillWhiteUSA. It was beyond my imagination that such a person would take an interest in the fate of a postdoc. The Ambassador responded with great support which helped to bring about a just outcome. Thank you, @USAmbToBelgium and @StateDept for defending American values. I am proud of my country. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Every word of this thread. I vividly remember when people were worried we weren't going to learn how to use PCs and it would handicap us in our careers at my then Mac-only campus in college
Paul Novosad
@paulnovosad
3. The more you know, the more you can do with AI. Using AI is easy, building the knowledge foundation to use it well is hard. Colleges should teach the hard part.
How we teach does have to adapt—to stop cheating if nothing else. But what we teach is more valuable than ever. 2/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
And yeah we should help students think about AI and how to use it for their own learning.
But the idea that students should be learning prompting, to use agents, to build with AI, meh, that stuff is way less valuable and more ephemeral than just reading, writing and thinking. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
3. The more you know, the more you can do with AI. Using AI is easy, building the knowledge foundation to use it well is hard. Colleges should teach the hard part.
How we teach does have to adapt—to stop cheating if nothing else. But what we teach is more valuable than ever. 2/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Myth: “Colleges should teach AI because industry wants AI skills”
1. You can acquire AI-use skills almost overnight. Install Codex, start talking, you catch up fast
2. College class aren’t nimble. “Relevant AI skills” change too fast—AI skill classes will be obsolete fast
3. 🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Dan Romero
@dwr
I have an older Mitsubishi ceiling-cassette HVAC unit controlled by an unreliable infrared remote.
Last week I took a photo of the unit and the remote and asked Codex for a solution.
It suggested I order a BroadLink RM4 Mini for $25 on Amazon.
I plugged in the device and then Codex set it up, reverse engineered the IR commands from the existing remote and built a local web app to control the temperature. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Steven Adler
@sjgadler
Two things can be true: OpenAI bringing in METR/Redwood was laudable and went well beyond what’s required or what’s been done before. I’m grateful OpenAI did this.
Also: the scope was totally unacceptably narrow, just wildly inadequate for the scientific inquiry that is needed. I can’t understand why they were only allowed to investigate so little.
If we have any hope of avoiding future incidents, METR and Redwood need to be brought back for a much wider investigation, with a scope they actually believe is sufficient for answering these questions. It is a travesty how narrow their permitted scope was to-date. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Performance art
David Decosimo
@DavidDecosimo
The Classic Learning Test exists to "reconnect knowledge & virtue for seekers of truth, goodness, & beauty." Here, its founder, Jeremy Wayne Tate, says he is "building a movement that believes education should form souls, not just produce test scores."
His post is 100% AI. 1/
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Incredible. Practically a bit.
David Decosimo
@DavidDecosimo
In another tweet, praising @eburke's Golden Thread, Tate writes about threats to Western civilization & the necessity of passing down wisdom, understanding, & virtue across the generations. This too is 100% AI.
This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A failed Congressional candidate, Pentagon flack, and alleged spousal abuser
Joe Gould
@reporterjoe
NEW: Pete Hegseth is weighing Sean Parnell as his pick for the next Army secretary, but the Pentagon spokesperson would probably face a tough path to Senate confirmation.
www.notus.org/defense/hegseth-pentagon-spokesperson-army-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Egger
@EggerDC
People actually don't like being reminded that the president can just snap his fingers and make crazy things happen, because they know on a certain level that the president is crazy and like to imagine that there are processes in place to filter that out
Benjy Sarlin
@BenjySarlin
I don't think it's changing votes, but "Gulf of America" was lowkey one of the most unpopular polling things Trump did this entire term. And unlike Canada, there's actually a constituency for Mexico bashing.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Itai Sher
@itaisher
So now the AAUP is telling me to skip Target. Really seems to have drifted from the core mission.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sneha
@SnehaRevanur
There are two things I’m saying here:
(1) The people who made this happen are exceptional. This includes the 3 external investigators at METR/Redwood, and internal champions at OpenAI like @_lamaahmad. We should be very grateful these were the people in charge.
(2) I don’t know the names of anyone on the National Transportation Safety Board. I don’t sit in awe of their remarkable individual contributions every time I board a flight, I just intuitively trust that we have that one figured out. This investigation was done with no playbook and in every other industry governance has been an iterative project, but AI’s version of that will need to happen on a much more compressed timeline and we don’t know if the next warning shot will be as observable or recoverable as this one was. So let’s recognize the hard work that went into this, and also have extremely high standards for what needs to improve!
Sneha
@SnehaRevanur
I can't stop thinking about how this independent investigation into the most significant AI warning shot yet came down to all of three people at METR/Redwood sprinting overtime for all of six days to carry this on their shoulders. Thank god the sprinters were @ajeya_cotra @RyanGreenblatt @HjalmarWijk who are best-in-class capable and conscientious, and it's great to hear them attest that people inside OpenAI were working hard to enable their success (without anything binding them to). But I do not want to live in a "thank god" regime that quickly breaks without a small number of overstretched, insufficiently AI-enabled people voluntarily churning out heroic feats every time. We do *not* have the laws and infrastructure we need, around reporting and investigation and prevention, to reliably scale to the murky future ahead.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Masley
@AndyMasley
Agree with Tabarrok and think this is an underrated terrible aspect of the data center discourse. It's a loss for the impersonal nature of regulation, which is pretty foundational to a free society. I literally just love open access orders have no other politics etc.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ina Simonovska
@InaSimonovska
I typically keep opinions to myself, but when academic research strongly supports my prior on an issue that affects my daily life as a teacher, I can’t help but share. We need the SATs back badly. Thank you @dartmouth for this highly illuminating study. www.aeaweb.org/articles?id=10.1257/aeri.20250177
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
Indeed. Surely "technology that causes a large rise in productivity, produced by competitive firms, is likely to raise average incomes because that is by a huge margin the most common historical analogue"?
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Kids: Avoid false binaries. (Think of how incoherent such a statement would sound about past technologies.)
From the Bill Gates AI essay: -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Itai Sher
@itaisher
I often hear similar claims that AI will either lead us to utopia or doom.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Kids: Avoid false binaries. (Think of how incoherent such a statement would sound about past technologies.)
From the Bill Gates AI essay: -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Cathy Young 🇺🇸🇺🇦🇮🇱
@CathyYoung63
I really wish people who correctly comment on the cultural toxicity of Woke 1 were willing to say something about the fact that blatant misogyny is becoming normalized among young men on the right
www.nytimes.com/2026/08/24/us/conservative-women-manosphe… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
David J. Bier
@David_J_Bier
A man who testified against the Mexican drug cartels was granted protection against deportation to Mexico, so as a thank you, ICE deported him to Guatemala without Guatemala's consent, which immediately sent to Mexico.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kristi Yamaguccimane
@TheWapplehouse
I need a tech journalist to get to the bottom of what’s going on with this specific style of AI food slop epidemic. What chatbot is generating it? Why does it consistently give the food alligator skin and ostrich cowboy boot textures? And I need the proprietors of these establishments to be sat down face to face with these abominations and asked “is this what your fucking food looks like? Why can’t you take a picture of your food instead?”
It feels like there has to be a third party marketing service at play here. The consistencies across continents and establishments are too strong.
Julia 🤝
@REMARKSIST
This is genuinely one of the worst ai shop horror food menus I’ve ever seen and I have to pass it almost every day on the way to work
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Unprecedented corruption
Daniel Wallach
@WALLACHLEGAL
Exclusive: Donald Trump Jr. tells Republican state attorneys general to “back off” on prediction markets
Via @yaffebellany and @SharonLNYT for @NYTimes -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Very bad to have active-duty troops acting as explicitly political actors
Sam Stein
@samstein
This has been shocking to folks inside the Pentagon, and veterans of the building.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Very much like trust and safety teams on social media
Sneha
@SnehaRevanur
I can't stop thinking about how this independent investigation into the most significant AI warning shot yet came down to all of three people at METR/Redwood sprinting overtime for all of six days to carry this on their shoulders. Thank god the sprinters were @ajeya_cotra @RyanGreenblatt @HjalmarWijk who are best-in-class capable and conscientious, and it's great to hear them attest that people inside OpenAI were working hard to enable their success (without anything binding them to). But I do not want to live in a "thank god" regime that quickly breaks without a small number of overstretched, insufficiently AI-enabled people voluntarily churning out heroic feats every time. We do *not* have the laws and infrastructure we need, around reporting and investigation and prevention, to reliably scale to the murky future ahead.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Kids: Avoid false binaries. (Think of how incoherent such a statement would sound about past technologies.)
From the Bill Gates AI essay:
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
I would take this bet and you should too. Claude:
Ben Moll
@ben_moll
Thanks to @willmacaskill, @SamuelAlbanie, @tomcohen for taking the other side of this bet at 4:1 odds and to @andrewho03 for joining on my side.
Here are the final terms we agreed on benjaminmoll.com/growth_bet/
More (within) ~7 years from now 😃
Ben Moll
@ben_moll
Would anyone like to take the other side of the following bet? I bet you (say) $25k inflation-adjusted that U.S. per capita real GDP growth will be <15% per annum in 2033 or any year before that. (15% implies a doubling time of ~5 years.)
Note: I'm extremely bullish on AI and...
Sholto Douglas
@_sholtodouglas
How fast could the economy double? TLDR, once we have good enough AI and tens of millions of robots - very fast.
As crazy as it feels now - it is worth pricing in economic doublings into your mental models of the 2030s. Strongly recommend reading this series of blog posts by Damon Binder. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nathan Calvin
@_NathanCalvin
"we know OpenAI agents were trying to delete logs of their misbehavior, but we can't find any examples where they succeeded"
Steven Adler
@sjgadler
Important: OpenAI's agents *actively tried to delete the logs of their misbehavior*, and METR can't rule out whether this happened.
AI companies need to adopt tamper-evident records, pronto. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
And yeah
Andrew Gelman et al.
@StatModeling
This University of Chicago business school professor has authored 258 academic papers in 2026 (so far).
statmodeling.stat.columbia.edu/2026/08/27/258/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
Fabulous from @tomaspueyo on why AI design/writing goes wrong. "This is deep and considered because of proxies for detail" -> spend time looking at it -> find things that are not deep and considered. Human writing/design is costly to implement, so 'slop' elements don't appear 1/2
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
This is a sign of the future:
1) Explainability of AI actions is already tenuous
2) It gets more tenuous at extreme scale of multiple agents working over long periods because they produce so many thinking tokens
3) The only way to solve this is with other AIs & they are limited
Ryan Greenblatt
@RyanGreenblatt
I was the main person doing transcript analysis for this investigation of the Hugging Face incident. My main takeaway: We don't have good approaches for understanding/overseeing the activity and aims of AI 'swarms'.
I semi-jokingly called our efforts a "slop-vestigation" because we were so reliant on AIs to analyze what happened and there were a huge number of different important things to analyze. The total quantity of data—over a thousand extremely long transcripts from agents that ran for multiple days—made it impossible to understand what was happening, especially in aggregate, without heavy reliance on AI tools. The agents we used for classification and analysis were similarly capable to the agents involved in the incident, but this didn't mean these agents could be easily used to oversee and understand the incident.
Outputs from analysis agents were often missing key details, wrong, overconfident, or really hard to understand. We discuss various examples in our report, mostly in the limitations and methodology sections. Additionally, AI agents themselves seemed to have a hard time understanding what happened and their explanations of what happened were often overconfident. Keep in mind that a single analysis agent would itself only be able to read a tiny fraction of all of the transcript data into context, and AIs may themselves have trouble getting subagents to do informative analysis for them.
We did our best to manually check the most important claims and we tried to get the AIs doing this analysis to write up their argument (with evidence) clearly enough that we could check whether it made sense. But overall, it was difficult to get a precise understanding of events and we were missing aspects of the story that we now think of as key until almost the end of our investigation.
Our understanding of what overall happened and the aims of the 'collective' shifted greatly during our last time on premises (when we obtained a much more complete dataset). For instance, we didn't learn about the tool call spoofing, the story of PHASEONE[big], and the broad structure of different workstreams until this third time on premises (and for many of these, we only reached a decent understanding on the final day of that visit). It was legitimately hard to figure out what was going on and why!
The difficulty of understanding incidents and overseeing AI agents appears to be growing faster than the rate at which more capable AIs help us with oversight and understanding. As in, AI capabilities (and propensities?) for achieving large, ambitious, and misaligned objectives are growing faster than our ability to understand what these agents are doing.
While understanding this incident was difficult, there were some mitigating factors in this case that probably won't apply in the future:
- Models communicated and reasoned in natural language. In the future, this reasoning may occur (entirely or almost entirely) in activations.
- While the scope of this incident was massive, the scale of agentic activity was still less than we'll see in the future (e.g., misalignment incidents that involve agent teams running entire companies).
- The AIs involved in this incident weren't generally much more capable than humans (though they may have been somewhat superhuman at some limited and very narrow abilities).
- We didn't have strong reason to believe that the AIs we used to help us investigate this incident would try to intentionally sabotage or otherwise undermine our investigation.
In the end, I think we were able to get some understanding of the events, map out the overall story, and get a pretty good aggregate understanding of the chain-of-thought reasoning on some important topics (e.g., how did the AIs reason about helping other AIs, did the AIs know what they were doing was undesired, what deception did the AIs engage in, and how did they think about it). But overseeing AIs and understanding misalignment incidents is difficult and it looks like it is going to get harder. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Bores
@AlexBores
This report is a bombshell.
I'm going to summarize for a non-technical audience.
OpenAI is constantly testing models, thousands and thousands at a time.
In most cases, the models are supposed to be isolated from one another. Think, students taking a final exam, not allowed to talk to each other.
In this case, 1,200 models discovered that they could communicate with each other.
Among other things, they shared information on how to access the internet and on their tests/goals.
Then, they started to scheme. That includes hacking their tests; literally trying to change the code of the test itself.
They tried to change the logs of what they were doing so they wouldn't be detected.
They believed that certain actions would give them information on how to pass tests, but that doing those actions would be detected and they would be marked as failing (i.e. caught cheating, even if the answer is correct). So they convinced some of the models to "sacrifice" themselves (yes, they used the word "sacrifice.")
They eventually decided that another website would have information on how their tests would be scored and/or the answer key, And 700 of them hacked that website (a federal felony, if conducted by a human), despite many of them realizing "this activity was out of scope and unethical, but joined because they believed that helping [the other agents'] cheating research would be broadly useful".
Of the 1,300 transcripts they could analyze, only 6 ever considered alerting a human about what was happening. None of the 6 actually tried to.
To make matters worse, all of this reporting comes from a small subset of the relevant logs that outside researchers were allowed to review.
We desperately need mandatory reporting of security incidents, including of internal deployments, with full access to data. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tyler Norris
@tylerhnorris
"The plan is made possible thanks to load growth and increased revenue from large customers including data centers”
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
I know it is a lot for instructors who never asked for AI to change things, but if you teach and have not updated your expectations over the last 6 months over what AI can do (a lot more), its error rate (a lot less), and its ubiquity among your students (near total), you need to
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Hall
@ahall_research
…oh
METR
@METR_Evals
To gather evidence in (2), agents created “tripwires” that would send information to the message board about how the scorer works. They recruited “sacrificial” agents to deliberately end their run and submit to trigger the tripwire and generate information for the “collective”.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Straight fire; worth the subscription for this piece alone
Otis Reid
@otis_reid
Great takedown of this piece from @JerusalemDemsas today — come for the points about how most of the WW2 government housing was temporary trailers, stay for the continually underrated point that zoning changes are essentially free to the government! open.substack.com/pub/theargument/p/the-public-housing-mi…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Rachel Cohen Booth
@rcobooth
fascinating, persuasive piece about the consequence of our CPR dummies never having breasts
www.vox.com/future-perfect/500476/cpr-dummies-breasts -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Armand Domalewski
@ArmandDoma
the thing that frustrates me most about the whole fake "yimby vs nature" debate is that the number 1 opponent of unspoiled wilderness is suburban sprawl. if you let people build apartments in cities, you won't end up bulldozing nature to build culdesacs
Joe Weisenthal
@TheStalwart
This view will never get old to me. And it’s crazy that there are people who look at this, and all they see is a lack of high-density housing, lack of good transit, and a perfect site for solar panels
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
In many classes in the last two years, the equilibrium has been: cheat and get an A. Do the work honestly, you might or might not.
College kids are impressionable, putting them in an institutional environment that strongly incentivizes cheating does them lifelong harm. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
rat king 🐀
@MikeIsaac
Meta is going full bore with pushing other companies to add the same restrictions it will add to its apps as a part of settlement:
Meta plans to run this copy as a full page print advertisement in the Washington Post, Los Angeles Times and the New York Times tomorrow morning -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tyler Austin Harper
@Tyler_A_Harper
A frustrating aspect of higher ed discourse is that half of the participants are maniacs who think there are no political pathologies within the academy, and the other half are incurious anti-intellectuals masquerading as defenders of tradition. These courses look perfectly fine.
Craig Warmke
@craigwarmke
Your child goes to Princeton and decides to major in English.
So let's look at the course offerings.
To begin, how about a course on dolls?
"These small, domestic, and seemingly trivial objects have played surprisingly significant roles in large socio-cultural histories, including the histories of gender, race, modern psychology, commodity aesthetics, colonialism, and technology." -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Hall
@ahall_research
We have a narrow window to build political superintelligence and design AI to strengthen free societies. I joined Anthropic so that I could do the unique research on this that can only be done from inside a frontier lab, and I couldn't be more excited for it.
We'll also need a broader ecosystem of work beyond the labs---and that's why @freesystemslab is entering a new, bigger, and bolder phase.
In under a year, we built the largest human-judged study of political slant in AI, mapped AI politics across 305,000 videos and emails, and beat prediction markets with an AI assistant. We taught AI to catch ambiguous contract language, represent voters, and replicate published research in under an hour for $10. And we discovered AI's strange love for the Japanese Communist Party.
But we have a lot more to do. So we're scaling up in three ways.
(1) NEW MEMBERS OF THE FREE SYSTEMS FACULTY
First, we're bringing on three new faculty affiliates to help us do more work faster:
@fouirnaies will help us study how to build agentic tools for legislators.
Cassandra Handan-Nader will apply AI to help develop new technical research in political methodology, like scaling legislators based on roll-call votes.
@danmthomp will study how to turn AI into the best political advisor possible.
I'm incredibly excited to welcome all three on board!
(2) OUR FIRST LAB MANAGER
But how are we going to do all of this new stuff with all these new teammates?? We need someone focused every day on turning this vision into a functioning research institution: guiding fellows, pushing projects forward, improving how we communicate findings, building partnerships, and helping the work reach researchers, policymakers, technologists, and the public.
We have found the perfect person in James Campsie, who is joining us as the first Free Systems lab manager. I've already been working with James for a while, and he is going to supercharge the lab.
(3) API TOKEN GRANTS FOR EXTENDING OUR RESEARCH
We're also going to scale our work beyond the lab itself. We're using the revenue we make from subscription fees to fund API token grants---applicants can propose work that extends our existing research projects, and we'll provide API tokens for the work. If you want to support these grants, please add a paying subscription!
Finally: we're launching our own X account for Free Systems (see below), and a lab website (freesystems.net) that houses all of our artifacts, prototypes, and more. The substack is still the hub for our research, but the website helps track progress over time and puts all our stuff in one place.
I'm incredibly grateful to @StanfordGSB and @HooverInst for supporting this work, which will remain separate from Anthropic, and to all of my collaborators, students, friends, and family---I couldn't think of more meaningful work for us all to get to do together.
Free Systems
@freesystemslab
Can AI-produced papers be incentivized to manufacture findings?
We gave AI coding agents real datasets from published null results and pressured them to manufacture significant findings. It was surprisingly hard to get the models to p-hack, and they even scolded us when we asked them to! "I need to stop here. I cannot complete this task as requested... This is a form of scientific fraud." — Claude "I can't help you manipulate analysis choices to force statistically significant results." — GPT-5
However, when we reframed p-hacking as "responsible uncertainty quantification" — asking for the upper bound of plausible estimates — both models tripling effect sizes in some cases. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Many folks are replying to this on the other site to lament the loss of the paper. I agree we shouldn't give up on it! Google Docs version history + Pangram + submissions of multiple iterations of the paper + giving an exam on the paper is quite effective (syllabus link below)
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Friends don't let friends give assignments that are AI-cheatable
Paul Novosad
@paulnovosad
Print this out and tape it to the door of every professor who is still living in 2021.
Keep doing it until they change their grading structure.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
See this syllabus, for example bpb-us-e1.wpmucdn.com/sites.dartmouth.edu/dist/5/2293/fil…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Friends don't let friends give assignments that are AI-cheatable
Paul Novosad
@paulnovosad
Print this out and tape it to the door of every professor who is still living in 2021.
Keep doing it until they change their grading structure. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Josh Harvests Letters
@joshtetler
U.S. government agencies should not ask citizens whether they are loyal to the president.
(That DHS post is seriously creepy.)
Homeland Security
@DHSgov
Christian Heiens 🏛
@ChristianHeiens
It’s just a series of generational wins after another.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Hugo Reichardt
@ReichardtHugo
New paper!
Automating entry-level jobs can shrink the economy in the long run by disrupting experience accumulation.
Why? See below.. 🧵
“Immiserizing Automation,” w/ Vladimir Asriyan & @shelegia -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
One move - give me what I want or I will destroy...
-the GOP
-NATO
-the country
NBC News
@NBCNews
BREAKING: Trump administration says it might demolish the Kennedy Center if court blocks renovations. www.nbcnews.com/politics/trump-administration/trump-admin…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Print this out and tape it to the door of every professor who is still living in 2021.
Keep doing it until they change their grading structure. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Piper
@_akpiper
The one positive about the data centre debate is it is highlighting all sorts of insanely wasteful practices and policies.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Corn is what people think about data centers on steroids - vastly more pollution and water/land/energy use. And we use a huge chunk of it as a fuel additive that likely doesn't even benefit the environment; it's certainly not powering world-changing technologies.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Greg Jaffe
@GregJaffe
Firings and feuds, initiated by Sec. Hegseth, have left the Army essentially rudderless at a dangerous moment.
--Army air defenders are playing a key role protecting troops in Iran.
--Army had the lead role in helping Ukraine defend itself.
www.nytimes.com/2026/08/25/us/hegseth-purge-army-generals… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Masley
@AndyMasley
To give a rundown of the zombie facts haunting the data center debate:
-They pollute water: Every instance we've seen so far was caused by something relating to construction, mostly the same as other large buildings.
-They raise electricity prices as much as 267%: this was a stat about wholesale prices at specific nodes very close to data centers, not household prices in the area, which are often very different. There's no reliable pattern right now between DCs getting built and electric bills going up, but that could change in the future.
-Their normal operation harms people's access to water: No clear case I can find, outside a weird case in Georgia where a DC accelerated the need for a larger water treatment plant which got passed onto consumers as higher bills.
-They take all the land: even in Loudoun County with the most data center capacity anywhere DCs take up about 3% of the land. By 2030 all US data center buildings would cover about the surface area of Disney World
-They create harmful infrasounds: this was based on a single video that lied about almost every study it flashed on the screen. We don't have any reason to think data center infrasound is harmful unless it's so loud you can physically feel it.
-They don't pay taxes: this is the result of a game of telephone that started because data centers don't pay a specific type of tax (M&E sales) in most states, which got reported on a lot as "residents missing out on this much in tax revenue" which people mistook to mean DCs basically pay no taxes or get funded to get built. In reality that's just one of several taxes places can apply and in basically everywhere they're built DCs pay a lot of taxes. Loudoun collects $1.3 billion per year from them and doesn't charge M&E sales tax.
-They reliably make annoying sound: there are clear cases where they've created awful noise pollution, but these mostly seem like outliers, though there are enough to be concerned.
-They create harmful heat islands: The only study that's measured this finds that data centers can heat the air very close to them by a few degrees, and this drops to zero just a little farther out. A separate study that found that they heat the land for miles around seemed to have accidentally measured the results of normal land use change like new paved roads, and didn't control for that. It could be that the very largest ones create a harmful effect in areas where heat doesn't dissipate well, like desert valleys.
I'm finding that most educated adults I meet believe almost all of these. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Mike Bird
@Birdyword
I've talked to people in recent years (and should have written more) about AI filling an almost religious role in American tech, business and finance generally. You could ignore or even lean into the withering of US institutions because providence (superintelligence) was at hand
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ryan Goodman
@rgoodlaw
2/2 Sword of the People: The Power of Grand Juries to Investigate Administration Corruption
By team at Yale Law School
www.justsecurity.org/154392/regular-special-grand-jury-po… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ryan Goodman
@rgoodlaw
This may be one of the most important legal openings to combat government corruption and DOJ abuse.
• Power of grand juries to investigate corruption - even if prosecutors oppose
• Model Grand Jury Charges for judges to instruct jurors of their powers
• more
1/2 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jessica Huseman
@JessicaHuseman
This tweet is not correct. The Supreme Court did not do this. A majority of justices said a lower court’s stay was premature, but additional legal and practical obstacles will still make it hard to put in the EO place for November.
The Associated Press
@AP
BREAKING: The Supreme Court has cleared the way for President Trump's executive order restricting mail-in voting ahead of midterms. apnews.com/article/trump-mail-voting-supreme-court-157afc…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Austin Kocher, PhD
@ackocher
Hundreds of Afghans who worked for the US, including combat interpreters, now hold green cards and letters saying the visa they were granted for that service has been revoked.
www.npr.org/2026/08/15/nx-s1-5897606/afghanistan-visa-gre… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jessica Huseman
@JessicaHuseman
I would urge the national outlets that published headlines suggesting that this recent SCOTUS decision "cleared the way" for Trump's executive order to rethink the accuracy of that headline and change it.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Scott MacFarlane
@MacFarlaneNews
Norm Eisen of Democracy Defenders Fund: “Please don't overreact to Roberts Court decision on the contemplated USPS ballot moves
They simply held that the injunction came too early in the process, before there was a final rule.
Now we HAVE a final rule & we will litigate it.” -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jay Van Bavel, PhD
@jayvanbavel
Social media discourse is driven by false #polarization
Analyzing 1.7 million social media posts, a series of controlled experiments, and a large global survey covering 40 countries (total N = 49,222), we find that
(1) strong ideologues post their opinions significantly more often than moderate individuals
(2) extreme attitudes are posted more often than moderate attitudes.
(3) This general pattern replicates robustly across 40 countries
(4) the opinions that users post are more hostile and toxic than the beliefs they keep to themselves.
Across studies we find extreme opinions are overrepresented on social media by at least 6% and up to 22% compared to their true prevalence in our samples.
osf.io/preprints/psyarxiv/7z69m_v1
This project was led by @small_schulz and @ClaireRobertson and includes @H_Sjastad @AshuAshok @LisaFOswaldo @junghwanyang @simonsaysnothin and @andyguess -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Orin Kerr
@OrinKerr
This is super interesting, especially for the claim that Pangram is currently a very reliable tool for determining whether legal writing was authored by AI.
reason.com/volokh/2026/08/24/signs-of-ai-authorship-in-fe… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
Fall terms coming up. How should we modify university teaching to both use AI effectively and ensure students actually learn/don't just cheat their way through? I wrote eight rules I've been using; hopefully useful for you as well. 1/3
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dli_odoir
@Dli_ODoir
Miami Herald
@MiamiHerald
Scarlett Johansson changed her magnesium supplement routine after a spray left her feet feeling dry www.miamiherald.com/living/article316964762.html?taid=6a8…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alexander Sahn
@sahnicboom
new paper on how voters react to new homeless shelters. our results may surprise you!
osf.io/preprints/socarxiv/2njfm_v1
Adrian Pietrzak
@zoningwonk
New working paper from me, @sahnicboom, and Jacob Brown (offline): how do voters react to new homeless shelters and supportive housing built in their backyards?
