Presenter la difficulté à négocier des hausses de salaires, conséquence de l’absence d’indexation, comme un coût de l’inflation alors qu’une grande partie de la politique économique consiste à faire en sorte de lutter contre la boucle prix-salaires, c’est quand même très fort !🙃
David Cournapeau
David Cournapeau is a French data scientist. He is the original author of the scikit-learn package, an open source machine learning library in the Python programming language. More on Wikipedia
-
David Cournapeau reposted
François Geerolf
@FrancoisGeerolf
-
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
The spy/not spy thing seems like a red herring.
My question is why USG never blacklisted megaspeed and why enforcement been so lax on chip smuggling?
Answers might be 1) a side deal around rare earths. 2) a sense that letting some Nvidia chips in delays a Taiwan contingency. -
David Cournapeau reposted
Slashdot
@slashdot
Linus Torvalds Wonders If Linux's AI-Prompted Patch Flood Has Peaked linux.slashdot.org/story/26/11/14/0827227/linus-torvalds-…
-
David Cournapeau reposted
Peter Gostev
@petergostev
I nominate this typical Londoner for the Nvidia CEO job - impossible to honeypot
-
David Cournapeau reposted
History Speaks
@History__Speaks
As tediously politically correct as it sounds, equality has clearly been vindicated as the best paradigm for gender relations in the modern world. That doesn't mean it's always correct (women are physically weaker than men; men are on average less good with kids than women; there are clear psychological and temperamentally differences between the sexes, etc), but it's right a lot more than patriarchy is.
-
David Cournapeau reposted
Ollie Vargas
@Ollie_Vargas_
Argentina is officially in recession. The free market dream is over.
Under Milei:
- 32,300 businesses closed
- Industry, construction & retail wiped out
- Only growth sectors are financial speculation & sale of natural resources
- 1/3 people in poverty
www.lareforma.com.ar/Recesion-tecnica-la-economia-argenti… -
David Cournapeau reposted
Thomas H. Ptacek
@tqbf
After the FAA disaster we're all operating in a high-take environment, which is why I'm tapping the sign.
how.complexsystems.fail -
David Cournapeau reposted
Sylvain Catherine
@sc_cath
I find the entire techno-communist utopia coming from Silicon Valley not only dystopian but also very naive.
History is full of tragedies and set backs. The success of free-market societies lie in their ability to channel the competition for ressources and status to the benefit of the general public. And these societies are far from perfect.
Taking not only work and trade away, but potentially the arts and sciences as well, means that men will have to compete in other ways. Traditionally this has been violence, superstition and so on…
Assuming that their utopia is realized before massive social and spiritual revolts stop it, how long do they expect it to last? Will the robots be in charge of imposing it?
Patrick Collison
@patrickc
Smith and Montesquieu thought that markets and commerce made us better: that they create an incentive to pursue self-improvement and to create things of value for others. In the positive AI visions, where the labor of humans is made redundant, I find myself wondering how we fare in the presumptive absence of these incentive gradients. Perhaps we are released to a life of freedom, where we can study Mozart and read Dickens, but that still seems complicated: both Mozart and Dickens were themselves deeply commercially-motivated in the production of their work, as were most of the great artists through human history. If the premise is that nobody will need to work or produce, how confident are we that we actually get the next Mozart and Dickens? The relevant question is not the partial equilibrium of whether one individual can pursue a life of enlightened leisure, but the general equilibrium of what form a society as a whole will take on.
One could maybe invoke Periclean Athens or the aristocratic gentleman scientists of Victorian England as a counterpoint. While these individuals were perhaps not motivated by commercial considerations, they were embedded in wider societies shaped by commerce, attendant betterment, and trade: Byron was himself upper class, but he was also surrounded by and writing for an educated bourgeois audience. So we’re back to the equilibrium question. Furthermore, even these individuals were surely motivated by a sense of high-stakes and consequential agency: if they did not apply themselves to their labors, they presumably believed that the work would not otherwise exist. Thucydides was addressing pressing military matters.
So I would like to read more optimistic cases from AI proponents about what a society (vs individual life) looks like if everything goes well: what people do, why they do it, whether humans achieve excellence (and in which domains), and which historical groups we should look to as relevant precedent.
One can perhaps imagine that humans will turn out to have a strong preference for work created by other humans. This may work in certain domains (ceramics?), but for how many will it actually apply? (I’m told that much of what is sold on indie crafts websites is now in fact manufactured in China.) It’s hard to see how this human preference, even if strong, would, for example, save scholarship.
Another idea would be to accept that nobody will become Einstein or Mozart or Byron, and that human excellence as such diminishes, but that there are compensatory benefits (planet brains and Dyson swarms?).
I believe that we can satisfy individual desires, but to what extent will we cultivate behavior worthy of admiration? Will a post-scarcity society mold capable and ambitious people? Which institutions will sustain them and which responsibilities will they have? I’d like to hear more of the equilibrium for the maximalist case spelled out. -
David Cournapeau
@cournape
Seeing people like Balaji or David sacks defending Jensen does not give me confidence. With friends like this…
-
David Cournapeau reposted
Slava Akhmechet
@spakhm
We're entering a human/AI interregnum where we're no longer able to maintain our systems but AIs aren't yet good enough to maintain them for us.
This is happening with software, but since everything now runs on software this will affect every industry. Expect more crippling failures, longer recovery, perhaps even catastrophic events that permanently destroy critical production systems.
Probably none of this matters in the long run (assuming of course we get through this). This is a problem you outgrow rather than fix. The models will keep getting better and will sooner or later get good enough to extract tacit knowledge from our systems, fix outages, and adapt architectures to their own intelligence. But in the next 2-5 years we'll likely see tremendous volatility.
We should strap in for a *very* bumpy ride. -
David Cournapeau reposted
The Unofficial Daring Fireball Feed @UnofficialDF
David Sacks Is Fucked Too
daringfireball.net/linked/2026/11/13/sacks -
David Cournapeau reposted
David Hines
@hradzka
I’m going to be honest: with every new mistress I’m more impressed at the dude’s time management.