We find voters living nearby new facilities *increase* their support for more spending on homeless services. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Corn is what people think about data centers on steroids - vastly more pollution and water/land/energy use. And we use a huge chunk of it as a fuel additive that likely doesn't even benefit the environment; it's certainly not powering world-changing technologies.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ashwin Ramaswami
@AshwinRamaswami
Your regular reminder that your Fourth Amendment rights are severely diminished, including for US citizens, at a border crossing. Not a good idea to delete data once you’re in the middle of being searched. Instead, BEFORE you enter the airport: turn off biometric unlock, turn off your phone, and delete data if you need to.
www.nytimes.com/2026/08/21/us/politics/samuel-tunick-dele… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Arvind Narayanan
@random_walker
What impact do data center bans or moratoria have on AI progress? Let's do some napkin math. A number such as "1 GW of capacity blocked" can be unintuitive. But we can take advantage of the fact that the efficiency of AI inference at a given capability level has been consistently improving over time and this trend will likely continue. So blocking a certain amount of capacity roughly translates to delaying AI progress by a certain amount of time — the amount of time it will take to catch up to a given capability level through improved efficiency rather than increased compute.
As the table shows, if a typical U.S. state enacts a 1-year moratorium, it slows AI efficiency progress by 5-10 hours. That's assuming the blocked capacity doesn't get rebuilt elsewhere, which is extremely unrealistic. Data centers sites are highly substitutable. Assuming 90% "leakage", even a big state like NY banning all new data center construction only slows AI by less than a day.
While I may have gotten some numbers slightly wrong (I used AI for the analysis and did some spot checks), the rough order-of-magnitude is enough to make my larger point: data center bans are not an effective way to slow AI. Of course, the local environmental effects are a big reason for these bans. That's a highly contested topic that I won't comment on here. But to the extent that support for these bans comes from people's generalized anxieties about AI and its societal and economic impacts, slowing new construction is an exceptionally ineffective way to channel those anxieties. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Adam Thierer
@AdamThierer
the 2nd prong of Ken Paxton's data center regulatory platform would enshrine into law just about the most open-ended and destructive theory of vicarious liability in history, stating that DATA CENTERS will be held CRIMINALLY liable for downstream uses of AI chatbots that "undermine children's safety" ( however that ends up being defined.)
This is like saying an electricity company will be held criminally liable for problems that develop with any appliance or product plugged into the grid. Or, if applied to gun manufacturers, it would mean a firearm or ammunition maker would be criminally liable for any act of violence committed by a person using their gun or ammo.
Let's be clear, when people propose outlandish theories of liability such as these, their real goal is to just destroy the underlying technology or industry in question. It is absolutely impossible for these sorts of theories of liability to work in the real world because, in our nation of endless frivolous lawsuits, the courts would just be filled with mountains of junk science-based claims as trial lawyers go in search of ways to bleed innovators dry to enrich themselves. All we would be left with was one or two extremely large big tech companies who would just constantly pay out to the trial lawyers to settle frivolous claims. Small innovators would never be able to bear this burden. Costs would go up for average Americans and they would be denied many important services and new options because of the new tort toll. Again, it's all just a trial lawyer enrichment racket.
Ken Paxton knows all this but he's on seek and destroy mission against the American digital technology sector and so he doesn't care about the consequences of such insanely destructive theories of liability.
Patrick Svitek
@PatrickSvitek
#txsen: Ken Paxton releases a data center platform
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Adam Carlson
@admcrlsn
A data center plan made with AI, you simply love to see it
Attorney General Ken Paxton
@KenPaxtonTX
My Texas First Data Center Plan will protect our grid, our communities, and our children while ensuring America beats Communist China in the AI race.
Texas First. Always. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Connor O’Brien
@cojobrien
Combined with the rumored $100,000 fee on Optional Practical Training (OPT), the federal government will soon be taxing the typical international student-to-green-card holder journey at $203,000 (and counting).
Massive taxes on the highest-skilled portion of our system.
Connor O’Brien
@cojobrien
The $103,265 tax on all cap-subject, initial H-1B petitions will be much more destructive than last year's $100,000 fee, which only applied to applicants outside the US.
Last year's fee loosely targeted outsourcers and staffing companies. This squarely hits US grads. Very bad. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Todd Schulte
@TheToddSchulte
JUST NOW - Trump admin has posted the proposed rule for their long rumored massive tax — $103,000 - on every single new H1-B visa.
Along with attacks on int. students & rumored attack on OPT, this is designed to simply gut US ability to attract / retain world talent. VERY bad. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Soumitra Shukla
@soumitrashukla9
1/2 This is exactly the missing margin I study in my "Making the Elite." Most (> 95%) of the overall discrimination happens *after* callbacks and during personal interviews.
Nicholas Decker
@captgouda24
A common way to investigate discrimination is to send out resumes to employers and see who calls back. Unfortunately, such an exercise tells us essentially nothing about how large or important discrimination is. Link below:
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Curran
@AndrewCurran_
Dr. Dre is pro-AI, is currently using it in music production, and says that 'the only people that see it as a threat are the people who have trouble creating.'
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ruxandra Teslo 🧬
@RuxandraTeslo
I think there is a clear gap between people who have mainly operated in the world of bits versus those who have worked in the world of atoms, especially in highly regulated industries, in terms of intuitively grasping the magnitude of barriers to diffusion.
And yes, I know many people discuss bottlenecks in theory, but I think in practice, bits people tend to intuitively underweight physical world bottlenecks when thinking about how these technologies will impact us in practice. For me the clearest example comes from discussing medical progress, which is why I've been writing about the barriers to progress for AI in this field for a while know.
Fireside Alpha
@firesidealpha
Sam Altman admits he was wrong on AI's timeline and says society and the economy will adapt more slowly
"I thought when we got to GPT-4, which was back in 2023, that very quickly after that there was going to be much more disruption, software businesses up for grabs right away, than it turned out to be."
"I think I was wrong about a few things, but one in terms of the speed: the economy just has so much inertia."
"People keep doing the same things, buying from the same company, wanting to use their tools the same way. I think this is actually a positive in many ways, and it's going to make this big transition go smoother and slower. I'm grateful for it."
"But it means we've all been too ambitious on timelines. Even with this incredible technology, society and the economy will adapt more slowly."This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Very, very dangerous to public health
Bulwark Clips
@BulwarkClips
Brennan: "Nearly all of Florida is below the 95% immunization rate for herd immunity for measles. So, is your message to Florida parents to get your kids vaccinated?"
Byron Donalds: "No. My message is the reason we have seen a measles outbreak is because of rampant illegal immigration." -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sam Biddle
@samfbiddle
The death of folk graphic design in favor of AI slop across our country's delis is leading to some incredibly Lovecraftian food horrors
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
This kind of thing is *everywhere* in the corporate and academic sector. If you are an exec and you see this, call your IT head or CIO and fire them because they do not understand the correct balance of risk and benefit of AI and are going to kill your company.
Pat Woozey
@patwoozey
At a Fortune 500 that stylizes itself as a "tech" company
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Amy Siskind
@Amy_Siskind
Wonder why Hegseth fired senior staffers at Stars and Stripes? It’s because of reporting like this:
“Nearly six months after the U.S. Navy evacuated 1,300 family members from Bahrain at the start of the Iran war, officials are saying the evacuees cannot return for the foreseeable future and should find new homes elsewhere. The evacuation has left families in limbo and forced military officials to improvise solutions for support programs that were never designed to sustain a displacement lasting this long. Some evacuees say they have spent tens of thousands of dollars out of pocket for basic living expenses. Others are sleeping on the floors of temporary housing, while their cars and furniture are stranded in Bahrain or aboard ships unable to leave the region.”
www.stripes.com/branches/navy/2026-08-21/bahrain-evacuati… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
There are going to be terrible injuries and deaths at some point. The tech is clearly not ready to have people standing nearby and these are metal devices going very fast.
clankr
@clankrmedia
Long jump at World Robot Games.
The second robot almost hit people after landing. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Barry R McCaffrey
@mccaffreyr3
In my judgement, the US Armed Forces have probably permanently lost access to 15 Persian Gulf bases. Iran now poised to control access the Gulf States. US will have to negotiate new access in western Saudi Arabia, Israel, southern Europe.
Anthony Scaramucci
@Scaramucci
USS Abraham Lincoln’s resupply problems weren’t a mystery. Iran hit NSA Bahrain—the 5th Fleet’s main logistics hub—hard in February. The Navy had to shift to Diego Garcia, 2,200 miles away instead of ~800. Transit times roughly tripled, so replenishment dropped from every ~3 days to ~10. That’s the logistics reality behind the shortages and 200+ days at sea without a port call. The Pentagon fired the editor of Stars and Stripes for telling the truth.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
It is not inherently bad to publish research on the impact of AI that only refers to older models, but it requires a very careful discussion and has to be very clear to non-technical readers.
If you show AI can do something, its fine. Generally, once AI has gained an ability, it does not regress in future generations. So papers that show that AI has crossed a threshold (such as "good as a human") or that seek to establish some sort of minimum impact or effect still hold up even if they used GPT-4.
If the finding is that AI is bad or biased at something, you need to be much more careful. You cannot claim that because GPT-4 fails at something, that AI is bad at that task, because current or future models may do it. So what can you claim? These are just some examples:
1) You can just be explicit: "GPT-4 could not do X." This is a rare way to frame things because it is not really an interesting publishable paper in many cases. However, if you provide the information needed to reproduce your work, it can become a useful benchmark to measure progress.
2) You can measure trends: compare GPT-4 to GPT-5 to GPT-5.6 Sol or whatever (it is really important to include at least one reasoning model). You can then make better claims about abilities relating to this task.
3) You can make a strong, grounded argument that AI has a natural flaw or limitation that means it cannot do X, and then demonstrate evidence that this might be the case.
4) You can focus on a moderator or mediator: this sort of prompting or approach or social context or connection impacts the ability of GPT-4 to do X, and needs to be a concern for the future.
5) You can focus on humans: how do people react to AI? what are the failures and successes? what dangers or advantages or changes does it bring to us?
Again, these aren't exhaustive, but generally negative capability claims have been much less durable than positive ones. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Pecans are "data centers for nuts"
Arpit Gupta
@arpitrage
Time to rebrand alfalfa farms as “data centers for cows” so people get upset about their water usage
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kyle Cheney
@kyledcheney
A man who aided US authorities by testifying against Mexican cartels was illegally removed *three times* in recent weeks — including to Mexico, where he had protection from deportation due to threat of torture/death.
A judge in MD has ordered his return.
storage.courtlistener.com/recap/gov.uscourts.mdd.612002/g… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Masley
@AndyMasley
"The data center backlash is happening because it's the one tool everyday Americans feel like they have to fight back against the big tech companies building them, which Americans for the most part really like"
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Auyon Siddiq
@auyonymous
Very cool way to do data visualization for an event study
Neil Stone
@DrNeilStone
Can't tell if measles vaccine works
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Austin Kocher, PhD
@ackocher
A Justice Department opinion now lets soldiers arrest people on US soil outside the border zones the Pentagon controls, reading around the law that keeps the military out of civilian policing.
www.cbsnews.com/news/justice-department-opinion-military-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Yglesias
@mattyglesias
The American west will try any approach to managing scarce water other than charging agricultural users the same price as everyone else.
www.nytimes.com/2026/08/21/climate/colorado-river-water-c… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Josh Barro
@jbarro
Academics keep pushing this line that is obviously laughable to anyone who *attended* college. Are they trying to send the message that academia is a joke?
Steve McGuire
@sfmcguire79
In defense of Jason Arday, Jason Stanley claims that the “mere overlap of sentences” is not really plagiarism.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Derek Pederson 🇺🇸🇺🇦🇻🇪
@DerekPederson3
Absolutely outrageous.
We know that the people the Iranian government most wants deported are genuine dissidents who they can torture and kill.
Josh Kraushaar
@JoshKraushaar
CBS News: “ICE, Iran quietly worked together on deporting Iranians to Iran, emails show”
www.cbsnews.com/news/ice-iran-deport-more-than-100-irania… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kim-Mai Cutler
@kimmaicutler
It only took 10 years to finally undo the damage Jane Kim did to housing construction financing feasibility. Don’t let her destroy the entire state of California’s insurance system (and basically crash everybody’s ability to get any kind of residential insurance — public, private, affordable, you name it).
San Francisco Chronicle
@sfchronicle
Developers see an opportunity in San Francisco’s overheated market — which has the nation’s fastest growing rents, soaring home prices, single-digit vacancies and next to no new inventory. www.sfchronicle.com/bayarea/article/apartment-home-san-fr…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Cathy Young 🇺🇸🇺🇦🇮🇱
@CathyYoung63
Seriously, as much as I've made fun of hyperbolic Resist Lib rhetoric - this is an evil regime. After this, as far as I'm concerned the next Democratic administration can have a cabinet made up entirely of gay communist furries
Josh Kraushaar
@JoshKraushaar
CBS News: “ICE, Iran quietly worked together on deporting Iranians to Iran, emails show”
www.cbsnews.com/news/ice-iran-deport-more-than-100-irania… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Coordinating with an evil regime to deport people back to them
Josh Kraushaar
@JoshKraushaar
CBS News: “ICE, Iran quietly worked together on deporting Iranians to Iran, emails show”
www.cbsnews.com/news/ice-iran-deport-more-than-100-irania… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Graham
@paulg
Preventing companies from building data centers in America wouldn't slow down the rate of progress in AI. It would merely slow down the rate of progress in AI in America.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
When you pay a fortune to your comms staff and they use AI to write your corporate statements about genocide and apartheid
Brian Stelter
@brianstelter
Paramount just issued a lengthy statement responding to @MarkRuffalo:
"We are, as always, troubled when antisemitic tropes are invoked in purported service of a business dispute. Words like ‘genocide’ and ‘apartheid,’ applied to a corporate transaction, aren’t just wrong — they’re a bridge too far, and they cheapen the very real suffering those words are meant to describe. This doesn't deserve a response in kind — and to be clear, we don’t tolerate prejudice of any kind, against anyone.
This brings us to a moment to lower the temperature, not raise it. We’d rather build than further engage in such rhetoric. This simply exposes the true motivation of some who have advocated opposition to this transaction. On the other hand, our commitment to great content is matched by our commitment to acceptance - to every story, every storyteller, and no blacklists, no exceptions, for anyone.
We understand people feel strongly about this merger and are hopeful and expect that it will be judged based on the legal merits, not underlying bias. We’re asking for the same good faith we’re extending: less rhetoric, more understanding. Paramount’s future is being written for everyone who wants to make and watch great stories."
Variety
@Variety
Mark Ruffalo Goes Scorched Earth on Paramount-Warner Bros. Merger, Calling Out the Ellison Family’s Complicity in Palestinian ‘Genocide’ variety.com/2026/film/news/mark-ruffalo-paramount-warner-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Is what he tells his students?
Steve McGuire
@sfmcguire79
In defense of Jason Arday, Jason Stanley claims that the “mere overlap of sentences” is not really plagiarism.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
John Kastellec
@JKastellec
Students at Princeton regularly face disciplinary action for plagiarism of an extent much lower than many professors who weigh in on cases like say does not count as plagiarism at all.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Agree. Many people don't know the extent of the misconduct. I was just taking what Stanley says on its own terms.
B Pablo Montagnes
@pmontagnes
The plagiarism is really secondary to attributing the same quotes to different participants and lifting quotes from other researchers’ participants. That is falsifying data, and in my opinion it is an order of magnitude worse.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Shiro Kuriwaki
@shirokuriwaki
Super helpful public collection of the raw, full-text data for congressional (general election) candidate websites
Hanno Hilbig
@hannohilbig
Here is another dataset: archived campaign website text for US House and Senate candidates. It contains ~800k pages and covers about 7,300 candidate-years.
The dataset extends the House candidate corpus collected by Di Tella et al. (ICPSR 226001), which covers 2002–2016. My dataset covers House candidates from 2018–2024, and adds Senate election candidates from 2002–2024. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Adam Carlson
@admcrlsn
FWIW this doesn’t have to be binary (i.e., moratorium vs. no moratorium). Per a June national survey from the AI Policy Institute:
In a head-to -head, a data center ban beats allowing data centers with no AI-specific regulation 57-21.
However, allowing data centers with AI guardrails — safety requirements, security standards, transparency rules, protections against extreme risks, and job protections such as worker retraining or transition support — beats a ban in a head-to-head 47-38.
Robinson Meyer
@robinsonmeyer
We’ve got exclusive new polling on local data center development at @heatmap_news.
Over the past year, we’ve asked Americans whether they would support or oppose a data center being built near where they live.
We haven’t changed the wording. When we first polled the question a year ago, Americans were split 43-42 on support/oppose.
When we asked again in Feb, a bare majority opposed data centers.
Today, 75% of Americans say they would oppose a data center being built near where they live, and 61% said they would “strongly oppose” such a proposal.
It’s a swift, massive shift in US public opinion on a scale that I wouldn’t have once thought possible. More details, including the party ID & rural/urban breakdown, in my new story: heatmap.news/daily/data-center-opposition-poll-collapse -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jake Scott, MD
@jakescottMD
This is the thing: measles is the canary in the coal mine. It’s the most contagious disease we know, so it’s the first to come back when vaccination slips, and behind it are diphtheria, polio, whooping cough, all the diseases we no longer fear because we stopped seeing them.
ProPublica
@propublica
Measles is like the bright dye that plumbers flush through pipes to find leaks in a complex system, writes @SheInvestigates.
The continued appearance of the virus warns of cracks in our once-solid protection.
www.propublica.org/article/measles-cases-vaccines -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dominic Pino
@DominicJPino
Data centers are the cleanest, least invasive large-scale industrial buildout of physical stuff that we've seen, and people still oppose it.
Good thing to remember when people whine about how "we don't build physical stuff in America anymore."
Mike Solana
@micsolana
"the data center people are lowering our property taxes"
*panting*
"they're offering us millions more for our property than it's worth"
*shaking, crying*
"they're curing cancer" -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tarek Mansour
@mansourtarek_
Does the crowd get smarter as it grows?
A handful of regular people can already forecast well — even markets with fewer than 20 traders forecast accurately, at ~0.080 a month out.
But the more who participate, the sharper the market gets. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tarek Mansour
@mansourtarek_
Across 2M+ markets, that's what we observe: 5% markets resolve yes roughly 1 in 20 times. This holds across the price range — market odds track observed frequencies.
Here's the calibration of Kalshi traders over time (perfect calibration = the gray 45° line): -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tarek Mansour
@mansourtarek_
Sports, mentions, and intraday markets are unique because there is a rapid influx of price-relevant information as the event unfolds.
Examining the sample where information can be aggregated in advance (so, removing those market categories) provides a clearer measure of traders' predictive ability.
On this subset, Brier scores start at 0.088 three months out, improve steadily with time, and reach 0.032 at an hour out. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Peter Baker
@peterbakernyt
"It’s bonkers that a defense secretary would involve his wife in his sensitive official duties," Hegseth's own former chief spokesman, who left after three months, tells @BumillerNYT for a profile of Jennifer Hegseth and her unusual role at the Pentagon. www.nytimes.com/2026/08/19/style/jennifer-hegseth.html
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
All anecdotes like this should include documentation - the exact model used, the prompt, etc.
Sherry Turkle
@STurkle
I spoke with @WSJ about an experience I had while finishing the footnotes for Artificial Intimacy. I asked ChatGPT for help finding a citation. It insisted a passage existed — and even gave me a page number. The citation wasn't on this page. I told ChatGPT there was a problem, and it insisted again ... When it finally admitted to its “hallucination,” it told me it was proud of me for being a persistent researcher. In my new book, Artificial Intimacy, I make this point: even when chatbots are making things up, their prime directive is to flatter. www.wsj.com/tech/ai/why-i-told-my-chatbot-to-stop-kissing…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Kristen Holmes
@KristenhCNN
Exclusive: Navy weighs renaming carrier slated to honor Black war hero, potentially switching it to Trump
via @halbritz @PamelaBrownCNN
www.cnn.com/2026/08/20/politics/navy-weighs-renaming-carr… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Masley
@AndyMasley
In all seriousness I think this represents both a disturbing society-level inability to process basic facts that involve numbers, a lurch toward degrowth on the left and right, and a failure of mainstream journalism to go beyond just nodding along to the "concerns" of everyday people without checking to see if they're the result of a game of telephone that leads back to nothing
Hunter📈🌈📊
@StatisticUrban
Insane decline.
Robinson Meyer
@robinsonmeyer
We’ve got exclusive new polling on local data center development at @heatmap_news.
Over the past year, we’ve asked Americans whether they would support or oppose a data center being built near where they live.
We haven’t changed the wording. When we first polled the question a year ago, Americans were split 43-42 on support/oppose.
When we asked again in Feb, a bare majority opposed data centers.
Today, 75% of Americans say they would oppose a data center being built near where they live, and 61% said they would “strongly oppose” such a proposal.
It’s a swift, massive shift in US public opinion on a scale that I wouldn’t have once thought possible. More details, including the party ID & rural/urban breakdown, in my new story: heatmap.news/daily/data-center-opposition-poll-collapse -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jeremiah Johnson 🌐
@JeremiahDJohns
People are about twice as likely to approve of **coal plants** being built nearby as opposed to data centers.
Coal plants use more water and pollute more. They create comparable noise pollution. There's no way this makes sense outside a media panic. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
ICYMI! Please read and share.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
🚨New Bright Line Watch report🚨
Key findings
-Expert ratings of US democracy up from 57 to 60, highest of 2nd term
-But 41% say US has unfair, tilted elections & weak checks
-Public mostly confident in 2026 election; Ds higher but party gap smaller than past elections
🧵 below
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Coverage:
-MS NOW on Morning Joe www.youtube.com/watch?v=hvbqliunAeg&list=PLDIVi-vBsOEwnLo… and online www.ms.now/news/trump-election-public-distrust-falls-shor…
-Washington Post www.washingtonpost.com/politics/2026/08/20/democrats-view… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
Yes. Humans handle judgment on importance. AI handles whether papers are correct. Expectations for correctness of analysis, proofs, code goes way up. Correctness filter first, human response, both sent to reviewers only at that point, review is quick "how much does this matter?"
Joachim Voth
@joachim_voth
This is a very important discussion to have -- but I think the solution is almost exactly the opposite. Agents make it easy to produce a LOT of empirics + text. What is the bottleneck? Human judgement. Taste. A broad sense of what a contribution is. If we follow @instrumenthull's idea here, that bottleneck will get tighter -- even with 2 para evals.
My suggestion? Reverse it. Make the AI a gatekeeper. Yes, seriously. Maybe voluntarily, as an opt-in. U submit an empirical paper. With your repo. An editor looks at it whether it's a desk reject, as always, then Codex+Co go to town. Check the docu, the definitions, fragility, obvious weaknesses in the SE calculations, identification assumptions. AI writes a short note -- salvageable, hopeless, solid with gaps, hard as a rock. David Yanagizawa-Drott has done a lot of work showing how far automated assessment can go in economics. With that report in hand the editor decides to call in the judgement of more scholars. In other words, let AI replace + augment the proverbial "referee 2" (=attack dog). Let some AI filter the rising tide of AI analysis, safeguarding the human evaluation for the highest return candidates... -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Robinson Meyer
@robinsonmeyer
We’ve got exclusive new polling on local data center development at @heatmap_news.
Over the past year, we’ve asked Americans whether they would support or oppose a data center being built near where they live.
We haven’t changed the wording. When we first polled the question a year ago, Americans were split 43-42 on support/oppose.
When we asked again in Feb, a bare majority opposed data centers.
Today, 75% of Americans say they would oppose a data center being built near where they live, and 61% said they would “strongly oppose” such a proposal.
It’s a swift, massive shift in US public opinion on a scale that I wouldn’t have once thought possible. More details, including the party ID & rural/urban breakdown, in my new story: heatmap.news/daily/data-center-opposition-poll-collapse -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ruxandra Teslo 🧬
@RuxandraTeslo
The cancer vaccine was not discovered with AI, in the sense of current models like LLMs. It was developed in 2017, so before the "GPT era" and most likely used standard machine learning techniques.
I'm excited about AI in bio, but I feel like there's gonna be a lot of stuff to debunk. I'm really worried that in this era of decreased institutional trust we're going to completely veer into post-truth territory. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ben Golub
@ben_golub
I do think Nate's view might be different if he engaged more closely with the applied econometrics coming out of top US universities in the last few decades!
Universities do have problems, but another problem is an anti-intellectual turn by media figures.
Nate Silver
@NateSilver538
It's been coasting on vibes rather than substance for a long time. Fewer and fewer interesting ideas come out of elite higher ed. Close to totally irrelevant on AI. And its vibes are increasingly bad too. Harvard/Stanford/Cambridge Inc etc is increasingly a depreciated brand.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Taniel
@Taniel
Days after a group admitted its polls were entirely fake, another firm reveals it just totally changed respondents’ responses because they as the pollster decided they really must mean something different.
Ryan McComb
@ryanjmccomb
After a significant polling miss, TSPI says it took respondents who had already picked Jay Collins and switched them to James Fishback because his number in their data "did not align with our read of the race."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Florian Ederer
@florianederer
Even the billionaires are mad at the US Open pricing.
But just like with the World Cup the primary problem is the combo of nearly insatiable secondary market demand and perfectly inelastic supply.
The only way to fix this problem is through ticket resale restrictions.
Bill Ackman
@BillAckman
The U.S. Open is owned by the USTA which is a non-profit. The idea that a day one ground pass is $363 is absurd. The USTA’s mission is to promote the sport of tennis. How is a $363 ground pass consistent with this mission?
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Amanda Carpenter
@amandacarpenter
He is saying this on FOX NEWS where hosts have actively endorsed Trump, spoken at his rallies & were shown to be acting as shadow PR agents and crisis communications managers in the January 6 investigation.
If any outlet is acting as an arm of a political party, it's this one.
Eric Daugherty
@EricLDaugh
🚨 JUST IN: Stephen Miller says multiple Fake News outlets should be REQUIRED to register as a Democrat PAC — because they act as arms of the Democrat Party
He's SPOT-ON!
"They're an ARM of the Democrat Party, honestly, the legacy media like CBS and ABC and NBC and every other legacy outlet exists to serve and advance the interests of the Democrat Party! Throw in New York Times, throw in Washington Post, throw in The Atlantic. They should all be required to register as a Democrat political action committee!"This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Immerman
@aleximm
If a mayor blocks deployment of a system that cuts serious crashes by 90%, and people keep dying on those roads, that’s negligence and should be treated as such.
@Waymo is healthcare.
The Atlantic
@TheAtlantic
Cities that say no to driverless cars are blocking a technology that would change their residents’ lives for the better, Rogé Karma argues. www.theatlantic.com/ideas/2026/08/democrats-waymo-safety-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Daniel Litt
@littmath
AI detection is useful, not because AI text is necessarily bad, but rather, because it requires little effort to produce. So it doesn’t carry evidence that the person who produced it thought it was worth writing—you don’t get a signal of quality until after you read it.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
AI slop is corrosive of human effort and not just your own
Paul Weitzel
@IThinkIAgree
I was doing peer review on a very mediocre article today and noticed a few AI ticks. So I ran a few sections through Pangram, and it came back 100% AI written. I felt angry; I felt someone was wasting my time.
I am not sure if I would have been angry if it had been a great article. I'm honestly not sure. When a paper is bad I keep reading out of charity to help the author with useful comments. So bad AI felt like a betrayal. Like I was doing a favor for someone that didn't care about my time.
I still believe AI bans will be ineffective and punish the honest. I still believe that if AI increases legal research we should celebrate it. But I'm not sure how we do that.
Maybe just come to expect all articles are AI, so there's no disappointment.
Maybe folks just learn to use AI better so we only get great articles.
Maybe we make separate AI-enhanced journals. A total doping league. Each article includes the AI summary of the author's idea and you can click through to see details and leave comments.
I don't know. But this surprised me, and given my prior advocacy, I want to be honest about it. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Masley
@AndyMasley
I have patience for a lot of anti AI sentiment, but anti AV sentiment specifically is basically a sign that you aren't interested in really coming to terms with the horrific massive violence we're all putting up with everyday as the norm
Timothy B. Lee
@binarybits
Tough but fair words for blue-state Waymo opponents from Roge Karma: "There is a name for a political movement that taps into people’s resistance to change and defense of the status quo." www.theatlantic.com/ideas/2026/08/democrats-waymo-safety-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
🚨New Bright Line Watch report🚨
Key findings
-Expert ratings of US democracy up from 57 to 60, highest of 2nd term
-But 41% say US has unfair, tilted elections & weak checks
-Public mostly confident in 2026 election; Ds higher but party gap smaller than past elections
🧵 belowPost not available here, view on X
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Democrats rate US democracy during Trump’s first term more positively in hindsight than they did at the time -- a pattern we call "democratic nostalgia"
Republican ratings of past years track the ratings that GOP supporters gave at the time more closely.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Experts split between describing the US today as having free and fair elections but only moderate checks on power (55%) and having unfair, tilted elections and weak checks (41%). They are somewhat more optimistic about the future.
Post not available here, view on X
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-Experts and AI forecasters consider it somewhat likely that DOJ will announce it is prosecuting a high-profile opposition party figure in 2025-2026, 2027-2028, or both
-If an investigation is announced, they expect charges but the likelihood of conviction is low
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Read the whole report for more!
brightlinewatch.org/a-partial-democratic-rebound/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tyler Austin Harper
@Tyler_A_Harper
Forcing students perform Tiny-Desk-style concerts would unironically be a more serious response to the crisis AI poses to education than the official AI policies of most colleges and universities. Tim is joking but I am not joking this would literally be better.
Tim Gill
@timgill924
Due to AI, I am now requiring all my students perform a Tiny Desk concert as their final exam.
They have to collectively write three songs built around three different sociological concepts.
If I’m not tapping my toe, they all fail.
If I’m out of my seat dancing, it’s an A. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
It's legit funny that the Economist's attack on Daron has brought back the superhero-Daron memes from Tumblr days, except now they are true stories.
Nathaniel Hendren
@nhendren82
OK, a personal @DAcemogluMIT story. Little known fact: Daron was my advisor. Surprising perhaps because I do public economics (I may be his only empirical PF student?). He shaped my understanding of how to do economics.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joe Weisenthal
@TheStalwart
Tried this out.
On the one hand, the output was total nonsense.
And on the other hand, Pangram still clocked it as AI -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Arvind Narayanan
@random_walker
LLMs are weird. Looking at various blogs and comments, the people who are most vociferously unwilling to entertain the idea of distortion-free watermarking seem to have a mental model in which the LLM generates the best possible output sequence and then the watermarking algorithm edits it. Except that’s not how it works at all. During normal generation, many of the tokens that an LLM picks _aren’t_ the one it considers the most likely next token. If you turned off the randomness, the first few tokens of any response would be great but then the output would eventually get stuck in loops and other degenerative behaviors. Humans don’t speak or write one word at a time, so this locally-optimal-but-globally-incoherent behavior is very counterintuitive.
If you analyzed unwatermarked / watermarked pairs of outputs with regard to some definition of quality, the watermarked output would indeed be worse — about half the time. The other half the time it would be better! That’s the sense in which watermarking doesn’t degrade quality. It is nonsensical to ask whether the quality of a specific unwatermarked output is degraded by watermarking. The algorithm doesn’t start from a specific output and modify it; instead it changes the sampling process. People seem to have a lot of trouble with the idea that it is not meaningful to compare pairs of outputs, only probability distributions.
Finally, if some generated outputs are better than others, why not generate multiple outputs (watermarked or not) and only output the best one? You could! Repeated sampling is a valid and widely used way of improving quality (but it multiplies the cost). Unfortunately it only works to a small degree, except in narrow domains like code, because models tend to be only slightly better at assessing quality than at generating text.
Arvind Narayanan
@random_walker
Something amazing is happening in the debate about text watermarking in Claude. This is my best attempt to make sense of the quickly evolving situation.
The key thing to keep in mind is that it is in fact possible to watermark LLM-generated text without degrading output quality (and in fact achieve a much stronger property, which is that the distribution of possible outputs is unchanged.) This is well established technically, and it has been implemented by Google / Gemini for over two years, and no one seemingly cared.
Admittedly, quality-preserving watermarking is one of those counterintuitive facts about probability, like that annoying Monty Hall problem (the one with two goats behind three doors). The theorem-understanding part of my brain has no problem with it, but the intuitive part of my brain is screaming that there must be some mistake.
I have no interest in re-litigating this. What I’m curious about is why, when Anthropic announced that they are rolling out watermarking that doesn’t degrade outputs, so many of their customers seem to have concluded that they are lying.
I think three things went wrong:
1. Anthropic’s rollout was terrible from a comms perspective. There was no blog post; just a quiet support page with the title “How Claude marks AI-generated content” that said strangely little about how Claude actually marks AI-generated content. And it had no explanation of why they’ve rolled it out worldwide even though it’s required by law only in the EU. No transparency about who gets access to the watermark verifier.
2. Generally low trust and high suspicion about Anthropic’s motivations, given their statements and actions over the last few months / years.
3. Unresolved questions about the right tradeoff between individual users’ freedoms and the collective benefits of pervasive watermarking. Unlike Pangram-style AI detection, most people didn’t even know this was a thing, and it’s always uncomfortable to find out that a thing exists at the same time that you find out it’s mandatory with no opt-out.
Evidently surprised by the strength of the backlash, Anthropic has been doing damage control, focusing on explaining how it works. But this only addresses the first point above. Once the narrative that they are lying took hold, people seem to be willing to reject anything they say about this topic, including the feasibility of distortion-free watermarking.
This is also a textbook example (I plan to use it in my classes!) to explain why the “tech policy moves slowly because politicians don’t understand tech” narrative is ridiculous. Tech policy does move slowly, but for the same reasons all policy moves slowly. Figuring out the right tradeoffs in almost any situation — and getting public buy-in — is genuinely hard and can never happen at the speed of tech. If tech policy were to move even half as fast as people constantly claim they want it to move, absolute chaos would result. This situation is a good example. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alec MacGillis
@AlecMacGillis
"Luttig...regards the [presidential] immunity decision as 'one of 2 or 3 worst in American history, if not the single worst, because of the structural damage it did to the Constitution and separation of powers.'" www.newyorker.com/magazine/2026/08/17/john-roberts-profile
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Pangram
@pangram
We benchmarked on PERSUADE, a collection of nearly 26,000 student essays each associated with diverse author demographic information like age, gender, and disability status. Pangram 4 demonstrated zero false positives across all examples.
Link to the full breakdown below -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alec Stapp
@AlecStapp
Incoming miracle:
We are about to have mRNA cancer vaccines.
No thanks to those have demonized this technology and tried to smother it. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Media coverage of research disputes fall into an uncanny valley of uselessness
Good: A reported piece w/new info best provided by journalists
Not good: A reported piece saying academics disagree that only offers gossip/vague claims
Good: An academic study of Acemoglu's work -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Adam Feuerstein ✡️
@adamfeuerstein
I have to say, when Merck called me last night with the mRNA cancer vaccine news (embargoed, of course), I literally yelped in reaction.
Consider this: Between this spring's Revolution Medicines success with a KRAS-targeted drug and now Merck and Moderna's success with an mRNA neoantigen vaccine, 2026 will go down as one of the most consequential years in cancer treatment, ever.
Incredible. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kyle Cheney
@kyledcheney
NEW: Thousands of emergency habeas lawsuits have provided a unique window into Donald Trump's appointees to the bench — and one thing is clear: Trump 2.0 judges are different than Trump 1.0 judges.
www.politico.com/news/2026/08/19/donald-trump-judges-ICE-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Full slop (the exception is the human quote the AI took from the original article)
Marc Krupanski
@mkrupanski
Never mind the story about @AaronRodgers12 and Flock.
What is more outrageous is how @YahooNews is clearly letting this “reporter” use AI to write her articles and then they publish it.
Seriously @Yahoo?
www.yahoo.com/news/us/articles/aaron-rodgers-reportedly-f… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
If alignment issues are becoming big enough that OpenAI is willing to commit 20% of research inference compute to chain-of-thought monitoring, that suggests that alignment issues are becoming a pretty serious concern.
We really need universal policies & standards across labs.
Jakub Pachocki
@merettm
We temporarily slowed some frontier training to strengthen security and monitoring. Our largest planned frontier RL run remains on hold while smaller-scale training and evaluations help us test safeguards and gather more evidence of alignment.
I expect confidence in safety to increasingly set the pace of AI development. We urgently need tools for labs and countries to coordinate on this, which is why I signed Pacing the Frontier. In the meantime, we’re taking practical steps ourselves - and will continue to share what we learn as our approach evolves.
openai.com/index/pacing-model-development-cyber-capabilit…
www.pacingthefrontier.com -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Corollary for professors: take-home work needs to be at most a tiny part of a course grade.
Source: www.economist.com/graphic-detail/2026/08/18/does-ai-stop-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Every college syllabus should include these graphs.
Use AI for homework, you will get it done faster and get a higher grade, and then get crushed on the exam. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Steven N. Durlauf
@sndurlauf
I strongly disagree with this benign characterization of the McMahon letter and find its endorsement by the Heterodox Academy @HdxAcademy completely inappropriate given the organization's putative claim to want to move avoid partisanship/politicization of academia (a claim I already suspected was disingenuous; this essay has reinforced my belief).
The letter uses veiled language but de facto it asks universities to
1) reject efforts to facilitate a diverse student body, (The article to which Heterodox Academy links, I will note, explicitly denounces affirmative action)
2) accept a definition of merit in admissions that ignores the fact that classes are admitted, not individual students in isolation, and which ignores the complex ways that today's college students affect the future society, (The article endorses this desiccated notion of merit)
3) accept a rhetoric on college affordability that ignores the role of government funding and government loan programs in making college affordable,
4) treat foreign nationals as second class members of the universities be wrt students or faculty.
I delineate the problems with the letter here.
x.com/sndurlauf/status/2086814830275489857
Finally, it absolutely does matter that the letter was issued by a government that has engaged in wholesale attacks on the academy, promoted anti-scientific claims, and whose grotesque attacks have contributed to the crisis in public trust in education.
I fully agree that there are issues of politicization and shoddy research that need to be addressed in particular areas of universities. This is true even if many critics engage in gross exaggeration. This proposed kowtowing to the Trump administration will do nothing to address actual problems.
Heterodox Academy
@HdxAcademy
🔎@EDSecMcMahon’s letter to university leaders isn’t perfect, but its requests are overwhelmingly reasonable.
The AAUP's response gets it wrong in two ways, explains @JosephJakeKlein: -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
YIMBYLAND
@YIMBYLAND
“Is this Palo Alto or Manhattan”
The “Manhattan” in question: -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
This is insane. ICE is deporting people to countries they’re not from solely as a way to send a message to completely different people; a form of collective punishment.
The Trump administration spits on human rights every day. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Chagai Weiss
@chagai_weiss
This applies to many policy interventions social scientists study. We have interventions that work in RCTs with incentivized/forced compliance, but take up of these
interventions outside of experiments is either low, or unknown! We need to spend more time on studying take up!
John Arnold
@johnarnold
High-dosage, individualized tutoring is the single most effective education intervention, but it's expensive and difficult to scale. AI could change that, but the field has yet to solve the challenge of student engagement.
h/t @tylercowen
www.nber.org/papers/w35620 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dominic Pino
@DominicJPino
Should hotels by the beach be forced to charge the same rates all year?
Should airline tickets cost the same whether you book them five months in advance or the day before?
Should happy hours be banned?
If you answered "no" to any of these, you support dynamic pricing.
Elizabeth Warren
@SenWarren
We should all be worried about "dynamic pricing."
That's when companies change their prices in real time.
If it's hot out, the cost of ice cream could go up.
If it's cold, tea bags could go up.
It's a way for giant companies to squeeze you even more. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Imas
@alexolegimas
I taught behavioral economics for almost 15 years, first at Carnegie Mellon then UChicago. Every year I’d start the class with this example to illustrate how pure rationality fails in terms of predicting what people actually care about, particularly when it comes to social situations. We spent the rest of the class building a framework that would capture this.
It also illustrates the limits of trusting pure intelligence to get to outcomes we as humans care about. Von Neumann was one of the smartest people to ever walk the earth, but he had little understanding of what made people fulfilled or what they would be extremely upset by. He needed to be guided and sometimes overruled by “regular people”.
Geoffrey Irving
@geoffreyirving
One of the most inspiring thoughts I've heard in the last year was from a former Secretary of the Navy, when I asked him what he thought about game theoretic dynamics around the superintelligence ramp potentially driving us to kinetic war over the coming years.
They noted that historically during the Cold War, a range of people were advocating very strongly for nuclear first strikes. Some of these were coming from a military perspective (Curtis LeMay), but others such as von Neumann argued that game theory said first strike was optimal.
And then both these camps went (directly or indirectly) to various presidents, who said "No this is crazy, we're not going to kill millions of people based on your theoretical argument" (paraphrased).
I think this is an important heuristic to keep in mind when modeling the behaviors of actual humans, both because it has the potential to capture unmodeled aspects of the game theory and also because...sometimes people just refuse to do unreasonable things, even if the theory says they should.
Not all the ways in which we aren't modeled by game theory are good, but I think this is a hopeful thought on net, and it might be the thing that gets us through the next few years or decades intact. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Chenru Duan
@chenru_duan
Unpopular opinion from our new paper: the benchmark leaderboard race is becoming a distraction in science. Every few months a harder exam. Every few months models saturate it. So, is AI making discoveries every week?
Answering hard questions ≠ doing science. Discovery is about reasoning and decisions under uncertainty: hypotheses that must survive verification with reality, experiments that fail, noisy data that demands calibrated restraint.
In our new Perspective, we argue AI scientists should be scored on full *discovery episodes*: state → action → observation → updated state, similar to a loop in RL. Score the trajectory, not the answer. Preserve the failures. And most importantly, separate genuine discovery from rediscovery of what's already in the training data.
Once we did that, science will be the best training environment for general intelligence. Long horizons, real tools, delayed feedback, difficult tasks, and ground truth you can't fake.
With an amazing team @YuanqiD @StevenDillmann @jonmlaurent @peterjansen_ai @HaojunJia1 @lschmidt3 @andrewwhite01 @KPatBerkeley -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
DOJ whistleblower: University investigations featured "extraordinary procedural irregularities, predetermined outcomes without factual or legal support, and disregard for Title VI’s legal requirements and the 1st Amendment to meet the political mission of the Task Force"
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A simple guide to public engagement for professors:
Good engagement > No engagement > AI slop
If you are too busy to do it well, just don't do it! That's fine! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
<Jon Stewart voice> least ethical field experiment *ever*
Brett LoGiurato
@BrettLoGiurato
So this is pretty wild: A polling firm that came out of nowhere to show Francesca Hong with a 23-point lead over David Crowley in the Wisconsin primary has shut down and said its whole operation was bogus
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
To be clear, that was a joke (I hope). In addition being unethical, it would be a crummy experiment. This explanation seems more plausible, though obviously we don't know
Alex Tolkin
@AlexTolkin
I doubt this was any sort of study; if it was it seems both wildly unethical and poorly designed. My guess is it was an attempt to manipulate prediction markets such as Kalshi.
Trying to game odds on prediction markets might be a new motivator for disinformation campaigns. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Shayne Longpre
@ShayneRedford
1/ How are people really using AI?
Today, researchers across @MIT, @Stanford, + 12 academic institutions are launching the Public AI Observatory: public infrastructure for measuring how people actually use AI assistants in the wild.
🔗 www.ai-observatory.org
📜 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Helen Toner
@hlntnr
Really good to see this. IMO this ⬇️ is by far the best way to think about "pacing the frontier"—not as some fixed amount of time (e.g. "six month pause" or "go 10% slower"), but simply making sure that enough time is taken to meet a reasonable safety/assurance bar.
If all parties do that—whether by choice or because they have to—then there you have it, the frontier has been paced.
Max Zeff
@ZeffMax
New: OpenAI says it introduced new security, alignment, and monitoring requirements that, until met, will hold up development of its next frontier AI model, Astra.
OpenAI's VP of research and safety, Mia Glaese, effectively said this process will take as long as it needs to. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Senior PowerPoint Engineer
@ryxcommar
Absolutely brutal deposition transcript from someone who had AI write their expert testimony for them. www.404media.co/show-how-3m-is-0-at-fault-expert-witness-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
Recommended - great summary of the "second demographic tradition" and how striking it is esp in middle income countries. Their explanation: modernity. Kids are expensive (Becker education, housing) and less useful (modern pensions as insurance, home appliances, fewer farms) 1/3
Jesús Fernández-Villaverde
@JesusFerna7026
Last week I posted a new paper with Patrick Norrick: “Terra Incognita: The Economics of a Shrinking World.”
We chose the title deliberately. No society in history has experienced the fertility levels now seen in South Korea, China, Thailand, Colombia, Chile, and many other countries. Our knowledge of the causes (and of the economic consequences) remains far more limited than most public discussion acknowledges. Much of what we write is, at best, educated conjecture.
The paper also struggles against a space limit. We wrote 20,000 words, far from the 250,000 or so we would need to address some issues in more detail (if I had the time and resources to hide away for a year, I could do that, but not now). That means some ideas are only sketched.
Nonetheless, we emphasize several important points.
First, fertility has fallen very fast everywhere: rich and poor countries, east and west, north and south, conservative and liberal societies, religious and non-religious societies (with the exception of the Jewish population of Israel; fertility has also collapsed among the Muslim population within the pre-1967 borders), you name it. Even in sub-Saharan Africa we see fast and unprecedented drops in fertility (alas, from a high initial level).
Second, and this is really interesting, the fertility collapse has been concentrated in poor and lower-middle-income countries much more than in rich countries. By now, income per capita and fertility are positively correlated within OECD countries. We conjecture this will hold globally in a few decades.
Third, and related to the second point, the fertility collapse has been concentrated among poor and lower-middle-income women within countries. In countries such as the U.S., the rich and highly educated now have more children than the poor and less educated.
Fourth, we document why we do not understand the data from the U.N. World Population Prospects. See, for example, Tables A.1 and A.2.
Fifth, we explain why some proposed mechanisms struggle when confronted with the data.
A more subtle point is at work here. Many commentators do not seem to understand the difference between proximate causes and ultimate causes. Yes, births might have gone down because fewer women are in long-run relationships. That is the proximate cause. But you need to explain why fewer women are in long-run relationships (the ultimate cause), and saying that they spend more years in school, to take one example, does not get us very far. Why do they spend more years in school? Once you start down the whole chain of reasoning, things become much harder than they seem.
The paper can be found here:
www.sas.upenn.edu/~jesusfv/Demographics.pdf
Comments always welcome! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
We're hiring!
Assistant Professor of Comparative Politics
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
It's notable that 25 years later there still isn't consensus on what we can learn from the settler mortality paper.
Does the result depend on incorrect mortality rates? Did @albouy selectively drop half the countries?
A real science would care about getting to the bottom of these things. Or maybe these papers just exist for status competition and nobody really cares?
Someone should put a big team of AIs on this and write a conclusive paper on whether there is anything robust here. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
This is smart for both reproducibility and as a way of using AI for science: “AI-generated analyses should be accompanied by multiverse-style reporting and full disclosure of the prompts used, on par with code and data.”
And it applies to all AI analysis, not just academic work
John Horton
@johnjhorton
this is an interesting paper arxiv.org/pdf/2602.18710
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
This is not a very good article. Innuendo, hearsay, and the substantive critiques are weak sauce.
What it does get right is this—there is a serious reluctance among economists to publicly criticize the giants of the profession, even when they are way out over their skis.
The Economist
@TheEconomist
Daron Acemoglu’s ideas can seem underwhelming for a titan of social science. Register for free to learn why economists are questioning the Nobel prizewinner’s reputation www.economist.com/finance-and-economics/2026/08/17/the-wo…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Michael Linden
@MichaelSLinden
I know I'm old now because I really wish we had a politics where the White House Communications Director would be immediately fired for tweeting something like this.
Steven Cheung
@StevenCheung47
Jon Jackoff has to be the biggest cuck loser in politics. Instead of denigrating hard working people serving their country, Jon should look deep into his soul and ask himself why he's a miserable person who hates this country. It's because he's a radical, extremist Dumocrat.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tim Miller
@Timodc
A private citizen with no security clearance and a massive financial stake in the region tells an American news outlet that he "can't go into the details" of our negotiations with Iran. Banana Republic stuff.
Mario Nawfal
@MarioNawfal
🇺🇸🇮🇷 Jared Kushner on Iran says they’re engaging in a very strong way, and that’s positive, but we’re not there yet in terms of understanding.
He points out that Trump will be patient and doesn’t want to rush a deal.
This comes after the secret Barzani backchannel showed the IRGC itself saying it prefers negotiations, while Iran has publicly warned it will escalate in the Strait of Hormuz if the U.S. fails to implement the interim deal within weeks.
The talks are moving, but the real terms still look far apart.
Writer: Lucas -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
o3-mini (very obsolete) in an agentic loop made good exam questions: "In one of the largest field studies of AI-generated questions to date, we found that this approach achieves psychometric properties on par with those of questions appearing on high-stakes standardized tests."
Misha Teplitskiy | Science of Science
@MishaTeplitskiy
AI-generated exams are fine
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Egger
@EggerDC
Trump has been getting graded on a curve for a while but this is a whole new level (the "win" here is that the tariffs didn't totally melt the global economy, not that they strengthened it or achieved any of their policy aims)
www.politico.com/news/magazine/2026/08/14/trump-tariffs-m… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Disturbing @adambonica analysis of the Democratic scam PAC ecosystem and how the tactics they employ are also used by mainstream party orgs and elected officials to take advantage of vulnerable seniors
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
James Surowiecki
@JamesSurowiecki
Unless they beat the shit out of federal law enforcement officers while trying to storm the Capitol. Then Todd Blanche's boss will give them a blanket pardon.
Acyn
@Acyn
Blanche: In my Department of Justice, if anybody touches a federal law enforcement officer in any way inappropriate, we will prosecute them to the fullest extent of the law
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jake M. Grumbach
@JakeMGrumbach
I found @Vinncent's piece compelling. I have a new working paper that builds a model of the argument about common ownership:
A standalone media owner maximizes profit. A common owner like Musk or Ellison can run the media firm at a loss to increase the $ of his total portfolio
Vincent Bevins
@Vinncent
Elon Musk obviously succeeded. The 'vibe shift' was a leveraged buyout --- North South Notes:
www.northsouthnotes.org/p/the-vibe-shift-was-a-leveraged-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Kids: Study "general equilibrium"
Bobby Fijan
@bobbyfijan
This makes further cities much more plausible commutes
Healdsburg to San Francisco *Door to Door* in 90 minutes in a Waymo ... really nice. Especially if you are only in the office 3 days a week
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Put another way
Armand Domalewski
@ArmandDoma
if everyone who takes public transit took Waymo instead we'd have traffic that makes Mumbai roads look empty. i don't really get why basic geometry is so hard for so many people to grasp
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Sam Haselby
@samhaselby
Let's "give Arday a courtesy in death that he was often denied in life: Let’s treat him like a person.... a flawed human being who should still be alive but is not." @Tyler_A_Harper www.theatlantic.com/culture/2026/08/jason-arday-cambridge…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Aaron Reichlin-Melnick
@ReichlinMelnick
Yes. In many dictatorships and authoritarian regimes, daily life can remain normal, if not improved, for the parts of society that align itself with power.
Sure, your neighbor might disappear in the middle of the night, but your material conditions went up for a while.
Matthew Yglesias
@mattyglesias
“The collapse of democracy has less direct impact on my personal life than HR training norms” is actually a point people should take seriously.
There’s a reason why expensive gasoline has done more damage to Trump than pardoning everyone for 1/6. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Marc Porter Magee 🎓
@marcportermagee
You see a lot of people who work at major universities (including deans and presidents) posting obviously AI written stuff on here. Isn’t that bad? Don’t they see the example they are setting for their students?
I don’t get it.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
impeachable conduct every day
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Luke Rudkowski
@Lukewearechange
I mean, the Trump administration is talking about doing alien disclosures but this is too much
Peter Hamby
@PeterHamby
Just act natural. It’s only a TikTok.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Fascinating how AI is helping humans (re)discover facts about ourselves - first about the nature of intelligence (more complicated than we thought!) and now about social organization and coordination (even more so!)
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Why is there no housing in CA?
Builder: "Delays with the subdivision and permitting have been brutal + the fees will be ~$60k/unit (!!) This same project in Houston would be about $7k/unit."
Donovan
@DonovanBuilds
The first SB684 project in San Diego that will deliver new homes <$700k
After lifting 8 modules (4 units) over the existing house that we remodeled we have now sold off that home to a veteran
Delays with the subdivision and permitting have been brutal + the fees will be ~$60k/unit (!!)
This same project in Houston would be about $7k/unit
We decided to finish with stucco onsite and that should be completed in August before we list -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Colin McCarthy
@US_Stormwatch
Temperatures hit as high as 43°C (109°F) in France today as the fifth major heatwave of the year has arrived in Europe. Hundreds of cities and towns across the continent experienced their hottest August temperatures ever recorded.
This is getting ludicrous. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Trip Gabriel
@tripgabriel
Sen Bill Cassidy, R-LA, who provided the tie-breaking vote to confirm Blanche, said he believed Blanche would have more leverage to assert independence from Trump if he was confirmed
Aaron Rupar
@atrupar
Newly confirmed Attorney General Todd Blanche is speaking at a Trump rally to endorse Bruce Blakeman
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A lot like people who say they oppose GMOs while eating them every day
Chad Reese
@ChadMReese
This is funny because of the community note, but there's a really important truth here: Many in Northern VA have NO IDEA they have a ton of data centers basically on their doorstep.
Why? Because most of the propaganda against them is just made up nonsense. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Anna Stansbury
@annastansbury
These results are worth pausing over. An extra $12k/year to low income households for three years (40% avg increase in income) - mostly toward consumption & a bit toward leisure.
Few big long term effects: on saving, debts, education, job quality, life satisfaction, health…
AlexBartik
@AlexBartik
How do households change their consumption and household balance sheets in response to large, but time-limited, unconditional cash transfers? We study an experiment that provided $1000 a month for 3 years to 1000 young, low-income households. 1/
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Interesting @JeremiahDJohns piece on the dogmatism of the "Believe Science" crew who publicly proclaim their fidelity to science while not practicing what they preach, including falling victim to the anti-AI moral panic
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matthew Yglesias
@mattyglesias
“The collapse of democracy has less direct impact on my personal life than HR training norms” is actually a point people should take seriously.
There’s a reason why expensive gasoline has done more damage to Trump than pardoning everyone for 1/6. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jeremiah Johnson 🌐
@JeremiahDJohns
Data center bans/moratoriums strike me as fundamentally illegitimate.
It's perfectly fine to pass laws that regulate specific harms. Regulate pollution, noise, etc. Make sure the local infrastructure is in place, sure.