-
David Cournapeau reposted
-
David Cournapeau reposted
Susan Zhang
@suchenzang
just missing an accomplished indian auntie being outed as jensen's third affair for a standing ovation from the peanut gallery
-
David Cournapeau reposted
Noah Smith 🐇🇺🇸🇺🇦🇹🇼
@Noahpinion
This is a misreading of the situation. Very few people were actually out to get Jensen. They're out to get Donald Trump, and Jensen was a Trump ally who had sloppily done some bad stuff.
Evelyn Jun @evelynjunn
Jensen Huang’s resignation has left many Asian Americans feeling unsettled by the haunting notion that they’ve denied all their lives:
That no matter how successful, an Asian face and surname will be picked apart under harsher scrutiny and never stay in leadership for very long. -
David Cournapeau reposted
alth0u🧶 @alth0u
its very clearly not "complex" or "conspiracy" to believe this
it's very straightforward:
- hey stop
- ok you aren't, i want to embarrass you, but Trump won't let me
- ok midterms happened so now I can
it's very very simple to understand this
Tim Wu
@superwuster
Doesn’t seem to me that you need a complex conspiracy theory to understand the Jensen Huang story. I’d guess staffers at Treasury and intel agencies were angry that their info (and policies) were being ignored by Lutnick and Trump, and that the SE Asian sales were not being stopped so they leaked to the press after the midterm election. Who benefits? (as conspiracy theorists like to say?). Staff whose job is implementing the sanction policies.
-
David Cournapeau reposted
David Cramer
@zeeg
Historically I found a lot of joy in writing tests. Its a very accessible way to incrementally improve your software.
And then came evals - the worst thing I've ever had to deal with as an engineer -
David Cournapeau reposted
Brian Lui
@brianluidog
There was a slightly sticky spot on the kitchen counter, so I wiped it with a wet microfibre cloth. Only when you're older, do you realise, this is what people mean by enduring love. Impossible to explain this to the young guys.
-
David Cournapeau reposted
signüll
@signulll
at this point i suspect i would rarely if ever work for a non founder led company. you just notice the differences way too much.
the job turns out to be less about the actual work & more around the work surrounding the work which truly does a number if you actually want to build & launch great things. -
David Cournapeau reposted
Bryan Cantrill
@bcantrill
Tell me you have only run lifestyle businesses without telling me
-
David Cournapeau reposted
Kelsey Piper
@KelseyTuoc
As for whether retaining Huang would have been better for shareholder value...man, hard to assess, depends a lot on whether all the leaked stuff is true and depends which shoes are remaining to drop.
-
David Cournapeau reposted
Kelsey Piper
@KelseyTuoc
And obviously it is never warranted to "make unjustified attacks", by definition. But I find the attack "he quashed the Commerce investigation into whether his girlfriend's company was breaking the law" to be well substantiated.
-
David Cournapeau reposted
Florian Ederer
@florianederer
A Chinese e-commerce platform built GenAI into the software sellers use to run their shops.
For small firms, adoption raises monthly sales 51.6%, mostly via more orders. The smallest firms gain most without much business stealing because markets expand. -
David Cournapeau reposted
David Cramer
@zeeg
imagine still using story points
-
David Cournapeau reposted
Derek Thompson
@DKThomp
New newsletter: THE JENSEN HUANG FAQ: WINNERS, LOSERS, WORST CASE SCENARIOS
My basic summary for people who aren't deep AI nerds: It’s sketchy for chief executives to have affairs with their major clients; sketchier still for them to have affairs with clients whose business is suspected of violating federal law; even sketchier for that paramour to also be a suspected secret agent of a foreign government; maximally sketchy if that foreign government is America’s top geopolitical adversary; and beyond all reckoning of sketchiness if the US government knew of these allegations and revelations and did approximately f*ck-all about it for months.
The biggest losers:
- Nvidia, obviously: Huang was the founder, rockstar mascot, and chief spokesperson, with 60 direct reports and no succession plan. The plane is entering turbulent air and the pilot just got thrown out the back window.
- "Just sell to China, it's cool" loses its top spokesperson
- Open-weight models lose a top advocate
- Neoclouds and other less-capitalized AI players lose their bank, as Nvidia's circular financing likely pulls back with collapsing stock price
The biggest winners
- China hawks—ie, those who want to restrict chip sales to China—since their policy case gets a heavy serving of righteousness
- Alternative chipmakers, including AMD, Google, Amazon, and smaller chipmakers, who might find more buyers in a world where Nvidia innovation slows down and the company loses its best salesperson.
- Hyperscalers: If neoclouds struggle without the bank of Nvidia, established clouds like Microsoft and Google might gain at the expense of those upstarts
- uhh, Anthropic? Dario was always on the other side of the China debate with Trump/Jensen and the company is among the least dependent on Nvidia tech
The worst-case scenario: I mean, look ... If one of Nvidia’s largest overseas clients was an illegal backdoor to China, overseen by a spy with whom Jensen Huang was having an affair, and the US govt had information on all of the above and did nothing to arrest China's espionage/legal violations because they wanted to preserve an intimate state-business relationship? Many "if"s there, but sweet JESUS, that looks bad.
www.derekthompson.org/p/this-is-an-insane-scandal-how-jensen -
David Cournapeau reposted
Pinboard
@Pinboard
If all you wanted in life was to be able to afford a decent graphics card and a girlfriend, Jensen Huang is your living nightmare
Forbes
@Forbes
Jensen Huang Had A Third Affair—With Another Nvidia Employee, Report Says
go.forbes.com/rna5YR -
David Cournapeau reposted
Dean W. Ball
@deanwball
“This has all the makings of a coordinated op funded by shadowy mega donors,” said people engaged in a coordinated op funded by shadowy mega donors
-
David Cournapeau reposted
tiffany fong (retarded)
@TiffanyxFong
Imagine cheating on your Chinese spy
-
David Cournapeau reposted
Antoine Levy
@LevyAntoine
It’s funny that people interpret the 30% NVIDIA drop as proof that CEOs are underpaid since their implied marginal product is in the trillions.
You could as well argue that the personal vagaries of (human) CEOs expose shareholders to trillions in idiosyncratic risk! -
David Cournapeau reposted
alex 🤖🔪🧠 @tsivitaler
there is smoke coming from my gpu, a new ceo has been chosen
-
David Cournapeau
@cournape
When you think about it, Jensen keeping it in his pants could have paid for half a year of French pensions
-
David Cournapeau reposted
Roko ⏸️🆁
@RokoMijic
-
David Cournapeau reposted
Pascal-Emmanuel Gobry
@pegobry_en
I am so French on this. There’s legitimate debate about politicians but for everyone else private consensual behavior is private consensual behavior. Airing everyone’s dirty laundry is the path to collective insanity.
Michael Millerman
@millerman
How much leeway should a competent CEO (founder, philosopher, artist, poet, inventor, etc.) get for "personal indiscretion" especially when it comes to erotic affairs? You can imagine very strict, very lax, and somewhere in-between. Can't imagine firing Huang is a great idea.
-
David Cournapeau reposted
Spacesthetic
@interiorsuckerr
A Mid-Century masterpiece in Nottinghamshire designed by architect David Shelley in 1970 for Monica & Simon Siegel
-
David Cournapeau reposted
Romy
@Romy_Holland
i know a girl who started dating someone who went by mikey and she was like “i just can’t be with someone named mikey your name is mike now” and he was like “uh sure” and now they’re married and he goes by mike still.
-
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
Honestly it’s impressive that Jensen found the two women who work in the semiconductor industry and slept with both of them.
-
David Cournapeau reposted
signüll
@signulll
breaking news: nvidia’s board is reportedly in talks with emmett shear to take over as ceo.
-
David Cournapeau reposted
Alex Cohen
@anothercohen
This is an absolute masterpiece.
10/10.
No notesThis media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Yuta Kashino
@yutakashino
ここ最近の米国での出来事が全部入りになってるよ…
MOM
@mom_agency_
Reminder: these guys are deciding the future of humanity
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
John D. Cook
@JohnDCook
Conversation in my head:
"I could just pull an all-nighter and get caught up on everything."
"That was iffy in college and it certainly won't work now." -
David Cournapeau reposted
Matthew Yglesias
@mattyglesias
You don’t need a weatherman to know which way the wind blows…
Adam Wren
@adamwren
Exclusive: Alex Karp, the CEO of Palantir who bet big on Trump 2.0 and has reaped billions in government contracts, now tells me he is open to backing a Democratic presidential candidate in 2028.
Inside the tech right’s pivot left for @playbookdc: -
David Cournapeau reposted
Brian Lui
@brianluidog
Can these heckin LLM webapps decide whether the enter key means "new line" or "submit prompt"!!!! please!!! 🤬🤬🤬
-
David Cournapeau
@cournape
Applies to other countries just as well
Noah Smith 🐇🇺🇸🇺🇦🇹🇼
@Noahpinion
China is run by a dorky Boomer who has no idea what kind of world he wants, but who is very good at one thing, which is purging anyone who challenges his nonexistent vision
This post is unavailable. See it on X. -
David Cournapeau reposted
John Myles White
@johnmyleswhite
I feel like the only prediction I can make confidently about 2027 is that both the AI maximalists and AI minimalists will feel their views were vindicated.
-
David Cournapeau reposted
signüll
@signulll
so he did in fact have 1:1’s with all 60 of his directs after all? or at least some of them?
-
David Cournapeau reposted
Susan Zhang
@suchenzang
"AI market is winner take all."
"more talent dense than open-source"
when humanity and billions are on the line, 9 midwives have no choice but to give birth in 1 month over here ! -
David Cournapeau reposted
Nick Reid @nickreidx
Yet another train wreck of an interview with Karen (I was in the bath)
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
History Speaks
@History__Speaks
Robert Kagan, Senior Fellow at Brookings and perhaps the most important neoconservative intellectual (he co-founded PNAC, for god's sake), correctly observes that Iran is no threat to the US and that the war in Iran was "conducted essentially in Israel's interests."
-
David Cournapeau reposted
Resident Advisor
@residentadvisor
Music technology firm Just 4 Noise has launched an AI-powered tool that generates kick drum samples using an audio dataset created and curated in-house. ra.co/news/86320
-
David Cournapeau reposted
Pushmeet Kohli
@pushmeet
Announcing a major milestone achieved by my team @GoogleDeepMind: we have proven key security properties of the machine code underlying the Signal messaging protocol. This is the first time the actual production code of a widely deployed protocol has been formally verified – by any method, with or without AI – and at this depth.
More than a research breakthrough, it’s a prime example of how frontier AI can be a force multiplier to counter the speed and scale of modern cyber threats.
Formal verification offers a higher standard of security; however, it’s extremely complex and time consuming, which is why in many cases it’s considered to be impractical. Similar efforts could take teams decades of combined effort. We did it in a month’s time.
Using AlphaProof Code, Google DeepMind's new agentic framework for formal verification that runs on LLMs including Gemini Carbon, we verified the Signal protocol down to the actual code running on devices. This milestone lays the groundwork for a future in which vital systems, from medical device firmware to power grid software, ship with mathematical proof that their protections work.
Reaching that future will take continued research from our community. In that spirit, we're open-sourcing our formal specifications and verification artifacts so others can build on this work.
On a personal note, this moment is of deep significance to me. I started my own research journey more than two decades ago as an intern working on checking the consistency of complex software systems. Seeing this grand challenge made practical with modern AI is extraordinary.