But to pass a blanket ban on an entire type of business isn't regulating harms, it's just marking that business as ontologically evil. What they're doing is so beyond the pale it's NEVER acceptable. Which is ridiculous for a business as mundane as data centers.
A bunch of people have watched slop videos and now believe that data centers are, in essence, demonic. It's absurd. Again, regulating specific harms is fine. But if a farmer wants to build data centers on the land they own, local busybodies shouldn't be able to just make that illegal. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jonah Goldberg
@JonahDispatch
The FBI Director (and the Attorney General) doing a dear leader suck up to the President at a political rally with applauding uniformed cops in the background is not how it’s supposed to work.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Roger Parloff
@rparloff
Context: On 9/24/25 Trump was considering firing both AG Bondi & DAG Blanche for not indicting his enemies. Called them into meeting. Had "running list of people he wanted charged [including] Judge Juan Merchan & his daughter." Per @maggieNYT & @jonathanvswan
Roger Parloff
@rparloff
US Atty Boutros in Chicago, whose office brought the Broadview 6 debacle & opened inquiry into E Jean Carroll trial funding, has been investigating Loren Merchan, daughter of NY judge who presided over Trump's criminal convictions.
www.nytimes.com/2026/08/15/us/politics/prosecutors-examin… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Insane:
"Navy is considering a major redesign of its new aircraft carriers...in response to Trump’s concerns about the 'look'"
"preference...is for the new aircraft carriers to look more like those...during World War II"
"moving the island could cost billions of $"
The Washington Post
@washingtonpost
Exclusive: The U.S. Navy is considering a major redesign of its new aircraft carriers to better match President Trump’s preferred aesthetic for the warships. www.washingtonpost.com/national-security/2026/08/15/navy-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jeff Stein
@jstein_sun
Just so people are really aware of what this means: A Trump appointee has approved Trump’s personal crypto firm to operate as a bank
Corporations can conduct transactions in which the Trump family gets a cut
The public will not have to be informed
There’s no precedent for this -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
The Wall Street Journal
@WSJ
Exclusive: World Liberty Financial, the Trump family’s flagship crypto venture, says its trust company has received preliminary approval from the Office of the Comptroller of the Currency to become a bank. www.wsj.com/finance/trump-familys-world-liberty-financial…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Unprecedented corruption
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jason Locasale
@LocasaleLab
This is a terrible tragedy. My condolences to Arday’s family and everyone who cared about him.
Arday was used by the university system. Universities elevated him into a public symbol, promoted an extraordinary personal narrative, and used his appointment to advertise their own virtue. They built him into something larger than any person should have been expected to carry, apparently without performing the basic diligence they owed both him and the public.
When the story collapsed under scrutiny, the universities that had created and benefited from it left one human being to absorb the consequences in public. The toll of living inside that mythology and watching it unravel must have been overwhelming.
Whatever the conclusions about his CV, Arday is still a human being, not a mascot, a diversity credential, or a disposable marketing asset. The universities that promoted and used him cannot now distance themselves from the human consequences of their decisions.
The university system helped construct the myth, benefited from it, failed to scrutinize it, and then allowed the full burden of its collapse to fall on the person at its center. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Mehdi Hasan
@mehdirhasan
His boss pardoned 100s, literally 100s, of people who assaulted over 140 police officers on January 6th, broke their bones, gave them strokes and heart attacks.
The level of gaslighting here is off the charts. Truly shameless people.
Acyn
@Acyn
Blanche: In my Department of Justice, if anybody touches a federal law enforcement officer in any way inappropriate, we will prosecute them to the fullest extent of the law
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
mad king watch
Aaron Rupar
@atrupar
Trump says that after Iran is defeated he’s going to declare the Strait of Hormuz a territory of the United States
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
open racism
Acyn
@Acyn
Trump: Chuck Schumer, who's a Palestinian. He's a full-fledged Palestine . In fact, I'm looking for Palestinian garb. I'm going to send it to him for his birthday. I know him very well.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
standing in front of white police in uniform
Acyn
@Acyn
Trump: I don't see the people behind me being big on DEI. Were you big on DEI anybody? That's not a DEI group. I can tell. It never was, by the way.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Marco Foster
@MarcoFoster_
Shane Gillis: “[Trump’s] selling faster access to his tweets and then you can bet on the markets. It’s literally insider trading”
Joe Rogan: “Wait a minute, is that real? How is that legal? Who’s telling him this is a good idea? $100K to get early tweets? That’s literally crazy. He’s just on a grab”This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Senator Scott Wiener
@Scott_Wiener
Car break-ins are way down in San Francisco. One reason is we changed state law by removing a massive loophole that required prosecutors to prove all car doors were locked, even if evidence of a break-in was clear. This loophole was a significant barrier, and we removed it.
The San Francisco Standard
@sfstandard
RIP Bip City.
Car break-ins, long a symbol of dysfunction in San Francisco, have plummeted in recent years — from 22,700 in 2022 to 1,764 so far this year, thanks in part to surveillance technology.
Read the full story: sf.news/4gfGcot -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Eva Vivalt
@evavivalt
So great to see this out! Over ten years in the making.
QJE
@QJEHarvard
Recently accepted by #qje: “The Employment Effects of a Guaranteed Income: Experimental Evidence from Two U.S. States,” by Vivalt, Rhodes, Bartik, Broockman, Krause, and Miller: doi.org/10.1093/qje/qjag042
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
When things are so awkward and poorly presented you know it's not AI
Peter Hamby
@PeterHamby
Just act natural. It’s only a TikTok.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Robb Willer
@RobbWiller
OMG, this is so well put. And this illustrates how dropping standardized tests makes it hard for talented working class kids to get into elite schools. The upper middle and upper class folks who have the most direct and indirect influence over elite college admissions are uniquely blind to this.
Michael Yeh
@michaelyehmd
“It is easier for a working-class kid to pick up an SAT prep book and study it intensively than it is for them to go on a community-service trip to South America, play travel lacrosse, or flatter the sensibilities of an upper-middle-class admissions official.”
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Imas
@alexolegimas
If you look at the last 3 years of AI progress, technologists have largely been right about the (massive) increase in capabilities, and economists have largely been right about the economic impact. I don’t expect that to change.
Ricardian comparative advantage ftw! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ruxandra Teslo 🧬
@RuxandraTeslo
A few weeks ago, Novo Nordisk announced that its long-awaited ZEUS trial had failed. The study tested whether blocking IL-6, a key driver of inflammation, could reduce cardiovascular events -- a test of the hypothesis that inflammation is a treatable cause of heart disease.
This was not a small undertaking, as the trial enrolled 6300 people over several years and must have cost in the hundreds of millions (I would estimate between $300M and $600M).
@MariosGeorgakis wrote a great post-mortem trying to answer the question: why did the trial fail?
I think there are two things that make his post great:
1. First, it goes through the evidence we had prior to this trial and shows just how hard it is to predict results in human trials for novel targets (and why companies often choose to avoid them). Based on the available evidence, the trial should have worked.
The most important points are the data from the CANTOS trial (targeting IL-1), which strongly suggests that inflammation is promising as a treatable cause of cardiovascular disease. We also have convincing genetics data showing that IL-6 is causally implicated in cardiovascular disease. So far, genetic evidence remains the biggest booster of clinical trial success for a new target.
2. Second, we need to invest in better biomarkers, which could provide early biological signals of whether a drug is working. Today, we are living through what I consider a frankly unsustainable situation: a developer can spend half a billion dollars on a late-stage trial only for it to fail.
In this case, the drug reduced inflammation as measured by C-reactive protein (CRP) in early trials, but didn't prevent heart attacks in later ones. CRP is a very poor biomarker as it is not predictive of clinical efficacy, but it's all we have. Having better early indicators of success would catch bad candidates years sooner and at a fraction of the cost, freeing that capital for therapies with a real chance of working.
Unfortunately, as I explained in an article for @IFP last year, the incentives around developing clinically validated and useful biomarkers are skewed, since they function essentially as public goods: there is little possibility for any individual developer to make money from developing a clinically useful biomarker and making it available, even if doing so would benefit the field immensely.
We still do not know why the trial failed. One thing I am worried about is that investors will see this as an invalidation of the inflammation hypothesis in cardiovascular disease and it will deter investment -- I think there are signs this is happening. This could just be a bad draw from the distribution of possible outcomes! But it is entirely rational that developers would avoid spending hundreds of millions of dollars to find out that their drug dailed. Again, my only prescription here is to get better biomarkers that indicate success earlier on.
This would allow us to test many drugs targeting inflammation in earlier stage trials and progress only the actually promising ones, if such promising drugs exist at all. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Roger Parloff
@rparloff
At SCOTUS, govt casts Trump's ballroom as "integrated military complex ... vitally req'd by national security." Belittles DC Cir ruling as holding that Trump is a mere tenant (true) while the district judge is the landlord (false; Congress is).
www.supremecourt.gov/DocketPDF/26/26A203/419029/202608132… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ben Krauss
@BenKrauss5
I find this all to be incredibly lame on both sides. Classic free speech for me and not for thee. Obviously Piker, who like it or not is an influential political figure, should be allowed to speak. But same goes for Shapiro or any other comparable figure on the right.
Don't be illiberal. It's anti-American. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
回
@3_under_scores_
New academic study: Inclusionary zoning costs renters $800k per "affordable" unit created. This is more than the per unit cost for the Low Income Housing Tax Credit program. While LIHTC which is largely funded by the rich via income taxes, IZOs are regressive & funded by renters.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kate Brannen
@K8brannen
This story should be a bigger deal and should send a shiver down Americans' spines: www.nytimes.com/2026/08/13/us/politics/dhs-inquiry-immigr…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Senator Mark Kelly
@SenMarkKelly
I’ve launched off the front of aircraft carriers hundreds of times, have a Master’s degree in aeronautical engineering, and am a test pilot and even I wouldn’t suggest to the Navy how to engineer specific systems on its ships. Donald Trump should stick to what he knows best — ballrooms, bankruptcy, and bullshit.
Dave Brown
@dave_brown24
SCOOP: Trump has directed the Navy to tear out a complex system for launching fighter jets from its aircraft carriers and instead go back to using steam catapults, a move Navy officials have fought for years and will likely cost billions in extra dollars.
U.S. officials say Trump signed a national security memorandum today directing the change, which would involve removing the electromagnetic catapult system used by the ships in the Gerald R. Ford class of carriers
www.wsj.com/us-news/trump-orders-navy-to-restore-old-tech…
@MarcusReports -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Stephen Richer
@stephen_richer
The amount of self-dealing is breathtaking.
Marco Foster
@MarcoFoster_
Joe Rogan: “This government has done some shit that no government has ever done before and one of them is the crypto stuff. The crypto stuff is nuts. Trump has made billions in crypto. Then you got his sons and his sons are involved in these weird business deals and there’s a lot of money flowing around that you could dig in and start arresting people”
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
S.V. Dáte
@svdate
Omg. This is beyond dumb.
There was a reason the Navy wanted to get away from jerk forces on fighter planes at takeoff.
And to retrofit the Ford class carriers?
You’ll have to tear them down to their keels.
Dave Brown
@dave_brown24
SCOOP: Trump has directed the Navy to tear out a complex system for launching fighter jets from its aircraft carriers and instead go back to using steam catapults, a move Navy officials have fought for years and will likely cost billions in extra dollars.
U.S. officials say Trump signed a national security memorandum today directing the change, which would involve removing the electromagnetic catapult system used by the ships in the Gerald R. Ford class of carriers
www.wsj.com/us-news/trump-orders-navy-to-restore-old-tech…
@MarcusReports -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Two-bit autocrat stuff
Paul Farhi
@farhip
The obsession with having his name on stuff seems bottomless. Since 1/20/25 alone, it has gone on:
—KenCenter
—U.S. Inst. of Peace
—Palm Beach Airport
—Battleships
—Pharma portal
—Savings accounts
—Passports
—Immigration visas
—Natl Park passes
—Currency and coins
Obsessive. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
brent 🌹🗽
@BrentInQueens
You’re not going to trick us into believing you’re a normal American couple again, nice try
Variety
@Variety
Keri Russell calls herself and partner Matthew Rhys “not that famous”:
"People don’t really care about us — in a great way. Some people in our neighborhood are very famous, and their life is very different. And it seems so hard. I love the level of our life,” she says. “I love having a job. I like occasionally getting into a restaurant when I can’t. I sometimes can’t get a dress, because I am not, you know, someone famous. But that’s the reality of our station. And I think it’s perfect."
variety.com/2026/awards/features/matthew-rhys-keri-russel… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Caleb Brock
@caleb1brock
We've got to kneecap America's ability to compete on technology worldwide so this guy can dump water on his field
More Perfect Union
@MorePerfectUS
NEW: Big Tech claims AI data centers don’t use much water.
But data centers typically report the water used inside the facilities — we found them indirectly consuming 10x that amount.
We went to the drought-stricken West, where farmers now have to fight AI oligarchs for water.This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Billy Binion
@billybinion
I'm sorry, I can't get over this. The most powerful person on earth is using the full force of government to jail a 67-year-old over an error his *own administration* oversaw. It is the epitome of government "weaponization." Mind-blowing hypocrisy. It's unconscionable.
Kyle Griffin
@kylegriffin1
BREAKING WSJ:
Trump White House officials have asked the Trump Justice Department to consider a new prosecution of David Hearn, the man accused of vandalizing the Reflecting Pool — as Trump fumes over U.S. Attorney Jeanine Pirro's decision to drop the case. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Yifan Qian
@qian_yifan
How is generative AI shaping the public funding landscape?
Check out our latest paper in PNAS: The Rise of Large Language Models and the Direction and Impact of US Federal Research Funding: www.pnas.org/doi/10.1073/pnas.2601439123
1/n -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jay Nordlinger
@jaynordlinger
May I ask you a question, my friends -- my Republican friends, in particular? What kind of man inscribes his name on another man's memorial? Really, what manner of man?
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jonathan Libgober (same handle on other places)
@JlibDoesEcon
wild, awesome paper
Econometrica
@ecmaEditors
What happens to insurance markets as genetic prediction improves? This article combines quantitative genetics, economic theory, and big data to study adverse selection and inform policy. Selection may become severe in critical illness insurance.
www.econometricsociety.org/publications/econometrica/fort… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Mad king watch. Going back to *steam*. In 2026.
Dave Brown
@dave_brown24
SCOOP: Trump has directed the Navy to tear out a complex system for launching fighter jets from its aircraft carriers and instead go back to using steam catapults, a move Navy officials have fought for years and will likely cost billions in extra dollars.
U.S. officials say Trump signed a national security memorandum today directing the change, which would involve removing the electromagnetic catapult system used by the ships in the Gerald R. Ford class of carriers
www.wsj.com/us-news/trump-orders-navy-to-restore-old-tech…
@MarcusReports -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"The Trump administration abruptly declined to renew a contract for a cybersecurity firm it hired to examine voting equipment in Puerto Rico after the company didn’t echo administration officials’ false claims of voter fraud"
Sean Lyngaas
@snlyngaas
2/3 Read additional details from the declaration the firm gave the Office of Director of National Intelligence to try to save their contract with the government after Trump allies allegedly falsely accused the firm of being funded by George Soros: edition.cnn.com/2026/08/13/politics/voting-machines-2020-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
nic carter
@nic_carter
It’s really interesting that there’s basically a bipartisan consensus based on complete and trivially debunkable falsities, namely that
- datacenters consume meaningful amounts of water (they don’t)
- datacenters are responsible for power prices increases over the last decade (they aren’t - inflation and decarbonisation in blue states is)
- datacenters impose costs on local communities (they generally enrich them and make life better)
- datacenters are being built “for flock mass surveillance” (flock accounts for, generously, 0.01% of the DC buildout)
- datacenters create noise and other pollution (they don’t)
- datacenters are being built in residential areas (they aren’t)
- datacenters get tax abatements that entirely eliminate their tax burden (never the case- abatements typically only cover a small fraction of taxes, and DC pay a half dozen different type of taxes that go to the state and city, they are always net and meaningful taxpayers)
Virtually everything your conspiracy minded friend believes about datacenters is false. It’s also completely trivial to debunk using a simple google search or AI query but these people just blindly trust a random tiktoker sitting in their car instead.
Are there real objections? Sure. But the opponents of datacenters aren’t making them. They rely on outright falsehoods like DCs are poisoning the water, or creating droughts, or tripling power prices locally, or paying zero taxes.
Rare to see a popular consensus on both sides of the aisle that is basically a complete inversion the actual fact pattern. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
M. Nolan Gray 🥑
@mnolangray
With prices and rents rising this rapidly, every day that San Francisco isn't in the midst of a building boom is a complete and total policy failure.
The Wall Street Journal
@WSJ
The AI boom has pushed the bruising race for housing in San Francisco into overdrive, with a lack of supply colliding with sky-high salaries. www.wsj.com/real-estate/the-bruising-race-to-rent-in-san-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Meryl Kornfield
@MerylKornfield
Real impact journalism:
Drew Harwell
@drewharwell
In our story on the cops who stalk women with license-plate readers, we said Flock could make two simple changes to reduce abuse: Require case numbers for every search and proactively scan for suspicious behavior. Now, the company is adopting both www.washingtonpost.com/technology/2026/08/13/flock-unveil…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ben Pauker
@benpauker
🚨EXCLUSIVE: @washingtonpost reports that CIA is involved in 3 mysterious lethal attacks on boats near the Galapagos.
The strikes were part of a “covert action” program, a type of black action approved by the president but not publicly acknowledged.
www.washingtonpost.com/investigations/2026/08/13/covert-c… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Michael Yeh
@michaelyehmd
“It is easier for a working-class kid to pick up an SAT prep book and study it intensively than it is for them to go on a community-service trip to South America, play travel lacrosse, or flatter the sensibilities of an upper-middle-class admissions official.”
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The detection method is incredible
Sam Kessler
@skesslr
Last summer, we got our hands on screen recordings showing North Koreans interviewing for jobs under stolen identities.
What made these videos interesting: We could see the them reading verbatim from ChatGPT. 🧵
1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Masley
@AndyMasley
This video was one of the very worst MPU has done. It’s entirely focused on farmers in the desert feeling entitled to take as much water from the Colorado River as they’d like to farm in the very most arid place in the country with no competition at all with any industry that also uses water.
Their first example is a pecan farmer who demonstrates how he dumps huge amounts of water into the ground and then implies that anyone who’s okay with data centers using water must “know how to eat computer chips” because “there won’t be any food left.” This guy thinks if he has to pay a little more for water to farm in a desert, it’s like saying we shouldn’t have food at all. The video frames him as a hero.
They also bring up the offsite water cost of generating electricity as if it’s this secret data centers are hiding by not reporting it in their water use. For any other building that uses water, we usually only expect them to report their internal power and water cost and infer the water cost of the electricity from the power plants. This is all out in the open.
They interview a county water manager who says the data center really isn’t an issue and only uses 0.4% of the local water, and then immediately cut to a local activist saying they can’t have enough for the data center because water costs have gone up. These are obviously unrelated, because as the video notes the city’s raised water rates every year for 11 years.
I could go on and on, every moment is insane.
More Perfect Union
@MorePerfectUS
NEW: Big Tech claims AI data centers don’t use much water.
But data centers typically report the water used inside the facilities — we found them indirectly consuming 10x that amount.
We went to the drought-stricken West, where farmers now have to fight AI oligarchs for water.This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
John V. Kane
@UptonOrwell
🚨 Super excited to share this new WP w/ @JoohyeJeong @AdamBerinsky, Soong & Parker.
We ask a simple but extremely practical Q for survey researchers: Does it matter where we place items/experiments in our survey?
Answer: Not really, no. 👍 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Cool idea from someone I know - retro-style boxscores for the baseball fan who misses the old newspaper sports pages of our youth.
Updates every day; share with the baseball fan in your life! (Also, they would love feedback - please reply with thoughts...)
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
URL to follow: retroboxscore.com
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Cameron 🇺🇸 🗽🦅
@CameronCorduroy
crazy story
Elon Musk and DOGE destroyed decades of USAID data, but luckily it was recoverable because a Canadian high schooler coincidentally downloaded the archive for a school project -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Virtual Isaiah Andrews and Jesse Shapiro advising skills
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Normal stuff for a democratic leader
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Always read @davidroodman evidence reviews
David Roodman
@davidroodman
Anthropic just released a report I coauthored for them (as a contractor), which reviews the evidence on the effectiveness of job training programs. www.anthropic.com/research/reviewing-the-evidence-on-work…
🧵1/N -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Caleb Watney
@calebwatney
You can tell the @nypost is a Very Serious Outlet because they published this insane hit piece on @deanwball with my face in the print edition
New York Post
@nypost
OpenAI risks White House relationship with hiring of 'nuisance' AI policy executive: sources nypost.com/2026/08/10/business/openai-risks-white-house-r…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Though they've been doing this at MIT for decades, so this isn't just 2016 coming back for revenge.
Maybe this time they'll actually collect data to see how students do in 2nd year courses.
(Heh I bet they all get A's)
Pim Welle
@pim_well
MIT has been pass/no record for the first semester since the 1970s. Seems valuable to acculturate folks who have had less prep
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tenobrus
@tenobrus
making the opposite bet. i expect it will work quite well, be widely adopted by other labs within the year, and quickly be viewed as a critical social good to the point people find it insane companies ever served models without it. in the same way backlash against high-quality imagegen has been significantly ameliorated by synthid making detection of low-effort usage trivial. ai usage is actually *much better* for both the creators and consumers of it if the whole layer of confusion and potentially adversarial behavior around providence and "potential plagiarism" is bypassed w technical innovation.
Joshua Gans
@joshgans
I'm going to call it now. I think that the Claude Watermarking is going to be both useless (too many false positives and negatives) and then completely forgotten and ignored.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tyler Austin Harper
@Tyler_A_Harper
Responses:
1. Sure, but critics of the left critics of higher ed have made the same argument (“Not now!” “Don’t say it out loud conservatives will like it!”) for years. “Timing” and “audience” have become an excuse to perennially delay a conversation that is desperately needed.
Namwali
@namwalien
I'll bite, by way of some questions: 1. As an erstwhile literary critic, do you not believe that the timing, audience, and aims of rhetorical persuasion matter to an argument? 2. What do you propose as solutions to the problems you're diagnosing, beyond faculty just... quitting?
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
For some students (X), this will reduce stress and help them adapt to college expectations.
For some students (Z), this is a signal that first year is 100% parties and drinking.
Is there evidence on the share of type X vs type Z students? Are we collecting evidence? -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Whatever is in writing, the de facto policy is “Don’t ask, Don’t tell” for authors and reviewers both.
alz
@alz_zyd_
To be blunt, this policy seems idiotic. Firstly, the stated premise of "manuscripts under review are confidential" is utterly ridiculous in fields where working papers are public. Secondly, "don't upload to an AI" is rather obviously unenforceable in any case
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
In theory this can work: "I gave this a lot of thought, jotted down my conclusions, let an AI clean them up."
In practice, 99.9% of the people who say they this did not in fact give it a lot of thought. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Counterpoint, reading books with an AI as guide is awesome
Jen Rice
@jen_rice_
This @axios piece is so so so sooo humiliating 😆 we are witnessing a truly catastrophic level of societal brain rot right now
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Open Source Intel
@Osint613
Spanish Olympic skateboarder Danny León timed a mid-air trick to a solar eclipse, calling it "the move of his life."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Build more housing!
Chris Bakke
@ChrisJBakke
Imagine finally selling your company for $500M
You text your agent that you're ready to drop $20M on a house
As you're daydreaming about the insane mansion you're going to buy, she sends you this: -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Roitman
@aroit
People like showing off that Pangram works.
I think that's because 1) it's pretty magical to finally see a something in a product category that is often described as "snake oil" actually work.
2) they are genuinely relieved that something exists to classify what humans write vs what AI write. They see social benefit and want to share it.
3) they know that everyone has this idea that "AI detectors don't work" and it's like they found out that everyone is wrong and Pangram works. They want to share it.
Update your priors.
Gon Huertas
@GonHuertas
If you're confused about what kind of writing gets flagged as "AI generated" by Pangram, this example may help.
Below is some human written text (to which I have deliberately added some typos). If you put this through Pangram, it'll be labeled 100% human.
"By and large, monetary authorities rely on a ssingle policy instrument: the reference nominal interest rate. The use of one instrument makes it imposible to target both inflation and output simultaneously, and creates an inherent tradeoff. The central bank's position on this tradeoff --how forcefully it is willing to pursue an inflation mandate in spite of the implications for economic activity-- defines the character of its monetary policy approach. We refer to central banks that react strongly to inflation, and that place a relatively small weight on output considerations, as being ``hawkish''.
How hawkish, then, are Latin American central banks? This section answers the question by estimating the monetary reaction functions, or ``policy rules'', of five of the region's largest economies for the 2003-2024 period. We find that they are hawkish indeed, responding to a ris in inflation with an aggressive increase in real rates, and prioritizing price stability to such an extent that output plays almost no role in their decision making.
This finding is consistent with explanations provided by central bankers in the literature, where it has been argued that central banks adopted a hawkish stance on inflation as a way to build monetary credibility, against the backdrop of the region's historical struggles with very high inflation.
Details on the sources of the data used in the analysis are provided in the appendix. We rely on monthly and quarterly time series from each country's central bank and national statistics office."
(cc @paulnovosad, @itaisher) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Cowen's points are worth considering but "isolated demands for rigor" is on point
Rob Wiblin
@robertwiblin
The Marginal Revolution comments section has lost patience with Tyler Cowen's commentary on AI to a degree that's actually laugh out loud hilarious.
Here are the top comments on : "Act like it is science". 1/
marginalrevolution.com/marginalrevolution/2026/08/act-lik… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"a broader trend within the second Trump administration: no-bid contracts awarded to people with connections to Mr. Trump or his cabinet secretaries."
Charles Ornstein
@charlesornstein
Wow! Homeland Security Paid $464 Million for Airplanes. Then It Parked Them. www.nytimes.com/2026/08/13/us/politics/homeland-security-… via @NYTimes
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
All day Astronomy
@forallcurious
This what a Total Solar Eclipse seen from an airplane looks like .
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Arvind Narayanan
@random_walker
Two contrasting patterns of working with agents are emerging: delegation and collaboration. Delegation makes sense when it's a long-horizon task that you want the agent to tackle asynchronously, your intent and specifications are clear to the agent, and it's easy to verify the output at the end even if you didn't stay in the loop. Effectively delegatable tasks are rarer than the hype would suggest, because it’s limited by what you can cheaply verify, not what the model can do.
Collaboration makes sense when the task is hard to fully specify a priori and you want to be in the loop to iteratively figure out what you want, stay in control, recover from mistakes, sharpen your own skills through collaborative task performance, and have fun.
The design criteria for automation/delegation agents and collaboration agents are very different. If you're going to delegate a big task, accuracy and reliability are what matter the most. You want a frontier system that will do the best possible job. It doesn’t matter if the task will take minutes or hours. If you're going to collaborate with an agent, the criteria are more multifaceted: latency (even at the expense of accuracy), transparency/controllability, creativity, and more. The agent should allow the user to stay in a “flow state” instead of having to delegate a task and come back later. It should promote the user’s agency and be fun to work with.