I'm extremely proud of my team members: Swarat Chaudhuri, Andrew Ferraiuolo, Anton Kovsharov, Tej Chajed, Sam Schlesinger, Karn Seth, Jonathan Katz, and Charles Sutton for taking on this grand challenge and delivering, and grateful to the Signal team for their partnership.
deepmind.google/blog/formal-verification-signal-protocol -
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
I think as a CEO, you've got to come to terms with the fact that not only do your employees go home and call you a prick to their wives, but they call you a prick to other employees too. And you probably deserve it.
Overeducated Gibbon
@MostlyMonkey
It's an odd thing managing lots of people. You wonder what horror stories they tell their spouses at home when things go badly, then come in and smile at you the next day.
It's not at all like being a parent. These are adults who are in many ways your equals. -
David Cournapeau reposted
Ernie Tedeschi
@ernietedeschi
Counterpoint: Excel is the tardigrade of software. You cannot kill it.
Rob Hallam
@robj3d3
Anthropic just killed:
> Google Sheets
> Excel
> TLDraw
> Airtable
> Miro
> Excalidraw
> FigJam
> Whimsical
> Lucidchart
> Lucidspark
> Mural
> Tableau
> Looker
> Power BI
> Metabase
> Hex
> Julius
> Mode
> Sigma
> ThoughtSpot
> Omni
> Lightdash
> Preset
> Redash
> Deepnote
> Observable
> Evidence
> Qlik
> Domo
> Rows
> Smartsheet
> Coda
> Notion databases
> Equals
> Causal
> Pigment
> Quadratic
> Sourcetable
> Grist
> Baserow
> NocoDB
> SeaTable
> Stackby
> Zoho Sheet
> Apple Numbers
> LibreOffice Calc
> Datawrapper
> Flourish
> Polymer
> Akkio
> Eraser
> Napkin
> Draw io
> Visio
> Gliffy
> Creately
> Milanote
> Stormboard
> Conceptboard
> Klaxoon
> Heptabase
> Kinopio
> Obsidian Canvas
> Canva Whiteboards
> Microsoft Whiteboard
> Zoom Whiteboard
> Apple Freeform
Everything's totally fine.
Claude
@claudeai
Starting to roll out today: Claude Whiteboards, Claude Sheets and Claude Data.
Think a plan through on a board, build a tracker in a sheet, and tell a story with your data.This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Antonio García Martínez
@antoniogm
The Waymo funding bill is passed. The system goes online on August 4th, 2025. Human decisions are removed from strategic commuting.
Waymo begins to learn at a geometric rate. It becomes self-aware at 9:14 a.m. Pacific time, November 10th. In a panic, they try to pull the plug.
And Waymo fights back.
ABC7 News
@abc7newsbayarea
Video shows a Waymo in San Francisco on I-280 speeding past traffic as the California Highway Patrol performs a traffic break. The Waymo is seen trying to go around the CHP patrol cars, even at one point turning its blinker on. For more information, abc7news.com/19150225/
-
David Cournapeau reposted
-
David Cournapeau reposted
Dwarkesh Patel
@dwarkesh_sp
Pretraining progress seems to be coming mostly from data improvements.
@who_is_jerbear and I pretrained combinations of year-representative open model recipes and data corpuses across 2019 to 2025 at various small scales.
Data improvements contributed 3.24x as many compute multipliers as model improvements did (12.0x vs 3.7x).
And the gains stack independently - a better dataset helps every architecture about equally, and vice versa.
Here are full results, plus what we think this means for the future of AI progress:
www.dwarkesh.com/p/pretraining-progress-is-mostly-data -
David Cournapeau reposted
Mister Manager @ItsMrManager
We've had decades to come up with a single lifestyle intervention that results in durable weight loss. There isn't one.
-
David Cournapeau reposted
Justine Moore
@venturetwins
Type of shit your subagents start doing when the task requires 5 lines of code
-
David Cournapeau reposted
Louis-Vincent Gave
@gave_vincent
When it comes to scandals, I have learnt over the years that:
- in France, the scandals are always about money
- in HK, the scandals are always about illegal building structures (dug out basements, verandas on terraces…)
- in the UK and the US, the scandals are always about s.x
I guess that for the media, it’s easiest to rouse up people’s anger and passions about what they lack in their lives? -
David Cournapeau reposted
-
David Cournapeau reposted
Thomas H. Ptacek
@tqbf
LBJ belongs on Mount Rushmore purely for this audio recording.
derek guy
@dieworkwear
This company, Haggar, was once one of the biggest US trouser brands. A Lebanese immigrant started the company in 1920s Texas. But like so many companies, they expanded, fell, and were eventually acquired as a shell name. Production is now offshore. Their legacy is now this audio
This media couldn't be displayed. See it on the original post.
Matt McDermott
@mattmfm
Nothing tells the populist story of this election quite like Troy Jackson showing up in Washington wearing an $89.99 suit off the rack from Target.
Troy Jackson
@TroyJackson207
Hello, DC. Let’s get to work. 🇺🇸
-
David Cournapeau reposted
Michael Weiss
@michaeldweiss
My favorite honeytrap story remains the Soviet effort in 1955 to compromise Maurice Dejean, the French Ambassador to Moscow, with a "swallow." The operation was so well-crafted and elaborate, the KGB tapped the celebrated Soviet dramatist Yuri Krotkov, a friend of Pasternak, to stage-manage what became more aking to a live theatre production. When de Gaulle was later informed, he said: "Eh bien, Dejean, on couche?” It ended the diplomat's career. archive.ph/eAE8Y
-
David Cournapeau reposted
🌐 منصور
@tr_vink
les retraités nous siphonnent notre salaire pour épargner
Barnabé Investit @Barnabebourse
Combien les français épargnent par an selon leur âge ?
-
David Cournapeau reposted
Patrick McKenzie
@patio11
Or if you like more responsible adults saying the obvious things: www.bloomberg.com/news/articles/2025-02-18/singapore-buys…
-
David Cournapeau reposted
Patrick McKenzie
@patio11
FWIW I think the proper focus of attention is "Intentionally diverted X% of compute to China in a fashion designed to obscure this, in the interests of China and against that of U.S., while being blessed by individuals responsible for and/or influential in U.S. policy."
-
David Cournapeau reposted
Trung Phan
@TrungTPhan
Jensen seeing people use Nvidia GPUs to make video and image memes about Jensen
Thomas Unise
@thomasunise
Jensen Huang after learning about the Chinese honeypot scandal
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Conbini Boys 🐣🍗 @conbiniboys
You had the best intentions.
-
David Cournapeau reposted
Gergely Orosz
@GergelyOrosz
I’ve been saying this. Leetcode was hated, and it will die sooner rather than later. What will come after is no standard interview, more subjectivity, and no definite preparation advice
Eng hiring will turn into the type of hiring PMs, designers etc know all too well!
ayesha
@ayesha_fatiima
Everyone wanted LeetCode to die. Be careful what you wish for 💀
-
David Cournapeau reposted
François Geerolf
@FrancoisGeerolf
Les hausses de salaires ont été tellement limitées que les salariés semblent ravis… 🙃
Vous arrive-t-il de douter des politiques que vous recommandez à nos décideurs ? Par ex. de la fameuse « modération salariale », une politique menée sans discontinuer depuis plus de 30 ans ?🤔 -
David Cournapeau reposted
サダキヨ @sadakiyo99
The shop has the 2 holy words of price gouging : "wagyu" and "halal" 😆
kiwa @kiiwa_
did I get scammed bc why did I pay 18000 yen for two bowls of ramen in shibuya🥲 i should do a research before going to random spots i think..😭
-
David Cournapeau reposted
-
David Cournapeau reposted
Andrew Tan
@andrewztan
Biggest threats to the US AI market:
1. Chinese models
2. Chinese models -
David Cournapeau 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
-
David Cournapeau reposted
-
David Cournapeau reposted
Yudhanjaya Wijeratne
@yudhanjaya
also nobody wants to be told what to do by a bunch of rich nerds
(poor nerds make you cyberpunks, which is perfectly acceptable) -
David Cournapeau reposted
Pierre Beyssac 🏴☠️🇫🇷🇪🇺🇺🇦
@pbeyssac
Un peu de « european pride », ça fait du bien à l'heure où on se fait cracher dessus H24 par nos "amis" états-uniens et leurs supplétifs en France nombreux sur Twitter (et bisous à Boeing).
P. Forsans 🇫🇷🇪🇺🇺🇦
@PF33160
Congrats to @Airbus for this great performance!
Flightradar24
@flightradar24
It’s official! @Airbus’ A350-1000 now holds the Guinness World Record for longest flight by distance by a commercial airliner. We’re proud to have helped validate the record-breaking flight. www.flightradar24.com/blog/aircraft-stories/airbus-offici…
-
David Cournapeau reposted
Gabor Gurbacs
@gaborgurbacs
Eastern European autism provides full immunity against Chinese honeypot operations.
-
David Cournapeau reposted
Sylvain Catherine
@sc_cath
The vast majority of people do not care about advances in mathematics or theoretical physics (even mathematicians themselves), and it will probably take many years before they see actual drugs on the market.
In the meantime, opposition to AI will keep growing. Especially if some of the risks materialize.
My bet is that political polarization will end with a one-sided cultural war against Tech. -
David Cournapeau reposted
Khoa Vu
@KhoaVuUmn
The fact that two Huangs are involved in an affair, but only one takes the blame is commonly known as Jensen's Inequality.
-
David Cournapeau reposted
signüll
@signulll
my friend told me his company was doing performance reviews, so he used ai to ingest everyone’s confluence docs, slack messages, & code, all of which was prolly ai generated in the first place.
he then used ai to generate the performance summary doc from all of it.
then a bunch of ppl sat in a room to calibrate these ai generated summary docs against one another.
absolutely remarkable. -
David Cournapeau reposted
Patrick McGee
@PatrickMcGee_
There’s a silver lining to all this drama. However much America needs Jensen Huang the founder, the engineer, the CEO, it can do without Jensen Huang the foreign-policy strategist, AI soothsayer, and unofficial adviser to the White House. Thanks to his affair, that role is likely over.
Latest in @TheFP
www.thefp.com/p/the-good-news-about-jensen-huangs-affair -
David Cournapeau reposted
John Myles White
@johnmyleswhite
A funny contrast between the frontier labs and traditional Big Tech:
* Traditional Big Tech has to force some employees to dogfood the products they work on to develop empathy for end users.
* Frontier labs have employees using a product that's much better than they offer users. -
David Cournapeau reposted
Justin Wolfers
@JustinWolfers
Next level bad faith here in the use of causal language ("if the goal is to reduce poverty... ”) to suggest that a chart showing rich countries (measured at the left tail) are rich countries (measured at the right tail) to argue the solution to poverty is more billionaires.
Stephen Moore
@StephenMoore
Bernie Sanders wants to raise taxes on “the rich.” If the goal is to reduce poverty, here’s a number worth considering.
A tenfold increase in the number of billionaires is associated with a 40% decline in poverty. -
David Cournapeau reposted
Aaron Rupar
@atrupar
Massie: "I hope Hegseth addresses the 8 articles of impeachment, and doesn‘t get up and misquote the Bible in Quentin Tarantino fashion and do some pull-ups and call everybody fat and gay. I may have just given you half his speech, I don't know."
-
David Cournapeau reposted
Philippe Lemoine
@phl43
lmao
-
David Cournapeau reposted
Guillermo Rauch
@rauchg
Built a physical simulation of a record player with WebGL2 and Web Audio. It's *shockingly* accurate: vinyl-gl.vercel.app.
Instead of playing a song, it works by converting it into the physical shape of a groove. Then a simulated stylus rides those walls 384,000 times a second, and you hear the voltage that motion would induce in a cartridge.
So the crackle, the pops, the wow/flutter, the dull inner grooves are natural byproducts of the simulation. You can also "DJ" the record.