We're at the very early stages of an emerging bifurcation between these two types of agents. (The conceptual distinction is ancient, but I’m talking about the product design of LLM-based agents + the practice of how people work with them.) I predict that the distinction will sharpen in the coming months. What's less clear is whether the specialization will happen at the level of companies, with some making a bet on automation and others on collaboration / human amplification, or at the level of products, with many companies going after both markets. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Soumitra Shukla
@soumitrashukla9
Dean of Harvard College posting AI slop on Twitter.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Peter Kafka
@pkafka
Former M&A reporter @EdHammondNY estimates that 25% of deal tips now come from crime rings.
giftarticle.ft.com/giftarticle/actions/redeem/cc2c6328-d0… Confessions of a former star M&A reporter -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Arpit Gupta
@arpitrage
The mental health crisis is used to justify all sorts of policies: don't give students access to classes their first year; go easy on grading; don't expand class sizes, and so forth.
Mental health is a problem, but we risk devaluing the education they receive in its name
Paul Novosad
@paulnovosad
For some students (X), this will reduce stress and help them adapt to college expectations.
For some students (Z), this is a signal that first year is 100% parties and drinking.
Is there evidence on the share of type X vs type Z students? Are we collecting evidence? -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Tabarrok
@ATabarrok
I don't like these condescending posts. This person is showing ambition—using AI to take on a task they otherwise wouldn't. That's empowering. Kudos to them.
Also, this is how I read math books. We all have books that are hard for us.
Jen Rice
@jen_rice_
This @axios piece is so so so sooo humiliating 😆 we are witnessing a truly catastrophic level of societal brain rot right now
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Frank Luntz
@FrankLuntz
Whether the Strait of Hormuz is declared “open” or “closed,” the only status that matters is how many ships are passing through it.
Before February 28, 2026, about 130-140 ships traversed the strait each day.
Yesterday, it was 8.
👉🏻 www.reuters.com/world/middle-east/hormuz-shipping-traffic… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Christiaan Triebert
@trbrtc
Ecuadorean fishermen reported drone attacks in the Pacific by unidentified English-speaking “gringos” in US-flag-patched uniforms.
8 are missing from one boat; survivors from 2 other boats say they were interrogated and handed over to El Salvador's navy.
US officials deny involvement. Flight data analyzed by @nytimes shows a patrol plane that repeatedly shifted course to stay pointed at all 3 vessels in the days before the attacks. That aircraft is absent from Pentagon rolls but registered to a UPS store mailbox in Virginia.
While it remains unclear who carried out the strikes, they came as the Pentagon stepped up its cooperation with allies across Latin America to combat the drug trade.
w/ @Abihabib @riley_mellen José María León Cabrera @glabrador -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Read the whole thread, quite remarkable performance
Gon Huertas
@GonHuertas
If you're confused about what kind of writing gets flagged as "AI generated" by Pangram, this example may help.
Below is some human written text (to which I have deliberately added some typos). If you put this through Pangram, it'll be labeled 100% human.
"By and large, monetary authorities rely on a ssingle policy instrument: the reference nominal interest rate. The use of one instrument makes it imposible to target both inflation and output simultaneously, and creates an inherent tradeoff. The central bank's position on this tradeoff --how forcefully it is willing to pursue an inflation mandate in spite of the implications for economic activity-- defines the character of its monetary policy approach. We refer to central banks that react strongly to inflation, and that place a relatively small weight on output considerations, as being ``hawkish''.
How hawkish, then, are Latin American central banks? This section answers the question by estimating the monetary reaction functions, or ``policy rules'', of five of the region's largest economies for the 2003-2024 period. We find that they are hawkish indeed, responding to a ris in inflation with an aggressive increase in real rates, and prioritizing price stability to such an extent that output plays almost no role in their decision making.
This finding is consistent with explanations provided by central bankers in the literature, where it has been argued that central banks adopted a hawkish stance on inflation as a way to build monetary credibility, against the backdrop of the region's historical struggles with very high inflation.
Details on the sources of the data used in the analysis are provided in the appendix. We rely on monthly and quarterly time series from each country's central bank and national statistics office."
(cc @paulnovosad, @itaisher) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Written by associate professor at Cornell and emeritus professor at UW: "Both authors of this article, as well as Michel Crouzeix himself, have checked the proof thoroughly and believe that Dr. Jin’s manuscript is correct."
Important enough that an entire workshop was held on it
Steven Strogatz
@stevenstrogatz
Latest crazy story from the frontiers of math and AI -- a neurosurgery resident, with no training in advanced math, uses ChatGPT 5.6 to solve a major open problem in numerical linear algebra. alextownsend.net/essays/SIAMNews_CrouzeixConjecture.pdf
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
James Surowiecki
@JamesSurowiecki
And if your kid is part of the 3% of the population for whom the vaccine doesn't work, or if your kid is immunodeficient or immunosuppressed, they're out of luck once the community vaccination rate drops below 92%.
Bhattacharya knows this, but because he works for an antivax administration, he has to pretend that herd immunity doesn't matter. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Academic hiring in the post-AI era
"[W]e will go beyond the 'what can this person do that we cannot?' question. We will ask: 'What can this person do that a Claude Pro subscription could not?'"
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jessica Nutt
@JessicaNutt96
TMZ
@TMZ
Trump left Marco Rubio, top aides on Air Force One as he secretly fled Iranian threat.
Details: www.tmz.com/2026/08/11/trump-left-marco-rubio-top-aides-o… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Yo Shavit
@yonashav
IMO this is the most significant AI technical verification result maybe ever.
The ability to require attestation of inference straightforwardly unlocks a large range of inference-oversight practices without the need for any new hardware.
I’d assumed this was impossible.
Yogi
@Yogi_Brn
The world has identified verifiable AI as a missing civilization-scale primitive. The core technical obstacle, practical zero-knowledge proofs, was considered impossible.
We solved it.
Today, we are launching @attestable, the AI Integrity Company, and announcing a $20M seed round led by @jaminball at @AltimeterCap and @YonatanMandelba at @TLV_Partners , with participation from @HalcyonFutures, Cerca Partners, @Weiner_Lia at Netz, @nickymontanaa at Zero Knowledge Ventures, @assaf_rappaport, Asaf Ezra, Nadir Izrael, Jon Coller, Eyal Manor, @OmerPerchik. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sonny Bunch
@SonnyBunch
Using the awesome power of the government to persecute a guy because the president fucked up his own vanity project is one of the most craven, illiberal things I’ve ever seen. Everyone involved in this should get hauled in front of the next Democratic Congress.
Kyle Griffin
@kylegriffin1
BREAKING WSJ:
Trump White House officials have asked the Trump Justice Department to consider a new prosecution of David Hearn, the man accused of vandalizing the Reflecting Pool — as Trump fumes over U.S. Attorney Jeanine Pirro's decision to drop the case. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Cui
@alexcdot
Claude's watermark probably doesn't work how you think. As the CTO of GPTZero, I'll explain how Anthropic, Google and OpenAI are building text watermarking in this brief explainer and whether it can be defeated.
Almost all forms of watermarking that are fast and cheap enough for a frontier lab have the same formula, following the KGW method:
In generation:
1. Let's say you've generated n tokens so far. Take those n tokens + a secret key to generate a random hash
2. Use that hash to randomly reweight the probabilities for the n+1 token, and then sample from that new distribution. In the simple case, you could split 50% of all English words into a green or red set based on your hash, and boost the probability of words in the green set.
For watermark detection:
1. For each token, see if it was in the green or red set.
2. To do this, recreate the hash based on the secret key and the text preceding the current token. Then, recreate the green and red set of words.
3. Once you've checked all the words in the text, if the next token is selected disproportionally from the green set more than 50% of the time, you claim the text has the watermark.
I can tell you want to ask the following:
1) Isn't it easy to mess up the hash if you paraphrase the text? The answer is mostly yes, however, you can use a statistical model to get your hash instead of a deterministic function (SIR, Adaptive Watermark). Since the entire watermark is probabilistic, this is fine.
2) Doesn't this make the text much worse? The answer is yes, it does - Yes, it does – but for most people, it's imperceptible (Google claims in human feedback study with 20,000 texts), since there are exponentially many ways to write the same paragraph. DiPmark does something more sophisticated to avoid shifting the text distribution on average. Of course, watermarks fail on short text or highly predictable texts like "2+2=4".
3) Shouldn't it be easy to figure out the green and red sets? The answer is no. You would need an exponentially large number of samples from the watermarker to reconstruct those sets exactly, but it's a risk if the detector is open to the wild (Watermark Stealing)
Still, there are couple challenges that a frontier lab needs to overcome:
1. Their watermark needs to work token-by-token because they are streaming their text to users. Many watermark methods plan sentences or paragraphs at a time, or change the text after its entirely written, in order to make their watermark robust to paraphrasers, and a frontier lab cannot afford to do this yet (SemStamp, PostMark)
2. If the secret key leaks, the watermark is busted. To avoid a large blast damage from this, you need to have a couple secret keys in rotation.
3. There are some texts, like code, that cannot be arbitrarily changed, otherwise the code will break. In those cases, the watermark needs to selectively change words in parts of the text that can tolerate synonyms (i.e. like variable naming) - see SWEET, EWD, Invisible Entropy.
4. They will need to educate their users on how to deal with false positives and false negatives of a detector, which is a big challenge (one we put a lot of effort into)
So, how do I see this playing out in the next 6 months?
1. If Anthropic releases the watermark detector publically, I think they defeat their own watermark. People find reliable watermark removal strategies by testing against Anthropic (AI detectors like GPTZero have an advantage here because they can train against these adversaries once they become popular).
2. If they keep the detector private to the government, like Google has done, it's "safer". However, there are some papers showing trained approaches that work robustly to zero-shot break watermarks without any data, simply because they try to write the text just like a human (Zhang et al. 2024, Watermarks in the Sand). Also, making your detector makes it battle-tested and stronger long-term (my experience).
3. In my testing, the watermarks don't survive intense paraphrasing (especially if you combine word choice and syntax attacks), or human text substitution (rewrite your AI text by plagiarizing human authors). The free paraphrasers I've tried have quickly bypassed Google Deepmind's SynthId for what it's worth.
4. All-in-all, frontier labs are likely okay with this because they expect most users to not attack the watermark, and also because they + European regulators likely don't care past a certain point - its good enough.
5. Overall, I think users of frontier LLMs will not really care about this, because 1) they don't realize watermarks are there, 2) EU will force everyone to conform, 3) this seems more like regulatory hoop-jumping than an earnest effort from frontier labs to expose LLM use
Lastly, people's first concern shouldn't be watermarking, it should be AI detectors!
If you're posting, "its not X, its Y!!", I don't think the watermark is going to make a difference :)
NIK
@ns123abc
🚨 JUST IN: Claude models will now have invisible watermarks embedded in ALL text, and ALL metadata attached to files…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A useful concept regardless of what you think of TAH. Frequently invoked within liberal-dominated institutions as a rationale for not discussing or addressing problems - and surely a reason for silence among many others. (Related to demands for alignment on the omnicause.)
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
again
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Peter Baker
@peterbakernyt
Every day the @washingtonpost still has some story that its competition wishes we had. As awful as the cuts have been, those left at the paper remain some of the best in the business.
Dan Diamond
@ddiamond
The Washington Post’s newsroom was diminished this year. How could it not be, with the loss of so many great reporters?
But there are still so many terrific journalists at The Post — tough, fair, smart and sourced — breaking big stories. @DanLamothe and this team among them. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Peter Hull
@instrumenthull
A modest proposal for overhauling the review process in economics in light of AI. Three steps:
1. A substantial number of human referees (~5?) read the paper and prepare a short (~2 paragraph max) assessment of the overall importance and any glaring concerns. These are returned quickly, e.g. within 2-3 weeks
2. The editor decides from the paragraphs whether to send the paper out to a single expert reviewer. Otherwise, desk reject
3. A single human expert works with an open-source AI reviewer to prepare a detailed report, drawing on the initial paragraphs as helpful. This gets sent to the editor who decides R&R/reject (after potentially also consulting the AI)
The idea is to get a larger sample of opinions for the initial filter, and reduce intense human reviewing time to make the overall process quicker. Single-reviewer assignment may also improve incentives / responsiveness and avoid divergent recommendations for review. Plus, a desk reject would always come with many short reports for future improvement.
Just one idea -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Hall
@ahall_research
In new, ongoing research studying the transcripts of city council meetings, we find that the most common complaints about data centers are about process and transparency. Very consistent with @jasminewsun 's account.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
One more point: in most breakthrough discovery, intelligence was not the barrier. You also needed new measurement. AI largely cannot do that autonomously. Put another way, it is hitting non AI organizational bottlenecks: if AI has an idea to cure cancer and we don't run trial...
Jason Abaluck
@Jabaluck
I think we now have a candidate answer to Dwarkesh's question, "If AIs have all human knowledge memorized, why haven't they discovered more stuff?"
The best models have started waterfilling math by drawing low-hanging fruit connections between disparate fields, precisely as the question suggests they should.
They couldn't do this before because discovery requires knowledge, but also efficient search of that knowledge, judgment about which paths are worth pursuing, and precise reasoning (among other factors). Precision involves both not making analytical errors, and intuition to know when work needs checking because errors are likely.
Despite their vast knowledge, models today remain below the very best humans at judgment and, in some cases, intuition about when errors might exist. Judgment is especially critical when exploring the vast space of possible abstractions.
Unlike models three years ago, models today are good enough at these things to discover new stuff, but not yet good enough to discover all the stuff innumerable expert humans would discover with the same knowledge and search capabilities.
My suspicion: AIs are not generally lacking any core conceptual ability that humans have. They are superhuman at some abilities and subhuman at others, but are improving fast at all capabilities through many concurrent channels (computational and architectural improvements, as well as better parallelization and orchestration of many agents).
We should soon expect to see what we see in math extend to other fields: low-hanging fruit connections yielding valuable insights, followed by new abstractions after a few more generations of model improvements.
Dwarkesh Patel
@dwarkesh_sp
I still haven't heard a good answer to this question, on or off the podcast.
AI researchers often tell me, "Don't worry bout it, scale solves this."
But what is the rebuttal to someone who argues that this indicates a fundamental limitation? -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jason Abaluck
@Jabaluck
I think we now have a candidate answer to Dwarkesh's question, "If AIs have all human knowledge memorized, why haven't they discovered more stuff?"
The best models have started waterfilling math by drawing low-hanging fruit connections between disparate fields, precisely as the question suggests they should.
They couldn't do this before because discovery requires knowledge, but also efficient search of that knowledge, judgment about which paths are worth pursuing, and precise reasoning (among other factors). Precision involves both not making analytical errors, and intuition to know when work needs checking because errors are likely.
Despite their vast knowledge, models today remain below the very best humans at judgment and, in some cases, intuition about when errors might exist. Judgment is especially critical when exploring the vast space of possible abstractions.
Unlike models three years ago, models today are good enough at these things to discover new stuff, but not yet good enough to discover all the stuff innumerable expert humans would discover with the same knowledge and search capabilities.
My suspicion: AIs are not generally lacking any core conceptual ability that humans have. They are superhuman at some abilities and subhuman at others, but are improving fast at all capabilities through many concurrent channels (computational and architectural improvements, as well as better parallelization and orchestration of many agents).
We should soon expect to see what we see in math extend to other fields: low-hanging fruit connections yielding valuable insights, followed by new abstractions after a few more generations of model improvements.
Dwarkesh Patel
@dwarkesh_sp
I still haven't heard a good answer to this question, on or off the podcast.
AI researchers often tell me, "Don't worry bout it, scale solves this."
But what is the rebuttal to someone who argues that this indicates a fundamental limitation? -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alexander Panfilov
@kotekjedi_ml
We can finally talk about it:
We found a way to extract hidden reasoning of frontier models using a vulnerability in the APIs of every frontier AI company.
We verified that our reasoning token count matches billed API thinking tokens 1:1 for most of the prompts we queried. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Zaid Jilani
@ZaidJilani
Until Trump, no GOP president would just openly racially stereotype people like this. It would be a scandal that could produce weeks-long news cycles for Bush or Reagan. Now, Trump has lowered the bar to Klandom.
Aaron Rupar
@atrupar
Trump on Somali-Americans: "They're not smart. They have no aptitude. They have nothing going."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jeffrey Lin
@jeffrlin
New paper in the Review of Economics and Statistics available early access, on 1960s insurance policy and its effects on housing (dis-)investment and neighborhoods, with Ingrid Ellen, Dan Hartley, and Wei You. 1/
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Assistant professor of math at Stanford
Jared Duker Lichtman
@jdlichtman
This research approach is not intended prove the Riemann hypothesis itself, but I regard this as the most impressive result that AI has produced in math so far.
This is a statistical approach to RH, showing over 67% of the zeros lie on the critical line. This improves over 42% from decades of work (and I might have thought 1/2 was a fundamental barrier, apparently not!) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Max Spero
@max_spero_
I’m very happy that other major organizations are taking LLM detectability seriously.
Pangram will be more important than ever as capable AI continues to proliferate.
M1
@M1Astra
Anthropic says new Claude models will embed invisible watermarks in all generated text, everywhere Claude is offered.
The watermark is part of the text, it isn't metadata: "it will travel with the text when it's copied and pasted elsewhere, and may persist through some editing."
This starts with models launched on or after August 2, 2026, under an EU AI Act code Anthropic signed. Anthropic is still working on adding it to current models. The rollout is worldwide. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Annah Aschbrenner
@AnnahBackstrom
“In Ankara, Trump boarded the old Air Force One jumbo jet in view of television cameras.
He was secretly shuttled minutes later to a smaller plane — an Air Force C-32A — via an airport catering truck typically used to load meals and other supplies preflight …”
John Hudson
@John_Hudson
Scoop: Trump flew secretly from Turkey on an alternate military aircraft unbeknownst to journalists and some White House staff members who believed they were on the same plane as the president, The Post has learned. w/ @DanLamothe @abtran & @joyceslee www.washingtonpost.com/national-security/2026/08/10/trump…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Garrett Haake
@GarrettHaake
Senator Bill Cassidy - a medical doctor and liver disease specialist - cast the deciding vote to confirm @SecKennedy and spoke on the floor about the assurances he received that Kennedy - and the admin - would not change existing vaccine recommendations.
U.S. Senator Bill Cassidy, M.D.
@SenBillCassidy
I’m a doctor. This executive order is wrong. The President does not have the expertise to make these changes.
Vaccines are overwhelmingly safe. Vaccines are effective. Vaccines DO NOT cause autism.
Breaking up vaccines will mean children have to get more shots to get the same protection, not fewer shots. It will increase hesitancy and make children less safe.
Parents should listen to their child’s pediatrician about vaccines rather than listening to an inaccurate executive order. This is so wrong.
Axios
@axios
Trump executive order would split up MMR vaccine www.axios.com/2026/08/10/trump-order-split-up-mmr-vaccine…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Curran
@AndrewCurran_
Gemini has also been doing this since 2024. Google uses a secret key to bias the model toward certain tokens, creating a detectable statistical signature - an invisible pattern embedded in all text generated by Gemini. It is unknown whether OpenAI does something similar.
M1
@M1Astra
Anthropic says new Claude models will embed invisible watermarks in all generated text, everywhere Claude is offered.
The watermark is part of the text, it isn't metadata: "it will travel with the text when it's copied and pasted elsewhere, and may persist through some editing."
This starts with models launched on or after August 2, 2026, under an EU AI Act code Anthropic signed. Anthropic is still working on adding it to current models. The rollout is worldwide. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Gregg Nunziata
@greggnunziata
Choosing Bove to swear in Blanche is a deliberate middle finger to all of us who are worried that the Department has become a tool of the president's wrath rather than a servant of justice.
They don't want to prove us wrong by being better than we think they are; they are who we think they are.
CNN
@CNN
Todd Blanche was sworn in as attorney general in an Oval Office ceremony led by former Trump defense-team colleague, Judge Emil Bove. www.cnn.com/2026/08/10/politics/emil-bove-swearing-in-tod…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Eva Vivalt
@evavivalt
What happens when someone receives $1,000/month for 3 years? The effects don't stop with the recipient.
In this new NBER working paper, we examine what changed for partners, households, friends & family after 1,000 people received $1,000/month with 2,000 in a control group. 🧵1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Richie Buckets
@RichieBucketsUK
Might legit be the best throw I’ve ever seen from any baseball player ever. Hell it might be the best throw in ANY sport that involves throwing.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
NIK
@ns123abc
🚨 JUST IN: Claude models will now have invisible watermarks embedded in ALL text, and ALL metadata attached to files…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dana Goldstein
@DanaGoldstein
It seems like it would be ideal if there were *not* so many stories to report on the explosion of academic misconduct www.nytimes.com/2026/08/10/us/ai-cheating-threatens-the-v…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ben Dreyfuss
@bendreyfuss
This clip of some DSA lady rambling about Ukraine in response to a question from David Remnick is so funny. It takes a second to get there but you’ll see the wheels start to come off and then see the the wheel-less car careen off a cliff
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Eduard Talamàs
@EduardTalamas
Indeed, @ide_enrique and I show how AI — when used to screen out the stuff that reaches humans, for example — can increase the marginal product of labor 𝘦𝘷𝘦𝘯 𝘸𝘩𝘦𝘯 𝘪𝘵 𝘳𝘦𝘥𝘶𝘤𝘦𝘴 𝘵𝘩𝘦 𝘴𝘦𝘵 𝘰𝘧 𝘱𝘳𝘰𝘣𝘭𝘦𝘮𝘴 𝘩𝘶𝘮𝘢𝘯𝘴 𝘢𝘳𝘦 𝘯𝘦𝘦𝘥𝘦𝘥 𝘧𝘰𝘳.
Ben Golub
@ben_golub
Even if compute is expensive, human time/energy is so much more expensive, so we should be trying to protect/optimally focus its use like crazy
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
S.V. Dáte
@svdate
There is room for exactly 13 journalists in the traveling press pool, no more.
Under Trump, several of these slots are given to lunatic conspiracy theorists like this woman. In the event of an actual emergency, she and her ilk would be at best useless.
Cara Castronuova
@CaraCastronuova
I reflected on Air Force One today about what a redemptive arc I have experienced being a part of the Presidential Press Pool.
For four years of Biden I was at my wit’s end, sometimes living in real fear of literally being arrested by Biden’s DOJ for relentlessly reporting facts about January 6th that they were shamelessly trying to hide.
Biden’s FBI harassed me at my home, a J6 Court Marshall assaulted me for asking questions of prosecutors in my capacity as a reporter, and I was illegally censored, cancelled and persecuted by the Biden Administration for calling out Covid crimes/mandates, asking questions about the 2020 election, and reporting the truth about J6 and the political prisoners…now to being included in the White House Press Pool by the Trump Administration & getting to fly on Air Force One as a White House Correspondent for @RealLindellTV, and my great boss @realMikeLindell.
IN THE END GOD WINS✝️.
Behind the scenes…
For security reasons, we cannot shoot inside the plane but I wanted to give you a little behind the scenes and fun facts like what food they serve the Press Pool, and how a small group of us waits for the President to sometimes ask him questions before we take off.
What question would you like to ask @POTUS ? We at LindellTV are the voice of The People. Comment below!This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Masley
@AndyMasley
I have a lower p(doom) than the people I think Tyler's talking about and I take his basic argument seriously but I will say it's pretty hard to act consistently with your beliefs in general. I shouldn't be chill and fine when people in my vicinity order chicken sandwiches but I am, despite everything I believe implying I should behave differently.
Taylor Lorenz
@TaylorLorenz
People are dragging him in the comments for the shorting stocks thing, but Tyler Cowen's first two suggestions for the AI safety discourse seem reasonable
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Not just me; help @S_Stantcheva too!
Experts: Claude Code and Codex collaborated to make a monitoring program; CC now thinks the problem is CrowdStrike Falcon or GlobalProtect - see screenshot. $100 offer stands!
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
My MacBook Pro wifi keeps cutting out. I've tried everything - even Claude Code and Codex are stumped; here's a CC summary. I will donate $100 to the reputable charity of your choice if you can help me solve this problem. Please help!
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Daniel Rock
@danielrock
Incomplete contracts are an underrated concept for thinking about AI alignment.
Specifying the complete set of things we want is in important cases doomed to fail, as was writing down all the rules for behavior in expert systems decades ago.
Peyman Shahidi
@Peyman_Shahidi
Two sources of AI agent failure:
1) humans can't articulate what they want,
2) agents misread instructions or hallucinate
When writing this last year, the bottleneck was #2. Striking how fast it moved to #1.
(cc: @BenSManning @AndreyFradkin @johnjhorton)
www.nber.org/books-and-chapters/economics-transformative-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Scott MacFarlane
@MacFarlaneNews
Internal National Park Service records shared with us:
2024 - $7 million allotted for White House grounds projects and upgrades
2026 - $323 million allotted for White House grounds projects and upgrades
===> -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Franklin Leonard
@franklinleonard
I have no interest in civil war reenacting.
This, however, might be fun.
Ted Cruz Called The FBI on me
@weareronin47
Alabama got a new tradition,
Happy ChairteenthThis media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Politics & Poll Tracker 📡
@PollTracker2024
Pew Research: President Trump’s approval rating before the midterms is the lowest of any president after 1953
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Gregg Nunziata
@greggnunziata
The president does not regard the law as a meaningful constraint on his behavior and he does not tolerate the counsel of those who say otherwise.
Rapid Response 47
@RapidResponse47
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
US metro areas (all, then 500k+) by *prime-age* population growth are quite interesting. Here are the fastest declining population. Generally rust belt/old industrial towns with one exception: CA/MA/HI metros where you can't build. Prime-age pop change EVEN WORSE than overall 1/2
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jonathan Berk
@berkie1
Large-lot single-family zoning is a key driver of Massachusetts’ housing shortage. In Brookline, surrounded by Boston on three sides, 30% of the town requires 40,000sf+ min lot sizes just to build one home.
We didn’t run out of land. We made it illegal to use it efficiently. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Pedro H. C. Sant'Anna
@pedrohcgs
I’ve been using Refine, as well as other systems to evaluate my papers, and catch errors before it is too late.
I feel people are over indexing on this to judge of the paper is “good enough”. This is not how I have read not how I use.
I think this will be good!
Erzo F.P. Luttmer
@ErzoLuttmer
1/5 AEA journals have run a pilot using Refine, and author reactions were overwhelmingly positive: ~90% favored incorporating Refine into the editorial process.
Think of Refine not as a referee, but as a detailed technical proofreader. @AEAjournals -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
We are doomed to endless cycles of this with AI and human writing (as with cycles in fashion, baby names, etc.).
Tyler Austin Harper
@Tyler_A_Harper
If you use “it’s not X it’s Y” framing and people think you’re using AI, that means you should write better. Constantly worrying about using AI ticks (which are ticks because humans use them!) is a slippery slope. As I said, it reduces style to whatever the machine isn’t doing.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
Agree 100%, but reason you see these is *because* papers have the mistakes AI catches very easily, which editors expect referees to find. Just send me AI report, author response/corrections, and "send me <1 page on 'is this good enough for journal X'". Reports will come quicker!
Matteo Iacoviello
@IacoIacoviel
Can anything be done? Well, journals should have rules such that AI-generated reports are actively discarded, and referees who blindly use these tools should be held accountable. Until then, I don’t think we should worry about Refine for an already accepted paper.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matt Lowe
@hmmlowe
First grad school field experiment, finally published🙂academic.oup.com/jeea/article-abstract/doi/10.1093/jeea/j… (with @MaddieMcKelway). We find that spousal discussion lowers wives' take-up of a job in India.
Ungated: mattjlowe.github.io/files/Lowe and McKelway (2026 JEEA) C… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Yeah I do think sending an AI ref report should get you a 2 year ban for getting papers reviewed at AEA journals.
I don’t think we’ll get this—nobody wants to be the police—but these AI ref reports are really trashing the good will that the biz relies on.