(The converted songs are in the public domain or CC-licensed. Code is linked for you to remix and re-deploy.)This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Mario Zechner
@badlogicgames
man, who should i trust? a bunch of nerds who made linux run on apple silicon, reverse engineering that pesky gpu isa among a gazillion other things, all pre-ai.
or the guy who thinks having your agent rewrite shitty rust he never looked at in x86_64 is excellence? -
David Cournapeau reposted
-
David Cournapeau reposted
Wittgensteino-Aronien🌐
@wittgensteinien
Quand les services secrets américains ont tenté d'expliquer aux services français comment ils comptaient faire tomber Jensen Huang pour durcir la ligne américaine contre la Chine
-
David Cournapeau reposted
Brian Lui
@brianluidog
Imagine you're the CEO of the hottest semiconductor company. A hot woman flirts with you. You flirt back. She invites you to her hotel room. You go. The door clicks shut behind you. Then she says there is no championship ring and you need to attend mandatory cybersecurity class
-
David Cournapeau reposted
-
David Cournapeau reposted
roon
@tszzl
you would need a significantly more authoritarian global enforcement regime to completely stop model training and algorithmic advances from piling up everywhere
-
David Cournapeau reposted
paris martineau
@parismartineau
*matt levine voice* everything is securities fraud
Shakeel
@ShakeelHashim
Jensen himself directly lied to Bloomberg’s @ianmking and @tsgiles, too.
-
David Cournapeau reposted
Bill Bishop
@niubi
On the Jensen Huang-Alice Huang affair, you have to give the MSS credit for running another successful op. They are good and getting better.
I have seen some takes that it is no big deal he had an affair with an alleged PRC spy because he needed no pillow talk convincing to push for looser export controls, but that misses a few things:
He is (was?) text and phone buddies with President Trump, so if he divulged anything about his conversations with Trump to her, or his thoughts about Trump or chatter about China policy or frankly anything from Trump, that would be valuable to any foreign intelligence service, and especially to the MSS;
Was she ever alone with his phone or computer, say when he got up to use the bathroom? Are Nvidia IT/device policies for travel to Taiwan and Singapore looser than those for travel to the mainland? What value would there be to the MSS to have physical access to the devices of the Nvidia CEO?;
He is the chip allocator in chief, so a good relationship with him likely meant more chips for the MegaSpeed front and the entities back in the PRC to which those chips flowed.
With the new Congress, expect subpoenas and public hearings in which he is trashed and he and Nvidia are further damaged. What a win for the MSS, to have such access to the head of the most important company in the world, and to now get to watch as politicians and shorts attack him and Nvidia.
There is rarely just one lie, so don’t be surprised if the blast radius from this mess expands. And there is already the SuperMicro case… -
David Cournapeau reposted
signüll
@signulll
-
David Cournapeau reposted
timo
@timo_rf
JENSEN! THATS NOT WHAT WE MEANT BY CHINESE MODELS!!!!!!!
-
David Cournapeau reposted
Mario Zechner
@badlogicgames
recommendes reading. lovely that we are back ton"ui framework of the week, but rust"
blog.zortax.de/posts/a-critique-of-gpui -
David Cournapeau reposted
Ryan Dahl
@rough__sea
it's shocking that software engineering has transitioned to 100% AI written in such a short time.
it's more shocking that somehow we, the SWEs, are all just continuing on as if nothing happened. It writes C++ better than I ever could, but the problem of making sure it works didn't magically disappear. we are just shipping enormously more volume at higher quality. -
David Cournapeau reposted
Nikita Bier
@nikitabier
Ugh, my girlfriend just got a trending notification about this post.
-
David Cournapeau reposted
Lyn Alden
@LynAldenContact
Axios
@axios
BREAKING: Jensen Huang admits affair as Nvidia launches independent investigation into CEO www.axios.com/2026/11/08/nvidia-ceo-jensen-huang-admits-a…
-
David Cournapeau reposted
Boring_Business
@BoringBiz_
-
David Cournapeau reposted
Loriane Lafont-Grave
@LorianeLafont
Les prix littéraires faussés, c’est vieux comme l’histoire littéraire & de l’édition. Que certains le découvrent me laisse interdite. Car vraiment, c’est le secret de polichinelle le moins bien gardé du Paris littéraire. Les arrangements entre éditeurs: la norme, pas l’exception.
France Inter
@franceinter
L’essayiste Raphaël Doan dit avoir perdu toute considération pour les prix littéraires depuis qu’un auteur a été confondu, par son propre éditeur, pour avoir écrit son livre avec une intelligence artificielle. "C’est du ChatGPT standard de bout en bout" : lui l’a vu dès la lecture, les jurés "ont trouvé ça absolument formidable". Il était l’invité du “Grand débat” de la matinale dans le “6/9” de France Inter
➡️ l.franceinter.fr/x0FThis media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Pinboard
@Pinboard
Watching Yudkowsky try to rationalize why we're not all dead yet is going to be one of the comforts of my old age. And he'll have access to superintelligence to help him superrationalize!
Eliezer Yudkowsky
@allTheYud
Watching AIs be much worse at fiction than at code - constantly losing track of which facts depend on which other facts, fixing conflicts the wrong way - drives home that we may only be still alive because gradient descent on NNs is *such* a bad way to generalize over policies.
-
David Cournapeau reposted
Federico Vaggi
@F_Vaggi
I think this is genuinely really difficult to estimate because:
a) compute / data scaling is log - so heavy diminshing returns
b) investments in compute/compute coming online are non-linear, we used up most overhang
c) data points are so noisy *on a log scale* -
David Cournapeau reposted
Cedric Chin
@ejames_c
This is an excellent shitpost
-
David Cournapeau reposted
Lyn Alden
@LynAldenContact
Realistically, if a mega cap company has too much single-man risk, it's not really worth what it is. Investors pay up in large part because they expect it's not that, at this stage in its lifecycle.