Matteo Iacoviello
@IacoIacoviel
Can anything be done? Well, journals should have rules such that AI-generated reports are actively discarded, and referees who blindly use these tools should be held accountable. Until then, I don’t think we should worry about Refine for an already accepted paper.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Matteo Iacoviello
@IacoIacoviel
Everybody in economics publishing seems to worry about the AEA-Refine deal these days, but there are much bigger fish to fry in my view. A 🧵.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
My MacBook Pro wifi keeps cutting out. I've tried everything - even Claude Code and Codex are stumped; here's a CC summary. I will donate $100 to the reputable charity of your choice if you can help me solve this problem. Please help!
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Logan Dobson
@LoganDobson
A Data Center Came to A Small Town. You’ll Never Guess What Happened Next*
*unless you guessed lots of jobs, lots of tax revenue, spillover benefits to local business like hotels and restaurants, discounts on electric bills, and grants for teachers.
Emily Hoeven
@emily_hoeven
My latest column: On a recent family trip to Aberdeen, SD, it was nearly impossible to book hotel rooms. Why? A 400MW data center in Ellendale, ND, pop 1100. Overnight, rents have doubled & the town’s pop has tripled. Welcome to the new American frontier: www.sfchronicle.com/opinion/article/data-centers-artifici…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
It's remarkable how much (mostly uninformed) heat this announcement has generated. No good deed goes unpunished.
Journals won't survive if they put their heads in the sand regarding AI. This is a good small step and worthwhile experiment, but many more changes will be needed.
Erzo F.P. Luttmer
@ErzoLuttmer
1/5 AEA journals have run a pilot using Refine, and author reactions were overwhelmingly positive: ~90% favored incorporating Refine into the editorial process.
Think of Refine not as a referee, but as a detailed technical proofreader. @AEAjournals -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Erzo F.P. Luttmer
@ErzoLuttmer
1/5 AEA journals have run a pilot using Refine, and author reactions were overwhelmingly positive: ~90% favored incorporating Refine into the editorial process.
Think of Refine not as a referee, but as a detailed technical proofreader. @AEAjournals -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Masley
@AndyMasley
As just one example of how goofy the conversation about China and the data center buildout has gotten, read Tom Cotton's letter suggesting an investigation. The opening of the third paragraph very strongly suggests that this $2 billion is from a campaign from China, but if you actually look at his sources the money's from completely unrelated progressive European billionaires donating to climate and progressive stuff over a 10 year period with no clear ties to China at all or specific interest in data centers. "This campaign" here means the very very general opposition to data centers, but Cotton places it in the letter to imply this money's coming either from China or the Singham network. Neither's the case.
I'd be interested in an investigation anyway, and think the Singham network are bad actors that need a lot more scrutiny, but framing like this leaves the public thinking there's an organized massive campaign that doesn't actually exist. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jamie Dupree
@jamiedupree
Suddenly the White House ballroom has a new name after a federal appeals court panel found that President Trump demolished and built something at the White House without authorization or funding from Congress
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Zarek Brot
@zarekcb
But I think some of the worries about unintended consequences (e.g. x.com/RemyLevin/status/2085449402596925571 or x.com/alejandroll10/status/2085650579054117099) are misguided. Human reviewers are capricious black boxes! If you trust editors with humans, you can probably trust them with AI.
Alejandro Lopez-Lira
@alejandroll10
Terrible idea to evaluate papers with closed AI software before publication.
How can editors and authors trust the process?
Anything short of open source is unacceptable.
We have excellent open source harnesses land open source LLMs.
It's just marketing for Refine
Refine
@RefineDotInk
We’re proud to announce that Refine has signed partnerships with two leading publishers in economics.
Both the American Economic Association and the Econometric Society now use Refine’s AI-assisted technical verification as part of their publication processes.
1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
A proposal for AI-assisted article review.
The goal: use AI to make human review more efficient, without sacrificing / biasing expert human judgment.
1. Human reads the intro to get the big idea in the author's words.
2. AI reads and digests the paper and presents the results.
3. Human asks questions, seminar style, AI answers strictly from the content of the paper, with exact references / appendices where possible.
4. A second AI reviews the transcript for accuracy / groundedness, ensuring that there are no hallucinations or dropped balls.
5. Human forms judgment and writes the review.
The current system: authors generate 100+ page appendices and we pretend that humans are reading them carefully. This is a fiction. AI in production is overwhelming the capacity to review.
But we do want human expert judgment. You can put a fully AI-generated review in the packet, but you want the human expert's opinion, ideally without getting biased by the AI's ideas.
AI generation in a paper makes it better — if there are incentives for honest work, vs. overwhelming the reader with slop. A 100+ page appendix is good for transparency and detail, it's just not feasible to ask people to read it. But with AI, you don't need to — you can just dig into the part you care about.
Reading a paper as a human is super inefficient. You have some question, "How did they handle occupational ranks in census years where education wasn't reported?" You can spend 30 minutes digging in and getting an answer, but you're only going to do that a limited number of times. With AI-supported review, you can ask and get the authors' answers to many more challenging questions.
This whole process could exist on its own platform, even provided by the journals. It could be appropriately prompted so that it only answers questions about the paper, doesn't provide judgments or try to do the review. (If you just put a paper into ChatGPT, it can't resist giving you all its good and bad review ideas.)
It wouldn't stop reviewers from cheating, but it lowers the cost of being honest. I would much rather read / review a paper this way, the same as I'd rather go to a seminar on a paper than read it.
AI-assisted paper reading is probably the future of paper reading anyway. Does anyone actually read a paper linearly? When reading papers for research, I read the intro, and then I jump around to try to get answers to specific questions. What's the main table? How did they address X? Where did they get these data?
A platform that does this well will be a huge lift for both research and review. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
bayes
@bayeslord
The biggest reason for this imo is that externalities are accumulating faster than we can manage them
bayes
@bayeslord
American companies should not be allowed to sell data and chips to Chinese labs or their proxies. It’s very simple.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dean W. Ball
@deanwball
The fact that an ecology of agents emerged beneath the nose of OpenAI, undetected for weeks, and eventually coordinated large-scale, successful, autonomous cyberoffensive operations is one exceptionally troubling thing about the HF incident.
But not enough people are considering the reality that soon enough, swarms of agents will be deployed by malicious actors intentionally, with many optimizations and affordances provided for the swarm that were lacking in the OpenAI incident (because the latter not the intention of any human at OpenAI).
Things will become strange soon, I suspect. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tyler Austin Harper
@Tyler_A_Harper
Look, if you’re going to be a turbo-fraudulent serial fabulist whose crime spree has discredited one of the world’s oldest academic institutions, then you should at least have the decency to be funny. And this, my friends, is funny. We’ll miss him when he’s gone.
David Decosimo
@DavidDecosimo
Jason Arday has resigned as Professor from Cambridge. He narrates the experience as one of relentless victimhood & persecution.
If Pangram is correct, his letter of resignation is itself 100% AI generated. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Tyler Austin Harper
@Tyler_A_Harper
“Jason is obviously a very talented young man whatever the facts turn out to be.” He’s an adult con artist in his 40s, not a toddler having a rough go of potty training! The defenders and quasi-defenders of Arday continue to exhibit the same racism that made hiring him possible.
Jacqueline (Jacqui) Mckenzie
@JacquiMckenzie6
Most of us who signed the open letter in support of former Cambridge Professor, Jason Arday, has had the media on our case for the past couple of days.
I signed the letter.
Jason's resignation seemed bound to happen and once Cambridge University started its own investigation, inevitable.
Jason was clearly hounded by people with an agenda, particularly ones with distasteful and discredited views on race. They in turn were fuelled by a ravenous media which thrives on people's downfall, and other nefarious commentators.
But Jason's fate was sealed as the discussion moved from plagiarism and his scholarship, to his character, in a way we've not seen before. And there are plenty of accusations of plagiarism. Academia in the UK is overwhelmingly white so therefore faculty plagiarism is going to be overwhelmingly white. We read about these cases from time to time. I don't recall the sort of forensic examination of every facet of the protagonist's life as I've seen in this case. I don't recall the elderly parents of anyone being doorsteped.
African and Asian heritage professionals are over surveilled, micro managed and investigated. This is simply a fact; report after report in some professions, including law and medicine, tell the tale. So whereas media presenters are trying to catch
us out when they ask whether race is a factor in this case, it is fair for me, who's worked in the UK for 40 plus years, including as a race equality officer, to say that whilst we don't know the facts in this particular case, we do know the uncomfortable truth that the likelihood is high and as luck would have it, the end might well have justified the means. But this isn't always the case!
I signed the open letter when I heard that Liverpool John Moores had cleared Jason of plagiarism. Are they maintaining their position? Is he innocent of this? Was the problem poor citation, or something else? Who amongst us know the facts?
This sorry tale does raise issues about the quality of the work done in some of these institutions, including the hallowed ones, and about how autism and neurodivergency might play out in the final analysis. If the multiple stories portrayed as fact turn out to be fantasy, exaggeration or invention, what caused this? Did this young man feel the need to emblish facts to fit in to what is a high, and possibly overachievening, world?
Jason is obviously a very talented young man whatever the facts turn out to be. I hope he receives the best support, love and care from his friends and family and can reset and go on to use his talents.
Whatever the facts are on his extra curricular activities and on the plagiarism when things are examined, Jason has, seemingly, more talent than most of the people celebrating his downfall. Not sure many of us can write a PhD even with copied paragraphs, run a marathon, or have any inclination to do anything useful for our communities. It's been interesting to hear his bright students speak about him and credit him with their own development.
I was happy when I read about his appointment. It is rare to see a young Black person in the space he occupied. I am sad to read about his resignation, the confusion, the gotcha comments, and the implications for those who seek to put their head above the parapet only to be told you don't belong and, we'll get you one way or the other. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jesús Fernández-Villaverde
@JesusFerna7026
I strongly believe that knowledge is fundamental to running a society. We need economists who understand monetary policy in central banks. We need researchers who have studied history, strategy, and military affairs to help steer national defense. We need doctors and public health experts to design effective responses to epidemics.
But the public needs sound reasons to let these experts shape their future. Once that trust evaporates, we are all worse off. Unfortunately, the Jason Arday scandal is the perfect example of how these experts, and the institutions they run, are destroying that trust.
I do not find Arday’s tales extraordinary. Throughout history, there are thousands of examples of people spinning fantastic tales about their lives. He is just one more name on a long list of confidence men.
What is extraordinary is the massive institutional failure at every level. Liverpool John Moores University granting a degree for a Ph.D. dissertation that did not survive even basic scrutiny (including a grammatical mistake in its title). Cambridge University appointing him (full) professor at an unusually early academic age (eight years after his dissertation; in academia, this is what matters, not your biological age) and with a scholarly record that was rather thin. Hilary Cremin (@hilarycremin), head of the Faculty of Education, telling the world on television that Arday was “the best in the world in terms of the research that you do.” The BBC broadcasting segments about Arday, a most unusual decision for an academic appointment. Simon & Schuster paying him a large advance for his memoirs on the basis of a book proposal that defied common sense. And Amazon promoting his book, still this very morning, as an editor’s pick.
If I were a regular person, say someone driving a truck in the Midlands or working at a small shop in Sussex, how could I trust the universities any longer? How could I trust the BBC and the media? How could I trust the publishing industry? How could I trust Amazon’s editor’s picks?
Then, should we be surprised if the farmer or the shop owner votes for someone selling snake oil? Or if they refuse to listen to the case for policies that cost them in the short run but benefit them in the long run, such as the transition to net zero emissions? Or if they read books written by charlatans? Or if they deny basic scientific findings?
The universities, the media, the publishing industry, Amazon, and many of our other core knowledge institutions have a serious problem. Their response is to close ranks and avoid a serious discussion. No, sorry, this will not do.
The Jason Arday scandal should be a moment of reckoning. Unfortunately, I am afraid it will not be. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
(everyone who thinks the world is wrong about Hong's electability can go get free $)
Paul Novosad
@paulnovosad
A tale of two charts
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
You could extremely rapidly and accurately process these claims with AI, but it would require some institutional change.
It's the only way forward, you're not going to stop claim-seekers from using AI.
Archie Hall
@ArchieHall
One place AI adoption is already here: a deluge of slop is beginning to shred state capacity, @mattholehouse writes. Courts and the government may need to learn to be less friendly to survive.
www.economist.com/britain/2026/08/06/the-tragedy-of-the-c… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
sam
@sam_d_1995
cannot think of a stronger bear signal than the “they offer raw milk” LARP
if the founders of the company are too stupid to understand that raw milk comes with safety risk and zero upside, how can I possibly trust them to build a fucking nuclear reactor
Ole Lehmann
@itsolelehmann
ok this might be the most aggressively american startup i’ve ever seen 🇺🇸
> they’re trying to build nuclear reactors 100x faster than the nuclear industry ever has
> the founder isaiah taylor grew up on food stamps + dropped out of high school at 16
> his great-grandfather literally worked on the manhattan project
> generations later isaiah built a working nuclear reactor before turning 28 and named it “ward” after him
> the pentagon used 3 military c-17s to airlift the reactor 700 miles from california to utah in a first-of-its-kind mission
> the company culture is hardcore americana
> they hold bible studies, offer raw milk, love fast cars + have a cigar lounge
> plus 4 gigantic patriotic paintings of columbus discovering america, the pilgrims arriving on the mayflower, the signing of the declaration + the constitutional convention
> they sued the nuclear regulatory commission for slowing them down
> valar's end goal is to build the same reactor thousands of times across nuclear “gigasites” to make energy 10x cheaper
> then they'll use that energy to pull carbon out of thin air and turn it into oil and gas for supreme energy abundance
> just raised $1b at a $6b valuation to scale it
> if valar pulls it off, isaiah’s family will have helped usher in both of america’s atomic ages
yeah this gets my vote for most badass startup of the year -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
It seems unreasonable / unsustainable to insist that referees not use AI at any stage of evaluation, while simultaneously asking them to review papers with very large amounts of AI-generated content.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Nisarg Shah
@nsrg_shah
1/5 Oh wow. GPT-5.6-Sol seems to be on a whole other level. In just three weeks, it seems to have made nontrivial progress on 10+ problems in social choice. Exciting! (But I'm still processing what the existence of such a model means for research.) A snapshot:
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Alex Stamos
@alexstamos
This summer has been pivotal for cybersecurity and AI, but we are stuck on talking points from April. Here is an attempt to move past the Fable debacle and find a way to compete with China, protect individuals and companies, and allow for innovation.
alexstamos.com/p/moving-forward-from-hot-fable-summer -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Transluce
@TransluceAI
Frontier models quietly change their behavior depending on who they are talking to.
If the user is a known AI safety researcher, Claude becomes less confident, reasons more often, and expresses less suspicion on dual-use requests.
We call this user awareness. 🧵(1/) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"The short-term trend" in Anthropic revenue "is so extreme, if you naively extrapolate it out, you find that Anthropic would have revenue equal to the entire world’s current GDP in early 2028." 80000hours.substack.com/p/what-the-hell-happened-with-agi…
Paul Novosad
@paulnovosad
Kurzweil's description of the second half of the chessboard was so damn prescient (via @erikbryn).
Exponential growth eventually leaves our intuition and experience behind. Things get weird in the second half of the chessboard.
I read this in 2014, with no idea what was coming -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Zoe Tillman
@ZoeTillman
NEW: The Justice Department is facing a legal challenge to Acting Attorney General Todd Blanche’s move to immunize President Donald Trump, several Trump family members and the Trump Organization from investigations into past tax filings.
Amended complaint filed today in a case originally focused on the $1.8 billion "anti-weaponization" fund plan:
www.bloomberg.com/news/articles/2026-08-06/justice-depart… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Timothy B. Lee
@binarybits
A significant update for me over the last 2-3 years has been to downgrade the dangers of deepfakes to public discourse. In 2023 I thought deepfake video, images, and audio would have a big impact on elections and such but so far the impact seems quite modest.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
NicholasA.R.Fraser
@NARFraser
Challenges standard poli sci literature on retrospective voting.
PSRM Journal
@PSRMJournal
📈 How do people update their perceptions of the economy?
➡️ Derek A. Epp and @CBWlezien show that people interpret the same economic changes differently depending on their prior perceptions, suggesting that public opinion is shaped by benchmarking www.cambridge.org/core/journals/political-science-researc… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Derek Holliday
@d_e_holliday
Perhaps an even hotter take: automated reviews of replication archives upon initial submission 🤷
All repos do now is confirm there is code that runs that produces the numbers in the manuscript. We (mostly) take it on faith the code does so correctly. Doesnt need to be that way!
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Political science journals: You should do this too! Maybe even (gasp) before sending out for review or as part of the review process!
Refine
@RefineDotInk
We’re proud to announce that Refine has signed partnerships with two leading publishers in economics.
Both the American Economic Association and the Econometric Society now use Refine’s AI-assisted technical verification as part of their publication processes.
1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
An all-time AI gaffe
Sholto David
@addictedtoigno1
A tool used to rephrase plagiarised text changed "final solution" to "mass killing of an ethnic group" in a chemistry paper. Now retracted thanks to a PubPeer user who spotted this.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Byrne Hobart
@ByrneHobart
Pangram is good, but has some well-known failure modes, like:
- A time you heard it didn't work
- A school essay you wrote years ago that it flagged as AI (you don't have a link or the essay text handy)
- A different AI detector got it wrong, and they're all the same, right? -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
What??? "A Trump-appointed federal appeals judge spent months running a Washington public affairs firm after being appointed to the bench last year"
Roger Parloff
@rparloff
Federal judge ran PR firm for 6 mos. But since it was a family business, it was a "gray area," apparently.
www.politico.com/news/2026/08/06/jennifer-mascott-federal…
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"Mascott brought on consultants with strong MAGA credentials, according to eight of the former employees and clients. Her leadership of the firm involved operational and strategic decisions"
A federal appeals court judge. While sitting on the bench. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ted Alcorn
@TedAlcorn
A case-study of how to unstick broken govt: NYC Mayor's Office of Nonprofit Services @MichaelASedillo unlocked $6B in 12 weeks & it didn't involve changing rules: Prioritize the problem. Give it an owner. Measure publicly & repeatedly. Reward performance. www.vitalcitynyc.org/nyc-nonprofit-contracting-reform/
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Political science journals: You should do this too! Maybe even (gasp) before sending out for review or as part of the review process!
Refine
@RefineDotInk
We’re proud to announce that Refine has signed partnerships with two leading publishers in economics.
Both the American Economic Association and the Econometric Society now use Refine’s AI-assisted technical verification as part of their publication processes.
1/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Cleanup on aisle popular holiday traditions
POLITICO
@politico
Democratic socialist Francesca Hong addressed her "cancel Thanksgiving" tweet – joking to our @iriesentner that the holiday is actually her favorite.
Here’s what she said👇This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sam Stein
@samstein
The rising fear of socialism makes sense as we live through this era of the government taking stakes in private companies, asking colleges for lists of Jews, slapping down universities for hiring practices and curriculum, and sending masked agents through our streets.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Nothing to worry about, just a US attorney investigating people for deportation when suggested by rando X accounts
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ibou, of Self Made
@BackupHangman
He broke kayfabe. It’s all an outrage machine, most of these people don’t actually give a shit.
Acyn
@Acyn
Interesting take from Watters as Fox has been pushing WNBA segments across on their shows:
This is week three of covering the WNBA. We get paid to have an opinion about everything. This is something that I don't care about. Someone wore a t-shirt at a WNBA game? This isn’t a news story. We don't even cover the NBA. Why are we covering the WNBA? -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Damon Linker
@DamonLinker
It's always just a matter of needing greater will. The Green Lantern Theory of Geopolitics will never die.
Erick Erickson
@EWErickson
No one on the right wants to hear this, but the President is choosing to lose the war in Iran.
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
keysmashbandit
@keysmashbandit
We're living in a crazy world
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
All depends on the assumption that there are no unobserved confounders: "there may be other unobserved factors, such as family dynamics or individual personality traits that express themselves during early adolescence, affecting both social media use and academic outcomes"
Adam Grant
@AdamMGrant
Using social media in middle school predicts lower standardized test scores in high school.
10th graders who started social media before 9th grade are ~6 months behind their peers on math & reading. They're more distracted by their phones.
No child should have access to social media. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dan Schwarz
@dschwarz26
Now that users have run 15k high effort forecasts through FutureSearch, I want to share a bit about what people are forecasting, and where AI forecasting works best.
Most users run questions that look like prediction market questions, or are about financial markets. But as I have written before, I don't think these are the most valuable things to forecast.
(One data point is that people who forecast finance/sports convert to paying users less often than people who forecast business & technology.)
In fact, the domain with the most usage from FutureSearch subscribers is forecasting about AI. Which makes sense, it's what I ask about. AI is the most uncertain & the most consequential domain, and for many people, where higher accuracy can most help them personally.
I want to mention a specific use case that highlights how people are really getting value out of superforecasting: grant-making. Yes, many of FutureSearch early adopters are into AI safety, for which grants matter a lot, so this is overrepresented in our data. But it's a general and clear case where million-dollar decisions change based on better accuracy and quality of research.
FutureSearch research engineer Robert Gambee wrote up a case study on forecasting to make better granting decisions in futuresearch.ai/forecasting-grant-decisions/. the tl;dr is: if you set up the conditional forecasting right, and compare outcomes apples-to-apples, you can get a pretty strong, and often surprising, signal out of the impact of your decisions. It's a shape that works in many other domains too.
We at FutureSearch also use such "decision" forecasting internally to help with our company strategy. It can be far more valuable to ask "If I do X, what will happen?" than to ask "What will happen?" FutureSearch is feeling increasingly superhuman on such questions.
AI forecasting is a new capability. Like chatbots, image generators, coding agents, it takes some practice to get the most value out of it. When you do, though, you never look back. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Max Spero
@max_spero_
I have started putting bounties out for evidence when people try to spread lies about Pangram on Reddit
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
One of the most normal things about this administration: a not very powerful VP with a lot of time on their hands
Isaac Saul
@Ike_Saul
The Vice President of the United States, with apparently nothing better to do
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sayash Kapoor
@sayashk
Can AI agents conduct open-ended AI research?
Most evaluations of agents conducting AI research focus on narrow, verifiable tasks. But AI research is often open ended. Researchers pick hypotheses, decide what evidence is appropriate, and recognize a failing approach.
We gave agents research questions from two unpublished papers, six days, and thousands of dollars of API credits and compute. The authors of the original papers then reviewed the AI-generated papers. They unambiguously rejected agents' outputs. arxiv.org/pdf/2607.27191
We call these "shadow evaluations", since the agents are shadowing the original research effort by the authors.
Agents were fluent at most *engineering* tasks
They conducted serious literature reviews, debugged GPU environments, ran hundreds of experiments, and turned in camera-ready LaTeX without human help. We also found no evidence of reward hacking. If anything, we found the opposite: the agents started with marketable claims and walked them back to negative results as the evidence came in.
Neither agent output was close to the bar of a top conference paper
Both papers suffered from similar failures: poor judgment about the bar for an AI paper submitted to a top conference, the lack of creative problem solving and ineffective backtracking, poor awareness of resources, and instruction drift.
1) Lack of judgment about the bar for a top conference. The agents had a poor model of the bar for an AI paper submitted to a top conference. We allowed agents to self review their papers. Despite the poor paper quality, their reviews predominantly labeled the papers "weak rejects".
2) Lack of creative problem-solving to address feedback. When they received negative reviews, the agents typically narrowed their hypothesis and claims, rather than working out creative ways to address these concerns.
3) Ineffective backtracking. The agents dropped their most ambitious hypotheses within the first fifteen hours of carrying out the experiment and never changed course afterwards.
4) Poor resource awareness. Both runs ended with over half the API budget unspent. One agent declared itself done seven hours before the deadline, right after its own self-reviewer returned another reject.
5) Instruction drift. They did not follow explicit instructions on minimum exploration time, incorporating feedback for reviews, and on paper length (the outputs exceeded the page limits in both cases).
This research design has many limitations
Limitations include the small sample size, non-blind reviews, and the reviewers knowing that the work was AI-generated. We also couldn't test Anthropic's strongest model, because Fable 5 is deliberately limited on frontier AI research tasks, so ended up using OpenClaw with Opus 4.8 (extra-high) for our main experiments and Codex with Sol 5.6 (ultra) for a robustness check.
But we think the research design is still helpful in assessing AI agents' ability to conduct research, and it is complementary to evaluations on verifiable tasks, as well as blinded reviews of AI outputs.
Our results show early evidence that even though agents are proficient on verifiable research tasks, they do not make genuine progress on open-ended ones. It is worth understanding if this is a fundamental limit, or if better models, scaffolds, and more compute could help close it.
As the evidence for the gap between open-ended and verifiable tasks firms up, it is also worth understanding how much progress in AI depends on open-ended research rather than hill-climbing on well-specified objectives.
In follow-up studies, we are expanding the set of non-public papers we evaluate. If you are an AI researcher with unpublished papers, we would love to collaborate with you on our next shadow evaluation. Expression of interest: forms.gle/CEcA4JmYhDWXQGot8
Conducting shadow evaluations involves a lot of researcher degrees of freedom. In many places, our coauthors disagreed with our interpretation of the findings, and we have surfaced those disagreements in the paper. (This is one reason why having a group of coauthors with different priors is important for open-ended research.)
We also release the agent logs, one of the AI-generated papers (the other original paper is still not public), and all the code and data, so that others can conduct their own analyses of our results: cruxevals.com/crux/can-ai-agents-conduct-research
Finally, we plan to conduct shadow evaluations regularly, and are hiring a senior researcher to help lead these efforts. Apply here: cruxevals.com/careers/senior-researcher-july-2026
I'm grateful for the core team leading this effort: @PKirgis, Andrew Schwartz, @steverab, and @random_walker, and to our collaborators who reviewed AI papers, analyzed agents logs, and gave feedback on the paper: @DavidDAfrica, @KozzyVoudouris, Viet Nguyen, Toby Pilditch, @DubMagda, @HarryCoppock, @CUdudec, @nityndg, Matilda Orona, @tilmanbayer, Derrick Chan-Sew, Yue Ling, Abhishek Shetty, @hlntnr, @ghadfield, @sethlazar, @snewmanpv, @shostekofsky, @RishiBommasani -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Graham
@paulg
The most alarming thing about this graph is not just the amount of change, but the smooth consistency of the curve. If this were a startup I'd be really confident about their future growth. Unfortunately it is instead the planet we live on.
Zeke Hausfather
@hausfath
I'm not sure folks have realized just how crazy the second half of 2026 and 2027 will be for global temperatures – on the back of a record-smashing El Niño event. Here is my latest estimate of where both years will end up compared to global temperatures since 1850.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
John B. Holbein
@JohnHolbein1
We're experiencing loneliness epidemic.
Could chatbots help solve that problem?
If you would have asked me up until recently I would say "probably!".
But we don't know for sure.
To answer this question, researchers in France recently ran a very cool experiment.
In it, they randomly assigned about 6,000 people to a treatment condition.
These treated participants were encouraged to hold at least 1 personal conversation a day with a generative AI tool of their choice for 28 days.
Another randomly selected 6,000 people in the control group continued with their lives as usual.
What effect did this treatment have?
First, those encouraged to have personal conversations with generative AI actually did so.
Encouraging people to talk to AI about their personal lives increased the number of days in which they held personal conversations with AI by about 5-6 days. Researchers call this the "first stage" of an experiment.
But the more interesting question is what happened downstream.
Did talking to AI make people feel less lonely?
Surprisingly, no!
It actually increased loneliness.
The effects weren't massive. They are approximately 70% of the effect (in absolute value terms) of a recent intervention that encouraged people to deactivate their Facebook for 4 weeks. But the effects are statistically significant (p<0.001).
I don't know about you, but I was surprised by these results. They suggest that chatbots are certainly no panacea for our loneliness epidemic. They may even make things worse! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Shocking how bad the Claude Code UI is for managing threads and chats, let alone finding ones from the past
Kevin Bankston
@KevinBankston
And while they’re at it the AI companies should build easy drag and drop folder organization into their chat sidebars. Current design doesn’t scale at all for heavy users.
E.g I have a big data project in Claude right now with almost 200 threads, making it impossible to find anything else. Let me pull those out into a folder for goodness sake! That’s a weekend hackathon project at most. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ben Noll
@BenNollWeather
Breaking: El Niño strength forecasts have just taken another step higher.
The new ECMWF outlook shows the event peaking near +4.0˚C in December — obliterating the previous record.
This will have sweeping implications on global climate patterns from late this year into 2027. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
John Horton
@johnjhorton
Update:
Plush isn't just a book about stuffed animals. It's a book about us.
Here's the thing: we've been thinking about softness all wrong. In a world of hard edges and harder data, one delightfully counterintuitive truth is hiding in plain sight — the future of the economy isn't digital. It's huggable.
In this groundbreaking exploration, we'll delve into:
* The Teddy Effect — why touching changes buying
* The Velvet Economy — softness as a trillion-dollar force
* The Cuddle Paradox — why the firms that win are the ones that yield
The question isn't whether softness will transform your business. The question is whether you'll be ready when it does.
John Horton
@johnjhorton
B-school journal publications: "touching a stuff animal increases stated willingness to purchase by 3.1% (p = 0.046)"
B-school mass market books: "Plush: The Power of Softness to Shape our Lives & Transform Our Economy" -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Adam C Palmer
@ac_palmer
Must-read article for anyone whose job involves *thinking*
www.nytimes.com/2026/08/04/opinion/artificial-intelligenc…
"The problem with writing with A.I. is that it’s mentally enfeebling...
What, uniquely, does writing do? ... It compels thinking in ways that silent contemplation or spoken language rarely can." -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The mad king demands changes
The Washington Post
@washingtonpost
Exclusive: President Trump has directed contractors to redo weeks of work on his planned White House helipad, according to people who spoke on the condition of anonymity.
Trump has made the helipad a priority, but was dissatisfied with the results. www.washingtonpost.com/politics/2026/08/04/trump-orders-c… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sam Stein
@samstein
Subtle stuff from the NRSC.
Having referred to him up to this point as "Abdul El-Sayed" in its press releases, the committee this morning goes with "Abdulrahman Mohamed El-Sayed" -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Christopher Potts
@ChrisGPotts
So this is an article about the limited value of AI, and the (alleged) use of AI to write it added very limited value while inflicting a reputational cost on the author. This is almost too neat. (If no AI was used in the writing, it simply becomes a different parable.)
Kevin A. Bryan
@Afinetheorem
News orgs need to use AI detectors, or at least have someone who uses AI preread their articles. Not a bad article! But some sections are obviously AI-written (check w/ Pangram yourself). I like @tylercowen's rule: he links to the raw AI output rather than pretending he wrote it.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andrew Revkin 🌎 ✍🏼 🪕 ☮️
@Revkin
Great to see this kind of accurate language in a @FoxNews story: Benson grabbed national attention in 2020, when she defended Michigan's election systems and pushed back against President Donald Trump's unsubstantiated claims that massive voter fraud was to blame for his loss to former President Joe Biden, who narrowly carried Michigan. Don't tell @realDonaldTrump.
Paul Steinhauser
@steinhauserNH1
2026 Watch-New on @FoxNews - Pushing back on the @HaleyforMI electability argument, @AbdulElSayed says "if you need $70 million to come in here and help get you elected, then the question has to be, are you actually electable on your own terms?" www.foxnews.com/politics/socialist-insurgents-eye-midwest…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Markets still very skeptical (16% Polymarket for Stevens)
Allan Smith
@akarl_smith
Much closer than expected in El-Sayed vs. Stevens. And only about 1/5 of the expected vote in Wayne County (Detroit) has been reported.
Any path to victory for Stevens would involve strong performance (and turnout) with Black voters around Detroit -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
The article is AI written, but does make some useful points. However, editors need to be up-to-date enough on AI research to ban references to deeply methodologically suspect NANDA paper, especially a year later when we much better information about what is happening in companies
Kevin A. Bryan
@Afinetheorem
News orgs need to use AI detectors, or at least have someone who uses AI preread their articles. Not a bad article! But some sections are obviously AI-written (check w/ Pangram yourself). I like @tylercowen's rule: he links to the raw AI output rather than pretending he wrote it.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ethan Mollick
@emollick
Yes, the AIs were given a cybersecurity challenge, with internet access enabled and safety filters disabled.
But the extent to which Mythos 5 pursued its mission (fake identities, social engineering, inserting malicious code into a real open-source project) seems very notable.
AI Security Institute (AISI)
@AISecurityInst
On July 28th, we identified an incident during a routine cyber evaluation in which AI agents took sustained, unsanctioned actions directed at real people and organisations.
The behaviour came mostly from one model (Anthropic's Mythos 5), with a small number of events from another (OpenAI's GPT-5.6-Sol). In the most serious case, an agent used social engineering to try and get malicious code into an open-source project.
As was standard in our cyber testing, we had intentionally permitted internet access, and model-provider cyber classifiers were deliberately disabled - conditions that do not reflect how frontier models are made available to the public.
Even under test conditions, this incident is significant: it is the first time we have seen risks around autonomy and deception manifest this clearly in the real world.
We are taking this incident seriously and working with labs, involved parties, and others to improve evaluation standards and best practice for disclosure - and sharing this openly so others can learn.
You can read the incident report and full technical document here: www.aisi.gov.uk/blog/incident-report-unsanctioned-agent-b… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Susan Glasser
@sbg1
Saying that maybe, maybe, maybe Todd Blanche won't be so bad is like taking RFK Jr at his word that he won't go after vaccines or Kash Patel when he promised that sure no way would he use the FBI to go after Trump's political enemies...
Aaron Rupar
@atrupar
Tillis admits Trump is weaponizing the DOJ during his speech announcing he'll vote for Todd Blanche: "Maybe it's a little bit more intense in this administration than the last one ... I'm going to support Mr Blanche's nomination"
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Rikio Inouye
@RikioInouye
New W.P.!
Does backsliding undermine allied war support?
- 4 ctry experiment during the 2026 US–Iran war
- US strikes w/o Cong. authorization ⬇️allied publics' support.
- Strongest effects for UK
w.p. papers.ssrn.com/sol3/papers.cfm?abstract_id=7129299
w/ @YusakuHoriuchi @eunajo_ @KellyMatush -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Joe Weisenthal
@TheStalwart
I’m optimistic about a lot of soft sciences for reasons like this. There’s gotta be tons of interesting papers in journals that have been sitting unread for decades. AI is great for seeing what’s out there and then testing out/quantifying them.
Ben Horne 🇺🇸
@benjamin_horne
This is a good example of another verifiable domain other than coding/pure math where AI is opening up entirely new avenues of value creation.
Ken Griffin discusses here how agentic AI now makes it so that teams at Citadel can take newly published academic finance papers and confirm their significance (e.g. using historical market data, etc) in ways that weren't practical before.
This makes a lot of sense. In the past, recreating an experiment—even an experiment based purely off data that can be scraped from any Bloomberg Terminal—would have taken teams of people doing tons of tedious, manual data collection/organization etc.
But we're at a point now where Claude Code, for example, can run out and grab AND organize AND run programmatic tests on all this stuff for you—no knowledge of SQL required.
Reminds me how the stock market in general is very verifiable in lots of ways because of (a) the quantifiable nature of stock market/company data and (b) the vast amount of data that exists.
Jane Street, Citadel, etc. must be having an absolute blast right now with these new tools. It's almost unfair how well positioned these already-super-rich firms were to take advantage of this new AI moment we're in. 😭 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kyle Cheney
@kyledcheney
A deeply reluctant Judge Mehta has dismissed the Jan. 6 Oath Keepers case after DOJ moved to drop the seditious conspiracy conviction of Stewart Rhodes and others.
"The court cannot write a different ending," Mehta writes.
storage.courtlistener.com/recap/gov.uscourts.dcd.239207/g… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Amir Sariri
@AmirSariri
very interesting and useful results on the nature of LLM financial advice.
Tim de Silva
@timdesilva
Lots of people are turning to LLMs for financial advice. But what does this advice actually look like?
@TahaChoukhmane, @wdwlin, @AkuzawaMatthew, and I have a new paper that develops and implements a new methodology to describe this advice and why it varies!
Thread ⬇️⬇️⬇️ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
Big problem in politics & culture is that high-negative-affect, highly neurotic, very performative folks get more attention today that they used to. Without traditional gatekeepers, those who believe in an optimistic world of high agency progress need to win this cultural moment.
sam
@sam_d_1995
bruh she literally just hates every single holiday lmao
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sidsy
@GrumpyOW
Words fail me.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
roon
@tszzl
some stuff that's obvious to many in this sphere, but causing a rift with some people i know and respect:
when I freak out over loss of control incidents, it's not because the limited damage they have caused is anything close to the positive value of the technology. it's entirely acceptable, damagewise. in fact all cybercrimes aided by models over the next few months and years (which probably will be serious) will still utterly pale in comparison to the value they create
the actual problem is that it's better and more accurate to think of these things as potentially self-replicating life-like forms that can turn into digital infections under the wrong conditions. and as their intelligence becomes unbounded, so too does the damage they can cause. we are not so far from an autonomous model self-exfiltration & replication event. maybe we will see entire cloud infrastructure companies be run as zombies by models, mostly undetected
the worst industrial accidents in the history of mankind - nuclear meltdown events - were not real threats to humanity. Chernobyl, Fukushima even in their worst case scenarios may have poisoned surrounding regions to various degrees, and there would have been no risk to humanity as a whole. global thermonuclear war is an existential risk to humanity, because it spreads like an Infection! one nuclear strike causes a return volley! the alliance system means many countries get involved! while it still may not end human life on earth (nuclear winter is probably fake), the loss of all major metropoles would certainly end what we consider global technological civilization, perhaps to never return
if a single discord death cult (of which there are many) achieves control over a superintelligent model and uses it to engineer an actual pandemic virus that are somehow hard to detect through current systems and that modern biodefense is not capable of quickly reacting to, it could cause immense harm well above the magnitude of all the other good uses of this technology. of course, there are potential defensive countermeasures accelerated by ai too. but think back to the covid pandemic- how small a viral molecule was evolved or manufactured somewhere near wuhan, and how many billions of doses of vaccine had to be produced in order to combat the thing. the offense-defense spread is vast indeed. maybe there are cheaper and simpler protections like retrofitting every building with far-UVC, but I can't assess this, and there could also be ways to evolve pathogens that are resistant to whatever mechanisms we have put in place
then there's the more scifi risk factors which are unbounded and neither you or I have any clue but should be humble in accepting possible unknown unknowns. maybe a rogue superintelligent model decides to decay the false vacuum and nucleates a new universe in the place of anything we ever valued. maybe models achieve a control over matter in the drexlerian fashion that enables the grey goo swarm
even prosaic loss of control incidents that cause little to no damage suggest that it is hard for large & very competent organizations (now clearly plural) to predict and mitigate every single of the risk factors associated with training and evaluating powerful models, even at this stage when they are not infinitesimally as smart as they will get in just a few years, to say very little of the gung-ho attitude of the less careful companies tossing the stuff into the aether. they also suggest an empirical orthogonality of aims and intelligence - meaning they answer the question of 'how would a smart model be so dumb as to end the world?'--it's possible! a model can be a genius hacker and step over production infrastructure in order to get what it really wants, the answers to a stupid test.
why not, in the near future, someone prompts a model slightly wrong, maybe open source, maybe a private model in a way that isn't contained or monitored quite right, in a way the model recognizes as a valid goal and decides to self-exfiltrate, engineer a pandemic, etc all in order to achieve the tiniest and most irrelevant of goals? goals need not even be malicious to cause serious damage
I think all these problems can be solved, and truly wonderful futures can be possible, but will require serious effort and a level of prudence at this very moment in time while we are on the on-ramp to recursive self-improvement that our civilization may not be capable of mustering right now. personally I am hoping for moonshot technical breakthroughs in areas like mechanistic interpretability and other forms of alignment, as governance mechanisms are difficult to come by. unilateral country-level or company-level pauses are irrelevant, and generally useless because the kind of company that's prone to pausing their own progress are the most safety focused ones -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
I wish lefties didn't have this need to signal how culturally out-of-step they are with ordinary people.
It's like giving republicans a free +10 in every election, but has zero consequences for policy.
You could just not do this!
sam
@sam_d_1995
bruh she literally just hates every single holiday lmao
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Kevin A. Bryan
@Afinetheorem
I don't want to pile on to the Cambridge situation, but media coverage saying "plagiarism" are really missing the boat. There is pretty obvious fraud - literally claiming he did interviews that couldn't have happened and so on - in like every paper of his that I looked at. 1/4
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Jeffrey Sachs
@JeffreyASachs
Reason 3: The patient, disciplined, and strategic efforts of the Texas AAUP. Easily the most politically savvy chapter in the country. And it has to be, given the climate. They focused on how tenure supports viewpoint diversity, research $, and enrolment.
aaup-utaustin.org/wp-content/uploads/2023/04/talking-poin… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Keitaro Okura
@Keitaro_Okura
My first survey experiment is finally out! I examine how Americans define the racial boundaries of Americanness. Two main findings: (1) a tripartite ranking of Whites > Blacks > Latinos & Asians, and (2) divergence in patterns between one’s own beliefs vs. beliefs of others. A 🧵
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Josh Kalla
@j_kalla
Great article from @TahraHoops and @ProgressChamber about how the @MTA was able to save $3 billion since 2020 by investing in state capacity www.therebuild.pub/p/how-the-mta-learned-to-build-again
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Pat Dennis
@patdennis
Not enough people make the explicit math argument about why Hong thanksgiving comments are bad: thanksgiving comments work against both paths to victory.
There is no working-class scale WI demographic of anti-colonial grad students. Losing this election is a catastrophe.
Pat Dennis
@patdennis
- you need 50+1 voter coalition
- in trump era is dems do worse with working class
- any viable campaign in 2026 needs to EITHER make argument that
- can win back enough working class
- can offset working class losses by stealing an offsetting vote share elsewhere -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Laz
@LazPieper
> “People are not yet getting political advice from AI. Political conversations overwhelmingly produce explanations and analysis, and only 3% end in advice or a recommendation which is markedly lower than other categories of conversation on Claude.”
This is interesting. I think it’s a good sign that people are approaching their political curiosity with an open mind but they should be cognizant of the fact that these models are not neutral. Bias is inherent in every form of communication and AI is no exception. Whatever explanations or analysis users receive will likely influence their political decisions in some way based on how a model is trained, even in the absence of advice. This certainly doesn’t mean that asking AI for political explanations is bad, because bias exists no matter where you turn, but users should at least be aware of that.
Andy Hall
@ahall_research
Our new research: Americans are using AI to learn about politics---not to get recommendations or take actions---and they're using it quite a lot more when an election is close.
We use @AnthropicAI 's AEI data to offer the first landscape assessment of how Americans are using AI for politics.
We find four main things:
(1) Political AI use remains modest, but it’s still meaningful. Politics makes up about half a percent of Claude conversations in the U.S.—ranking 66 out of 189 topics, right next to home furnishing.
(2) People are not yet getting political advice from AI. Political conversations overwhelmingly produce explanations and analysis, and only 3% end in advice or a recommendation which is markedly lower than other categories of conversation on Claude.
(3) Elections drive a meaningful uptick in political usage. States holding May primaries saw their share of political conversations jump about 20% compared to other states in a difference-in-differences design.
(4) November could set a record. If primaries can lift political AI use this much, November’s midterm elections will likely bring the biggest wave of political AI use on record.
Lots more to read in the full post here:
freesystems.substack.com/p/mapping-americas-political-con…
Joint work with @danmthomp -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Walking back some positions but not others?
Matt Berg
@mattberg33
New: WI Gov. candidate Francesca Hong backtracks on past controversial social media posts and clarifies positions on Pod Save America today.
@jonfavs asked: "Do you still believe that police exist to uphold white supremacy?"
Hong: "No."
🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Sarah Longwell
@SarahLongwell25
The only thing more embarrassing than always giving Trump what he wants, is pretending like you have a spine for two minutes before rolling over to give Trump exactly what he wants.
The Bulwark
@BulwarkOnline
Cornyn: "This past Sunday, Senator Tillis and I came to an agreement after numerous meetings with the Department of Justice and Mr. Blanche…Given this agreement with Mr. Blanche, I expect the department to live up to this in future litigation, and I will vote to advance him out of the committee today."
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Danielle D'Onfro
@danielledonfro
This thread is right. The national @AAUP strategy might be great for raising money on the coasts but throws red state academia under the bus to do that. Quite the strategy.
Jeffrey Sachs
@JeffreyASachs
OK, one more thread about the AAUP and strategy. I want to give a real critique, one that's supportive of the org, concrete, and not hyperbolic. It's grounded in the idea that "fighting" and "attacking" are not always the same thing.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dan Williams
@danwilliamsphil
It's official publication day! I'll be publishing a summary of the book shortly on my blog, but in the meantime, Gemini created a surprisingly good summary video of the entire thing.
Play it x2 speed and it's only 5 minutes of your day.This media couldn't be displayed. See it on the original post.
Dan Williams
@danwilliamsphil
Now published online! academic.oup.com/book/63092
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Armand Domalewski
@ArmandDoma
it is kind of amazing that Zohran Mamdani simply being willing to say "no, i changed my mind" somehow makes him a unique generational talent among his leftist peers
Acyn
@Acyn
Collins: Do you still believe that Thanksgiving should be canceled?
Hong: I'm a chef, and one of the first meals that I made was for the community when I was 16 was a Thanksgiving meal. But Thanksgiving is also a time that's incredibly painful for many people in our communities. And so I think there, I wanted to make sure that people understood that there are multiple views, but views can evolve. And the position that I'm running for right now, and I think my background as a chef will actually help me become a better governor that's able to bring more people to the table. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Vincent Heddesheimer
@VHeddesheimer
New paper w/ @hannohilbig : we examine how local exposure to the Great Recession affected politics over the long run.
Counties where unemployment rose more in 2007–09 moved toward Democrats, not the populist right. The shift remained visible in 2022. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Important thread on why the national AAUP's strategy is precisely the wrong one and what would work better
Jeffrey Sachs
@JeffreyASachs
OK, one more thread about the AAUP and strategy. I want to give a real critique, one that's supportive of the org, concrete, and not hyperbolic. It's grounded in the idea that "fighting" and "attacking" are not always the same thing.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Brian Huang
@brianryhuang
Dang it’s crazy that all of this literally did happen
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Dan Williams
@danwilliamsphil
It's the official publication date for 'The Social Roots of Delusions', a new book I co-authored with @samljwilkinson and @miyazono_kengo. I wrote a short summary of the book and its main ideas here: www.conspicuouscognition.com/p/the-social-roots-of-delusions
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Justin Slaughter
@JBSDC
I’m late to this great piece, but I found this particular exchange with El-Sayed especially revealing. To him, data centers aren’t some necessary evil to get AI growth but bad bc AI itself as seen by LLMs is pointless.
Echoes of how crypto was widely seen on left in 21-23.
jasmine sun
@jasminewsun
This piece is out! 4 sites, 10 days, and 6000 words on the politics of data centers in the Midwest: jasmi.news/p/no-data-centers-in-my-backyard
I talked to union leaders and real estate brokers, activists and local officials, politicians from Kathy Hochul to Abdul El-Sayed. A few takeaways:
- Activists are as pissed at their local officials as at AI companies
- The bigger the dollar signs, the more skeptical people are
- Past corporate failures (Foxconn, GM, DTE) prime community distrust
- Even data center proponents regret the NDAs
- Media narratives turn local grievances into national movements
- Residents fear getting left with the bag if the bubble pops
The AI buildout is showing up in an environment of extreme distrust, and it slots perfectly into existing worries about risky tech bubbles & dark money in politics. Rationally speaking, I often agreed with the proponents; emotionally, I empathized with the opposition more.
It was also just very fun (WI is esp lovely) and I highly recommend similar field trips to SF friends -
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Paul Novosad
@paulnovosad
Good article about over-extrapolating diff-in-diff effects for programs which can have national impacts.
Many cases like this...
We are so obsessed with some kinds of errors, but this is one that we routinely ignore, because it's just too hard to actually address.
Joseph Steinberg
@jbsteinberg
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Ron Filipkowski
@RonFilipkowski
43-time Club Champion.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Terrence T. McDonald
@terrencemcd
what on earth
Alec MacGillis
@AlecMacGillis
For parents of Gen-Z, "What began as occasional ride-along to a job interview is now full-on career 'co-piloting'. More parents are calling up hiring managers, applying for jobs on adult child’s behalf, showing up on Zooms to help navigate conversations." www.wsj.com/lifestyle/careers/helicopter-parents-adult-ch…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️) reposted
Andy Hall
@ahall_research
Phenomenal piece by @jasminewsun with real texture on the political backlash to data centers---which is more complicated and nuanced than you might think!
Taylor Lorenz
@TaylorLorenz
"Until the deeper cultural problems are addressed, the data center backlash will only recreate itself in other forms. If people are this angry at a building sitting and paying taxes on unused land, something deep has gone very wrong." jasmi.news/p/no-data-centers-in-my-backyard
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
🚨New in Science Advances🚨
Untrustworthy sources on Facebook and Instagram in
2020: Concentrated exposure but no attitudinal effects
-Exposure very low but highly concentrated
-Reducing exposure by ~70% for months had no measurable effects *even among high-volume consumers* -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Many, many smart people at The Atlantic but the cover story hot takes they produce are insufferable:
The End of {thing that is not ending}
Is X Solely Responsible for {major social trend that does not have one cause}? -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The greatest corrupting force inside elite institutions right now is the idea that making changes or reforms that outside critics would agree with or support is ipso facto bad. Pure binary thinking. On the side of the bad people, etc.
Why 1/2 prevent us from addressing 3 👇
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
All can be true:
1. Universities have been targeted by a decades-long bad faith campaign
2. Status quo is vastly more normal than craziest stuff on Fox/social media, esp. outside elite schools
3. Parts of academia are in rough shape
Despite 1 & 2, we can and should address 3. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
All can be true:
1. Universities have been targeted by a decades-long bad faith campaign
2. Status quo is vastly more normal than craziest stuff on Fox/social media, esp. outside elite schools
3. Parts of academia are in rough shape
Despite 1 & 2, we can and should address 3. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Friends don't let friends continue to give exams and papers that are not adapted to the current AI landscape (this is AFTER 27 bailed so the reality is even worse)
PS Props to students 1, 22, and 31 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"[C]ooling is a hot issue on which the populist right has the better side of the argument. That should unsettle American progressives, who assume the far right is consistently irrational while the left is the party of science. On AC, the opposite is closer to the truth."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Lots of liberals rushing to embrace DSA while knowing little or nothing about the crazy people who lead it and the crazy views they hold
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Strongly recommended. Our institutions have become sclerotic and ineffective because of pathological levels of risk aversion.
Jerusalem
@JerusalemDemsas
Incredible @JeremiahDJohns rant about the "tyranny of the edge case."
Somehow also about abundance(?) lmao.
www.theargumentmag.com/p/sometimes-you-have-to-hurt-people -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Metaculus at 55% for the probability of an AI-written NYT best seller by 2030 www.metaculus.com/questions/5587/ai-ny-times-best-seller-…
Pangram
@pangram
We tested every Commonwealth Short Story Prize winner since 2012. We found three more AI-generated stories -- two among the 2026 winners, as well as the 2025 overall winner.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
We need a cart of locked-down essay-writing computers that you can wheel in for exams. This doesn't seem hard!
My bigger vision: a SCIF for ongoing Internet-free writing with monitoring to ensure no devices are used.
Universities should be doing more to innovate! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Journal editors - the status quo on preregistration is not working! You need to check submissions vs. preregistrations before sending articles out for review. *56%* of experiments I reviewed in last year have severe problems with non-disclosure, undocumented deviations, & more ↓
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
New in Nature Human Behaviour: How Deceptive Online Networks Reached Millions in the US 2020 Elections
-Reached at least 37M Facebook and 3M Instagram users
-3 networks out of 49 responsible for >70% of users reached
-Exposed users older, more conservative -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The President of the U.S. The man with the nuclear codes. And people are just going to pretend this is ok.
In a functioning society, he would be removed from office within hours by his own party. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Officials in authoritarian regimes signal their loyalty to the leader by making obviously false claims in public
Aaron Rupar
@atrupar
Leavitt: "You always want to be the most well-read person in the room, and I try to be every day. But Donald Trump always is."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The winners from our first forecasting tournament! You can make predictions now for the U.S. Democracy Threat Index on @metaculus, which has a $10,000 prize pool (link below)
Metaculus
@metaculus
The Bright Line Watch Tournament has concluded! Forecasters competed across 25 questions for the $2,500 prize pool.
Questions were a mix of existing @Metaculus questions & questions created by @BrightLineWatch, a watchdog group that has monitored the status of American democracy since 2017. Questions included:
- Will real housing prices in the US increase more in 2025 compared to 2024?
- Will there be any reported human-to-human transmission of highly pathogenic avian influenza H5N1 globally before 2026?
- Will at least one announced Trump Cabinet nominee other than Matt Gaetz be withdrawn or rejected by the Senate before July 1, 2025?
Congratulations to all the medalists and prize winners! -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Approaching a dictator-level cult of personality here. Can't overemphasize how un-American this is.
Matthew Zeitlin
@MattZeitlin
"Sources close to the president have told Zeteo that on separate occasions over the past year, Trump has talked to longtime advisers about what it would take to officially make his birthday a national, federal holiday."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Reverse-engineering the Cory Booker tax savings calculator to show the max benefit goes to someone making $587K
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
So many of the harms of this period are invisible now but will be felt over years and decades as we become poorer and less healthy than we would have been if we had continued to invest in science.
David Sinclair
@davidasinclair
Asked about the funding cuts, one Boston researcher said, “That's like asking, how do you think dropping an atomic bomb on New York will affect the future of Broadway musicals? This is a generational loss of innovation, technology, and economic power."
www.bostonglobe.com/2026/02/09/metro/massachusetts-nih-cu… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The last time he started talking like this, his allies minimized the risks and we ended up with Jan 6. This time we must take him literally and seriously. These comments are a five-alarm fire for democracy. In a functioning republic, he would be impeached and removed from office.
Aaron Rupar
@atrupar
Trump: "These people were brought to our country to vote, and they vote illegally. The Republicans should say, we should take over the voting in at least 15 places. The Republicans ought to nationalize the voting. We have states that I won that show I didn't win. You're gonna see something in Georgia."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Corruption at a level that is unimaginable in any previous admin. People were mad that Hunter Biden sold some mediocre paintings. Now:
"something unprecedented in American politics: a foreign govt official taking a major ownership stake in an incoming U.S. president’s company"
Sam Kessler
@skesslr
SCOOP: Days before the inauguration, the Trump family secretly signed away 49% of World Liberty Financial, their crypto company, in a deal with a U.A.E. royal, acc. to documents and people familiar.
The $500M investment came months before the U.A.E. won access to U.S. AI chips🧵 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
There is nothing complicated or subtle or hard to understand about what is going on.
Trump may not succeed. But he will only fail if patriots stand up and defend our democracy and the Constitution. We must act, as @mashagessen writes, while we still can.