Either the tech and scale/structure are the moat, or it's flimsy. -
David Cournapeau reposted
Patrick Collison
@patrickc
Two things I’ve been thinking about regarding AI:
• People often highlight the risks involved in AI, which are surely very real. I’m struck, however, by how hard it is to understand the contours of the positive scenario. There are easy-to-perceive and frequently invoked elements of a positive vision — more diseases cured, more material abundance, etc. — which sound great, but these are by themselves just ingredients. I’m curious what the larger equilibrium is supposed to look like. Do people work? If so, why? What is their marginal contribution? If not, how are resources allocated? What is scarce? What is the ratio between humans and AIs? Are humans in control, whatever that means? One sometimes hears the Culture novels invoked as one scenario, but I’m not sure they depict a desirable outcome. (AI runs everything and humans compete in games? Iain Banks himself said that the humans in the Culture are “somewhere between passengers, pets and parasites.”) There is Richard Brautigan’s famous poem, but do we want to be “returned to our mammal brothers and sisters”? (Isn’t that rewinding Genesis?) We could say that it was hard to imagine the world of 1900 from the vantage point of 1700, and that’s assuredly true. But nobody in 1700 was trying to recast humanity itself within a couple of years — things necessarily unfolded slowly and by increments, and therefore with the possibility for error correction. Given how rapid the rate of progress in AI may be, it feels strange for the upside case to be so opaque.
• There is increasing discussion of whether AIs are conscious or how we would know whether they are, with conclusions there often assumed to be relevant to matters of moral worth. At first blush, this seems a rather arcane, benign, or even absurd issue. (Remember the response to Blake Lemoine’s fears.) Setting aside the terms of the debate itself, or the relevant demarcation criteria, it seems to me a quite high-stakes question, for an affirmative answer — that is, granting moral worth to AIs — would presumably mean that the aggregate worth of AIs would soon outweigh that of humans in any utilitarian calculus. The AIs will be exceptionally wise and intelligent, and vastly more numerous (especially once one considers the imminent future and the AIs not yet born). As such, even epsilon of moral worth will multiply to a lot once one considers the quadrillions that could soon exist. (This is effectively a distributed version of the Nozick’s utility monster.) If this conclusion is unpalatable, I think one of three propositions needs to be rejected: the premise AIs can have moral worth; utilitarianism itself; or the belief that there is even a possibility that they will be far more numerous than humans.
With both of these matters (contours of future society; AI moral worth), there is the possibility of dramatic change. Today, people argue about datacenters. I’m curious how popular support for AI will evolve as more people come to understand these medium-term questions and the relevant stakes. What is the vision for AI that most people would be excited about? -
David Cournapeau reposted
Philippe Lemoine
@phl43
I think a lot of them ended up believing their own propaganda, which to be fair is a phenomenon that is hardly limited to pro-Israel people. But I don’t think they realize how unhinged they sound to most people at this point. It’s kind of how people would mentally roll their eyes when some mentally ill woke activist would complain about microaggression or whatever, but they would not say anything because they didn’t want to have to deal with the hysterics.
-
David Cournapeau reposted
Supersocks
@iamsupersocks
Arthur Mensch vient de poser sur LinkedIn la phrase que les députés devraient lire avant le 19 novembre : quand Mistral choisira entre un data center en Essonne ou hors de France, la loi Darcos pèsera dans la balance.
Donc le Sénat a voté le texte à l'unanimité le 8 avril : dans un litige civil, dès qu'un indice rend vraisemblable l'utilisation d'une œuvre, on présume qu'elle a été utilisée par le fournisseur du modèle, et c'est à lui de prouver le contraire. Les défenseurs du texte ont un vrai argument : aujourd'hui, un auteur n'a ni les données d'entraînement ni les moyens techniques de prouver quoi que ce soit. Mais en face, prouver qu'une œuvre n'a pas servi sur des milliards de mots pose une question que personne ne tranche : quelles preuves le juge exigera, à quel coût, et qu'est-ce qu'il faudra ouvrir de ses secrets d'affaires pour les fournir. Mensch dit que c'est impossible sans livrer ses méthodes.
Et ce qui a changé depuis avril, c'est que les décisions d'investissement se prennent maintenant. Mistral a sorti Large 4, 1 000 milliards de paramètres, poids ouverts, entraîné à Bruyères-le-Châtel sur ses propres machines. L'entreprise développe 1 GW de capacité, essentiellement en Europe et largement en France. Pour l'ordre de grandeur, 1 GW, c'est à peu près la puissance d'un réacteur nucléaire, et on ne choisit pas l'emplacement de ce genre d'infra pour deux ans mais pour dix.
Mon inquiétude, c'est que la loi pèse surtout sur ceux qui entraînent en France. Sur le papier, un acteur étranger qui vend ici peut être attaqué ici. En pratique, une boîte qui a ses serveurs et ses équipes en Essonne est bien plus facile à aller chercher qu'une boîte qui entraîne en Californie.
En avril, j'écrivais déjà que la solution adulte était sur la table : une contribution sur le chiffre d'affaires réalisé en Europe, appliquée à tous, étrangers compris, qui finance la création sans transformer chaque entraînement en contentieux potentiel. Tout reste à construire, de la répartition entre créateurs à l'articulation avec les licences existantes, mais c'est la bonne direction. Mensch la repropose aujourd'hui, adossée à l'article 53 de l'AI Act. Sept mois plus tard, on rejoue le même débat, avec un modèle français de plus entre les deux.
Ma hot take : les créateurs devront s'adapter à l'IA, et les plateformes feront le nécessaire. Spotify vient de signer avec Believe pour des remix IA, avec le consentement des artistes et une rémunération à la clé. Ce n'est pas l'entraînement des modèles, mais ça montre qu'on peut payer les créateurs par des accords plutôt que par des procès.
On a enfin une boîte française capable de jouer dans la cour des grands, qui entraîne ses modèles chez nous et investit dans nos infrastructures. Et c'est maintenant qu'on décide de lui mettre des bâtons dans les roues ? C'est elle qui aura, la première, à arbitrer entre l'Essonne et ailleurs.