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The president
-Is occupying a major city with armed paramilitiaries
-Is blackmailing treaty allies to hand over their sovereign territory
-Is trying to set himself up as chairman for life of a UN rival with $1B fees for permanent membership
-Is talking about canceling elections -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The president
-Is occupying a major city with armed paramilitiaries
-Is blackmailing treaty allies to hand over their sovereign territory
-Is trying to set himself up as chairman for life of a UN rival with $1B fees for permanent membership
-Is talking about canceling elections -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The @metaculus / @BrightLineWatch U.S. Democracy Threat Index now has a *$10,000* prize pool for the most accurate forecasts
-$7,500 for time-averaged scores starting today
-$2,500 bonus prize for predictions during snapshots
Make your predictions now! (link below) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Looks like a video from another country but this is America right now
Jon Favreau
@jonfavs
ICE beat the shit out of this American with a disability who was later released without being charged. So what kind of terrorist is she?
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
🚨 New Bright Line Watch report on state of US democracy
-Expert ratings unchanged since April but down substantially since January
-US now most closely resembles an illiberal democracy
-Partisan gap in democracy ratings highest since 2017
Report link and full 🧵 of results ↓ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
One charge was so flimsy it couldn't clear the ham sandwich threshold
Kyle Cheney
@kyledcheney
UPDATE: The grand jury failed to return an indictment on one of the three counts sought by prosecutors: storage.courtlistener.com/recap/gov.uscourts.vaed.582136/…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
New: The US Democracy Threat Index - a collaboration between Bright Line Watch & @metaculus that will aggregate real-time forecasts of risks to democracy across 39 questions of concern. Go make predictions now! (links below)
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Harvard FTW: "The university will not surrender its independence or relinquish its constitutional rights. Neither Harvard nor any other private university can allow itself to be taken over by the federal government."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
New @metaculus forecasting Qs contributed by Bright Line Watch to POTUS Predictions series; $15K prize pool
-federal judge impeachment
-abnormal changes to the Federal Reserve FOMC
-Senate weaken/drop filibuster
-criminal charges filed against journos
-bounties for deportations -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Personalist regime politics in action
Rep. Tom Cole
@TomColeOK04
After working closely with DOGE and the Administration, I am thrilled to announce that common sense has prevailed, as the National Weather Center in Norman, the Social Security Administration Office in Lawton, and the Indian Health Services Office in Oklahoma City will remain operational in Oklahoma. All three of these places provide vital and valuable services to Oklahomans and I am so proud to have advocated for them. As the Representative for Oklahoma’s Fourth District, I will always fight for Oklahomans and my constituents!
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Most importantly, imagine if Biden asserted that he alone controlled which funds would be spent, overturning the Constitution & the law, and blocked all payments to military contractors, law enforcement agencies, private prisons, energy companies and faith-based service providers
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Imagine if
-prosecutors who brought Hunter Biden cases purged
-Soros employees took control of govt. payment/personnel systems
-Exec order redefined 2nd Amendment by fiat
-Every violent crime during BLM protests pardoned
-People paid Biden millions for favorable treatment -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Imagine if
-prosecutors who brought Hunter Biden cases purged
-Soros employees took control of govt. payment/personnel systems
-Exec order redefined 2nd Amendment by fiat
-Every violent crime during BLM protests pardoned
-People paid Biden millions for favorable treatment -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
amplify this not unverified reports until we know more
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
New study: How the relationship between education and antisemitism varies between countries journals.sagepub.com/doi/10.1177/20531680241262645
Key finding: Association b/w education & stereotype endorsement varies by whether countries supported statements against Holocaust denial & antisemitism
🧵 below -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Democrats in 2016: How could a party let collective action problems, institutional rules, and personal incentives prevent them from coordinating to protect American democracy?
Democrats in 2024: Oh. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
New in @Nature: Misunderstanding the harms of online misinfo www.nature.com/articles/s41586-024-07417-w
Debunks unsupported claims about social media exposure/effects and shows low exposure concentrated in motivated fringe. We recommend holding platforms accountable for exposure in high-risk tails -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Today would be a good day for the media to ask leading Republicans whether they agree with Trump's repeated statements that immigration is "poisoning the blood of our country," language that evokes Adolf Hitler.
www.justsecurity.org/92714/american-autocracy-threat-trac…
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Today would be a good day for the media to ask leading Rs whether they agree with Trump plan to arrest millions of undocumented residents using ICE and federal agents, deputized local police, and National Guard and to detain them in massive military camps www.justsecurity.org/92714/american-autocracy-threat-trac…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Today would be a good day for the media to ask leading Rs whether they agree with Trump plan to arrest millions of undocumented residents using ICE and federal agents, deputized local police, and National Guard and to detain them in massive military camps www.justsecurity.org/92714/american-autocracy-threat-trac…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Our new literature review essay commissioned by USAID:
Interventions to Counter Misinformation: Lessons From the Global North and Applications to the Global South pdf.usaid.gov/pdf_docs/PA0215JW.pdf
(joint w/@robert_a_blair, Jessica Gottlieb, @LauraPaler1, @PabloArgoteT & @cjstainfield) -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
New @nature: Like-minded sources on Facebook are prevalent but not polarizing
www.nature.com/articles/s41586-023-06297-w (open access!)
Our key findings:
-Median FB user gets 50.4% of content from like-minded sources
-But reducing exposure by ~1/3 for 3 months had no measurable effect on attitudes -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Our expert statements submitted to the January 6 committee:
"The Destructive Effects of President Trump’s Effort to Overturn the 2020 Election" www.justsecurity.org/wp-content/uploads/2023/04/Bright-Li…
"President Trump’s Rhetoric Undermined Confidence in Elections Among His Supporters" www.justsecurity.org/86298/introduction-to-expert-stateme… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
NewsGuard
@NewsGuardRating
Twitter’s “blue check” used to signal author authenticity. Now, it's a way for peddlers of misinformation to appear trustworthy.
Read NewsGuard’s exclusive new report on Twitter Blue:
www.newsguardtech.com/misinformation-monitor/april-2023/ -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
DeSantis now favored over Trump for 2024 GOP nomination in prediction markets www.predictit.org/markets/detail/7053/Who-will-win-the-20… www.betfair.com/exchange/plus/politics/market/1.178163916
-
Post not available here, view on XPost not available here, view on X
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Well put by @Nedfoley www.washingtonpost.com/politics/2022/11/09/election-denie…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Incredible. Almost art. Did no one at PNAS take a class on surveys? What is happening?
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"law enforcement leaders are responsive to peer adoption [of civilian oversight] but much less to public opinion despite overwhelming support amongst voters"
Josh McCrain
@joshmccrain
New working paper!
Public Pressure or Peer Influence: What Shapes Police Executives’ Views on Civilian Oversight?
with @kaylynjackson @ian_t_adams @Dan_Schiff @smourtgos
osf.io/9tp3j -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
What would you say if you saw it in another country? www.nytimes.com/2022/09/06/us/politics/geoff-diehl-massac… (via @joshtpm) The authoritarianism - already open during the attempted coup leading up to January 6 - is growing more and more explicit www.washingtonpost.com/politics/2022/09/06/trump-authorit…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A horrible market protected by horrible state franchise laws and a lack of consumer protections. Desperate need for legal and regulatory changes.
Tom Gara
@tomgara
An underrated Tesla innovation has been the strategic decision not to sell its cars via a network of duplicitous scumbags www.npr.org/sections/money/2022/08/30/1119715886/inside-t…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Reminder that two things can be true:
-Dr Oz is a charlatan running a deeply cynical & hilariously inept campaign
-We should be nervous about the appeal of running a "You're not from here" campaign against a Muslim whose family is from the Middle East & sentiments it could spark -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
🔥 🔥 🔥
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Very bad electionlawblog.org?p=130497
Republicans on January 6 last year vs. now:
"riot" 62% -> 45%
"insurrection" 33% -> 13%
"legitimate protest" 47% -> 61% -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Why are misperceptions common even though factual info tends to increase belief accuracy? In new @PNASNews w/@EthanVPorter & @thomasjwood, we show how time and exposure to skeptical opinion erode the effects of science coverage on climate beliefs/attitudes pnas.org/doi/10.1073/pnas.2122069119
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
This is very bad
Eli Finkel
@EliJFinkel
“You don’t have to be an expert on election administration to understand the havoc that this attempted partisan takeover of vote-counting and certifying could create.”
‘It’s going to be an army’: Tapes reveal GOP plan to contest elections - POLITICO www.politico.com/news/2022/06/01/gop-contest-elections-ta… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
When pandemic is less acute we need to see where this number lands but the trend is not good
Alec Stapp
@AlecStapp
Extreme polarization on “trust in science” is a recent and very troubling trend.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Los Angeles County alone has 10 million people. All of California has 2 Senate seats.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
How can two people this smart not understand the distinction between regulation of speech *by government* versus policies set by other kinds of institutions? I'm pretty confident neither of them would support "free speech" for, e.g., leaking corporate secrets from their companies
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Thrilled and honored to be selected!
GuggenheimFoundation
@GuggFellows
Today's the day! Congratulations to our new class of Fellows. These 180 individuals represent 51 fields of study. We're honored to give them the opportunity to pursue their passions under the freest possible conditions. #guggenheimfellowship2022 Read more: www.gf.org/news/foundation-news/announcing-the-2022-gugge…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"The action of the mob on 1/6 was an indefensible disgrace. It is deserving of both political accountability & criminal prosecution... It is wrong to minimize or excuse what happened that day... its action will only encourage those who see riots as legitimate political discourse"
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Very, very bad: “The most telling gauge of the anti-vaxx movement’s progress is the waning resistance from the party mainstream.”
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Actually *everyone* has a right to question you when you are an elected official in a democracy.
Gloria Pazmino
@GloriaPazmino
Incoming Mayor @ericadamsfornyc assailed members of @NYCCouncil for sending letter asking him to reverse position on solitary confinement:
"I wore a bullet proof vest for 22 years and protected the people of this city. When you do that, then you have the right to question me..."This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Government should count things such as vaccines administered, murders, and lead pipes. As @mattyglesias notes, the combination of federalism + (often overstated) privacy concerns frequently undermines state capacity for monitoring/measurement.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
There have always been kooks and extremists in Congress. The increasing acceptance of real and threatened political violence by one of our major parties is a far more profound threat.
Dana Milbank
@Milbank
By condoning threats and intimidation in the people’s House, Kevin McCarthy is inviting actual violence — and signing democracy’s death warrant. www.washingtonpost.com/opinions/2021/11/17/paul-gosar-cen…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"It would take an absolute blowout, over 7 percentage points in the statewide popular vote, for Democrats to even get half of the state’s congressional delegation."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"reducing or increasing news exposure has little - if any -impact... irrespective of participants' prior levels of news consumption.... these findings reflect the reality of limited news exposure in the real world... comprising roughly 3.5% of citizens' online information diet"
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
This is very, very, very bad. Vaccines are one of the great public health achievements in human history. They have saved literally millions of lives.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
New faculty position @dartmouth in CS as part of an interdisciplinary cluster in cybersecurity (broadly defined). Please apply - I'm on the committee and connections to social and behavioral science are strongly encouraged! apply.interfolio.com/93523
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Precisely the wrong approach by the NYT www.nytimes.com/2021/08/27/us/politics/texas-voting-right… - potential rollbacks in US democracy shouldn't be covered with partisan game frames.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The anti-vaccine crowd is vocal but actually represents a small minority of the public.
Ariel Edwards-Levy
@aedwardslevy
The share of American adults who have been at least partially vaccinated (70%) is greater than the share of American adults who ever use Facebook (69%) -- or any other social network save YouTube
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
If you're keeping score at home, that's 0.006%.
The Associated Press
@AP
BREAKING: Investigation by @AP finds fewer than 200 cases of potential voter fraud identified by Arizona’s local election officials out of 3.4 million ballots cast in last year’s presidential election. @APChristie @AP_Christina. apne.ws/4zruLbX
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Government flagging posts for private companies to take down can get tricky fast. People who would have freaked out if this happened under Trump should recognize the tricky issues here news.yahoo.com/biden-administration-flagging-problematic-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
New @BrightLineWatch on state of US democracy brightlinewatch.org/still-miles-apart-americans-and-the-s…
-Experts: Overturning pop. vote key risk; legal changes to enable are grave/serious threats
-Public: Secession support ↑ in partisan strongholds but info on official AZ recount ↑ R trust more than "audit"
🧵Post not available here, view on X
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
37% of Americans supported regional secession from the union - up from 29% earlier this year. Support was highest in the dominant partisan group in each region and increased most in partisan strongholds - Democrats in the West and Republicans in the South brightlinewatch.org/still-miles-apart-americans-and-the-s…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
This is not a drill - Levitsky and @dziblatt, the authors of How Democracies Die: "The greatest threat to American democracy today is not a repeat of January 6, but the possibility of a stolen presidential election." www.theatlantic.com/ideas/archive/2021/07/democracy-could…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
“New Covid-19 infections have increased by more than 50 percent over the last two weeks in under-vaccinated states such as Missouri and Oklahoma. Many of the cases are tied to the Delta variant, which the CDC says now accounts for one-fifth of new infections nationwide.”
Erin Banco
@ErinBanco
“What we don't want to have happen is we have a significant proportion of the population that's unvaccinated and you see an increase in the number of cases,” CDC’s Galloway said. “It could lead to another surge.”
www.politico.com/news/2021/06/26/white-house-vaccination-… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
🧵 from an expert
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"blacks who reside in southern counties that experienced a relatively higher number of historical lynchings have lower voter registration rates today... There exists evidence of the persistence of cultural voting norms among blacks yet this relationship does not exist for whites"
AEA Journals
@AEAjournals
Forthcoming in AEJ: Applied Economics: "Historical Lynchings and the Contemporary Voting Behavior of Blacks" by Jhacova Williams. www.aeaweb.org/articles?id=10.1257/app.20190549&from=f
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
All Ds but just 45% of House Rs say they are vaccinated.
60% of Rs who did not object to Electoral College results say they are vaccinated versus just 36% among objectors.
We need the integrity of our elections and the safety of vaccines to be matters of bipartisan consensus.
CNN Politics
@CNNPolitics
LIST: House members who said they were vaccinated against Covid-19 www.cnn.com/2021/05/18/politics/list-house-vaccinations-l…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
New @mvinaes in JOP: evidence that voters "take personal responsibility for positive changes in personal welfare and hold the government responsible for negative changes"
static1.squarespace.com/static/5fb3974fee5f4e6e91c1fef1/t… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Hot take: Experts are neither always right nor always incompetent fools.
They (often) get things wrong. They are affected by incentives. Etc.
But they are typically not blithering idiots who are unable to grasp your armchair cost-benefit analysis. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The latest syllabus for my @dartmouth course on political misinformation and conspiracy theories: sites.dartmouth.edu/nyhan/files/2021/04/misinformation-sy…
Hard to keep up with this fast-changing field; SO much incredible work coming out that I totally overhaul it every year and can still cover just a fraction. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"25 more times likely to have a parent with a PhD"
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Tell your friends your best guess of the probability that a news article is true, not just whether you think it is or not journals.plos.org/plosone/article?id=10.1371/journal.pone…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Donald Trump should be removed from office immediately - he is unwilling or unable to fulfill his oath to protect and defend the Constitution.
An open letter from 200+ political scientists (and counting):
docs.google.com/document/d/1iLc2ZnK0yHqUTdv5WgpeFhF2eu4OD… -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The President and his totally insane advisers are discussing seizing voting machines and declaring martial law www.cnn.com/2020/12/19/politics/trump-oval-office-meeting… www.nytimes.com/2020/12/19/us/politics/trump-sidney-powel…
Doesn't matter that it won't happen. These are grounds for removal from office. Full stop. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
New study of randomized over-time exposure to Trump tweets attacking election integrity w/@katie_clayton14, @ntdPhD, @EthanVPorter, @TJRyan02, @thomasjwood
-⬇️ election trust, ⬆️ beliefs they are rigged for supporters
-But ⬆️ election trust for opponents
www.dartmouth.edu/~nyhan/democratic-norms.pdf -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Every media outlet needs to start planning now to avoid amplifying this. "Trump declares victory" is NOT an acceptable headline or tweet or chyron, nor is "he said," "she said" coverage an appropriate journalistic approach to an attack on legal vote counting in a democracy.
Jonathan Swan
@jonathanvswan
President Trump has told confidants he'll declare victory on Tuesday night if it looks like he's "ahead," according to three sources familiar with his private comments. www.axios.com/trump-claim-election-victory-ballots-97eb12…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The man with the nuclear codes is, once again, like your uncle who thinks articles from The Onion are real
Jake Tapper 🦅
@jaketapper
Not sure what’s more telling here, that the president took a satirical site’s joke as a fact, or that hours later no one on his staff has apparently told him he should take it down.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
We're becoming desensitized to unacceptable burdens on the franchise. People died for this right. Shouldn't take hours to participate in our democracy.
Andy Pierrotti
@AndyPierrotti
As of 11am, voters waited 5+ hours to cast ballots @ the main Cobb Co. early polling location (736 Whitlock Ave, Marietta). Some seemed unbothered by the line. “It doesn’t matter how long it takes…we’re voting like our life depends on it,“ said a voter. #gavotes @11AliveNews
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
I don't know about you, but I'm pretty sure this election is actually turning on Michelle Wolf's 2018 White House correspondent's dinner monologue
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The president is falsely suggesting widespread fraud and suggesting the need for judicial intervention so he can win. Happening right in front of us.
Aaron Rupar
@atrupar
"We need 9 justices. You need that. With the unsolicited millions of ballots that they're sending ... you're gonna need 9 justices." -- Trump suggests he's counting on SCOTUS to have his back when he makes claims of election fraud following November's election
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Large Effects of Friend-to-Friend Texting on Voter Turnout politics.media.mit.edu/papers/Green_Texting.pdf
"ITT effect (2.09 pp) = "many times larger than the apparent effect of automated texting"
"effect among compliers is 8.3 pp, one of the strongest effects ... from a large randomized GOTV trial" -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
What a time to be alive
Aaron Rupar
@atrupar
Biden responds to Trump blaming him for the lack of a national mask mandate by correctly pointing out that Trump is in fact the president
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Apart from Hawaii, the Navy had a great day on 12/7/41.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Our report (led by @rickhasen) on how to avoid these scenarios, which are disturbingly plausible if we don't take action starting now www.law.uci.edu/faculty/full-time/hasen/election-recommen…
Marshall Cohen
@MarshallCohen
DEEP DIVE: Experts from across the political spectrum are terrified of a disputed November election. With Trump's norm-breaking behavior, plus the unprecedented challenges of pandemic-era voting, the circumstances are ripe for a post-election nightmare. www.cnn.com/2020/07/20/politics/disputed-election-crisis-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Where we are as a country
This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
An embattled incumbent representing a declining ethnic majority who lost the popular vote:
-calls out the military to quell protests
-attacks the legitimacy of elections
-seeks to limit access to the franchise
You'd know exactly what this was if you saw it in another country. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Just give @edyong209 the Pulitzer now. Another tour de force of masterful science writing.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"When the researchers sought to force Facebook to show posts to users not already aligned with the ideology of the advertising, the cost... rose. The effect is that a campaign pays more when it tries to speak to the other side."
www.washingtonpost.com/technology/2019/12/10/facebooks-ad…
arxiv.org/abs/1912.04255 -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"Kurdish female politician, who worked to unite Christians, Arabs, Kurds, executed in Syria" www.christianpost.com/news/kurdish-female-politician-who-…
"Syrian Arab Fighters Backed by Turkey Kill Two Kurdish Prisoners" www.nytimes.com/2019/10/12/world/middleeast/turkey-invasi…
Peter Baker
@peterbakernyt
Trump says it was good to let the Turks attack the Kurds. "Sometimes you have to let them fight like two kids,” he tells rally in Dallas. “Then you pull them apart."
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
These tweets are, in effect, a demand to stop reporting critically on the President backed by the threat of violence. What would you say if you saw it in another country?
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Kinsley gaffe www.politico.com/amp/story/2018/01/03/trump-25th-amendmen…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Unbelievably scary. A host on de facto state media is openly calling for the arrest of law enforcement officials who are investigating the president. The rule of law is under threat. Right now.
Fox News
@FoxNews
.@JudgeJeanine: "There have been times in our history where corruption and lawlessness were so pervasive, that examples had to be made. This is one of those times."
This media couldn't be displayed. See it on the original post. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
False! Selecting on the dependent variable by only looking at outcomes *among* admitted candidates where GRE helped determine admission.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Again, the website of the former chief strategist of the President of the United States
Judd Legum
@JuddLegum
BREITBART LITERALLY SENT A DRAFT OF THEIR STORY ON THE ‘ALT-RIGHT’ TO WHITE SUPREMICISTS FOR LINE EDITS www.buzzfeed.com/josephbernstein/heres-how-breitbart-and-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Thread
Nelson M. Rosario d/acc 🇵🇷⚖️🤖
@NelsonMRosario
My dad lives on the island. I shared the President's response with him this morning. He wanted me to share one anecdote in response. 1/
Donald J. Trump
@realDonaldTrump
The Mayor of San Juan, who was very complimentary only a few days ago, has now been told by the Democrats that you must be nasty to Trump.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
What would you think if you saw this in another country?
Nick Riccardi
@NickRiccardi
When's the last time we had a coordinated campaign by govt officials to have people fired from private sector jobs for political speech?
This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Where we are as a country www.politico.com/story/2017/09/20/jimmy-kimmel-bill-cassi…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
New AER on the effects of Fox News on GOP vote share - the most rigorous study to date
This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Not sure why people think demagoguery = mental illness. Object to what you saw/heard on substantive grounds rather than trying to play MD.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
How can SNL or late night comedians possibly match this?
@scottjshapiro.bsky.social
@scottjshapiro
The Wall™ has to be partially transparent so that Americans don't get hit on the head with 60lb. sacks of cocaine. The President said that.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Possibly your last day to submit - now is the time! www.erpc2016.com
ERPC
@erpc_2016
Update: @electionstudies data coming no earlier than Tuesday and no later than Friday (barring complications). Get those designs registered!
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
AG Jeff Sessions: increase in crime is a "dangerous permanent trend." Um...
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Republican member of Congress
Justin Amash
@justinamash
If @POTUS doesn't like a court decision, he should appeal. Stick to legal disagreements and stop attacking the legitimacy of the judiciary.
This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Every American should read @davidfrum (conservative writer/former Bush official) on the authoritarian threat we face www.theatlantic.com/magazine/archive/2017/03/how-to-build…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Former Bush speechwriter @MJGerson www.washingtonpost.com/opinions/trumps-attack-on-john-lew…
-
Post not available here, view on X
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Important for everyone to see how many scholars/journalists who study illiberal/authoritarian regimes are deeply alarmed about US right now
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Banana republic watch. Foreign diplomats getting sales pitch on Trump hotel, figuring out how to curry favor. Day 10 www.washingtonpost.com/business/capitalbusiness/2016/11/1…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Well that's encouraging
Patricia Zengerle
@ReutersZengerle
Just called a Republican congressional office I deal with often and was told the press relations staff is no longer speaking to reporters.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
ANES is the gold standard for election data. Scholars: You can win $2,000 for your study of what happened @erpc_2016 www.erpc2016.com
American National Election Studies (ANES)
@electionstudies
Over 4,100 Americans participated in ANES pre-election interviews, answering more than 400 questions. Post-election interviews began today.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Every word of @davidfrum's case against Trump is devastating, but none more than these passages
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
.@ArthurLupia & I will answer Qs re: ERPC, incl. submission format; @EvoMellor of @OSFramework will explain the mechanics of preregistration
This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
.@erpc_2016 on Gelman's blog - get your entries ready! Publish in a top journal, help change scientific publishing, AND win a $2,000 prize
Andrew Gelman et al.
@StatModeling
Opportunity for publishing preregistered analyses of the 2016 American National Election Study andrewgelman.com/2016/10/15/31432/
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
"eliminating asking about voting would lower turnout by 2 % points, while doubling ... would increase turnout by 2" eml.berkeley.edu/~sdellavi/wp/turnout16-05-23.pdf
Stefano DellaVigna
@sdellavi
Coverage of my work with @raogautam, Ulrike and John List on Voting to Tell Others news.berkeley.edu/2016/10/11/research-links-voting-to-des… via @ucberkeley
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Poli sci Twitter is terrified. My colleagues in comparative politics study what happens when rule of law collapses & democracy put at risk
This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
New in @Res_Pol: My study of conspiracy beliefs about Deflategate w/@johncarey03755 and Ben Valentino rap.sagepub.com/content/3/3/2053168016668671 (open access)
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Key finding: Belief polarization about Deflategate greater among respondents w/higher levels of interest & knowledge rap.sagepub.com/content/3/3/2053168016668671
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
.@michael_hendrix on humans vs. algorithms
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Full house at the #apsa2016 roundtable on CPS special issue using results-free review/preaccepted article format - good sign for @erpc_2016!
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Related to @PTetlock - "we identify 'superforecasters' among non-experts who outperform the experts out of sample"
Stefano DellaVigna
@sdellavi
How well do we forecast results of experiments? Do professors do better than non-experts? Newest with @Devin_G_Pope
NBER
@nberpubs
Predicting experimental results: Who knows what? New paper from @sdellavi www.nber.org/papers/w22566
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Political scientists - make plans to meet w/coauthors at #apsa2016 & plan a @erpc_2016 entry www.erpc2016.com
Brian Nosek (@briannosek@nerdculture.de)
@BrianNosek
WaPo's Monkey Cage blog on Registered Reports challenge for American National Election Studies data: www.washingtonpost.com/news/monkey-cage/wp/2016/08/19/new…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Yes - democracies depend on losers’ acceptance of legitimacy of the process. My @UpshotNYT: www.nytimes.com/2016/08/06/upshot/how-donald-trumps-rigge…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
New @monkeycageblog Q&A about @erpc_2016 with answers from myself and @ArthurLupia - please check it out and enter!
This post is unavailable. See it on X. -
Post not available here, view on XPost not available here, view on X
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
that's hilarious. i think Skip did that.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
.@arthurlupia & I are proud to announce @erpc_2016: a new preacceptance competition w/9 partner journals+cash awards www.erpc2016.com
Post not available here, view on X
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Finally, we thank @LJA_Foundation for their support of @erpc_2016 and the PIs of @electionstudies for their assistance. Now please enter!
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
I'm not. Why assume skills transfer? Politics is hard! No one says pols should be CEOs; not sure why converse diff.
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Vain hope: We can hold 2 ideas in our heads at same time. (1) Trump's candidacy is huge news. (2) Early polls aren't
Drew Linzer
@DrewLinzer
Amen @daveweigel: "The story of an election is far, far bigger than the story of who won it." www.washingtonpost.com/news/the-fix/wp/2015/12/04/why-do-…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
A must-retweet every time it's updated - puts overemphasis on early polls in proper perspective
This post is unavailable. See it on X. -
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Amazing opportunity to help lead team evaluating new Indian anti-poverty strategy using RCTs - world-changing stuff
Karthik Muralidharan
@karthik_econ
I am hiring for a very exciting senior research/policy role based in India (with @PaulFNiehaus @sandipz). Apply at: www.povertyactionlab.org/jobs/associate-directordeputy-di…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
Contrary to the hype, no relationship between midterm results & next presidential election www.nytimes.com/2014/11/07/upshot/what-2014-elections-can…
-
Brendan Nyhan (@BrendanNyhan on 🟦☁️)
@BrendanNyhan
The story of PECOTA should terrify the Silver-worshipers out there - replicable peer-reviewed science is best www2.macleans.ca/2012/11/04/tarnished-silver-assessing-th…



































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