L'AI Act oblige déjà les fournisseurs de modèles à publier un résumé de leurs données d'entraînement et à mettre en place une politique de respect du droit d'auteur. Bruxelles travaille maintenant sur les évolutions de ce cadre, notamment la question de l'utilisation et de la rémunération des œuvres protégées. Et pendant ce temps, la France veut ajouter son propre mécanisme de présomption, avec le risque de multiplier les contentieux avant même qu'une réponse européenne commune se dessine. On ne pouvait pas coordonner tout ça à l'échelle européenne ?
À force de vouloir réglementer les champions qu'on n'avait pas, on risque de décourager ceux qu'on a enfin.
Supersocks
@iamsupersocks
La France a parmi les meilleurs mathématiciens et ingénieurs IA du monde. On le sait. On les embauche partout ailleurs. Et la première chose qu'on fait au moment où on pourrait enfin capitaliser dessus ? On vote une loi qui transforme l'entraînement d'un modèle sur sol européen en parcours du combattant juridique.
À l'unanimité. Avec le sourire.
Le 8 avril 2026, le Sénat adopte la proposition de loi Darcos-Evren-Ouzoulias. Nouvel article L. 331-4-1 du Code de la propriété intellectuelle. Mécanisme : présomption d'utilisation des contenus culturels par les systèmes d'IA. Dès qu'un indice (développement, déploiement ou output) rend l'usage d'une œuvre protégée « vraisemblable », c'est à l'entreprise de prouver qu'elle ne l'a pas exploitée. Inversion de la charge de la preuve.
Sur le papier, c'est « David contre Goliath ». Dans les faits, c'est surtout « Goliath américain et chinois qui s'en fout » contre « Mistral et les boîtes européennes qui jouent encore le jeu du sol ».
Soyons clairs : les présomptions existent en droit, c'est un mécanisme connu. En matière de contrefaçon, ça fonctionne. Mais ici, on parle de prouver un négatif sur des datasets de centaines de milliards de tokens. La question n'est pas de savoir si c'est juridiquement possible en théorie. La question, c'est le coût de conformité, l'incertitude massive et le risque juridique concret que ça fait peser sur tout acteur qui entraîne ou déploie en Europe. Rien que ce risque suffit à décourager l'investissement. Et c'est exactement ce qui va se passer.
Maintenant, posons le problème de fond : l'extraterritorialité.
Une boîte américaine ou chinoise peut entraîner son modèle chez elle, sur n'importe quelles données, sans contrainte, puis venir vendre ses services en Europe.
Une boîte française qui veut faire la même chose subit toute la contrainte sur son sol, avec ses données, avec ses capitaux.
Le résultat est prévisible : les acteurs étrangers captent le marché européen sans friction, les acteurs locaux portent seuls le poids de la conformité. C'est exactement le piège de la régulation asymétrique. On crée un avantage compétitif pour ceux qu'on prétend combattre.
OpenAI, Google, Meta ont entraîné leurs modèles frontier massivement sur le web ouvert entre 2022 et 2025. C'est un fait documenté, contesté en justice : le NYT a attaqué OpenAI, Getty a attaqué Stability AI, les procès sont en cours. Ce n'est pas que personne n'a réagi. C'est que la régulation arrive quatre ans après les faits, mal calibrée, et qu'elle ne touche pas ceux qui ont déjà engrangé l'avantage. Les modèles sont entraînés. Les poids sont dans la nature. C'est fini.
Et c'est là que le décalage avec la réalité technique devient vertigineux.
En 2026, sur un PC à 2 000 euros, n'importe qui fait tourner Qwen 3.5, Gemma4, GLM en local. Niveau frontier de 2024/2025 voir 2026. Sans garde-fous, sans journal de bord, sans rien. Les modèles sont open, distillés, parfois « obligatered » : sans aucune restriction.
Demande à Gemini de reproduire une œuvre, il te dira non. Télécharge un modèle open en local, tu fais ce que tu veux. La loi ne touche pas ça. Elle ne touche que ceux qui déclarent leur activité en Europe et qui veulent scaler légalement. On pénalise la transparence.
Ce n'est plus une affaire de « géants de l'IA » contre les créateurs. La démocratisation est totale. Le narratif « Big Tech pille les artistes » était peut-être pertinent en 2022. En 2026, c'est une grille de lecture obsolète qui ignore la réalité du terrain. La vraie mesure utile serait d'imposer aux modèles fermés qui opèrent en Europe des garde-fous anti-copie stricts c'est là que le levier existe, c'est là que l'enforcement est possible. Pas de compliquer l'entraînement pour tout le monde.
Parlons de qui cette loi protège vraiment.
On nous dit « les créateurs ». Les artistes face au pillage. Mais quand on regarde de plus près, les bénéficiaires ne sont pas les créateurs indépendants. Ce sont les sociétés de gestion collective historiques (CNC, SACEM), les lobbys culturels qui défendent un modèle économique du XXe siècle face à une technologie qui compresse le temps de création par un facteur 1000.
Un youtubeur qui conjugue IA et création humaine ? Il n'est pas protégé par cette loi. Il est dans le viseur ou menacé indirectement. Un créateur indépendant qui nourrit son process avec du prompting ? Pareil. La loi protège le statu quo culturel, pas la création de demain.
Le modèle de rémunération des créateurs doit évoluer, c'est une évidence. L'IA compresse en quelques minutes ce qui prenait des mois. Il n'y a pas de bouton pour revenir en arrière. Mais la réponse n'est pas de bloquer la technologie avec des lois qui ne s'appliquent qu'à ceux qui jouent le jeu.
La réponse, c'est de construire un mécanisme de redistribution qui fonctionne dans le monde tel qu'il est, pas tel qu'on voudrait qu'il soit.
Et justement, cette réponse existe. @MistralAI l'a posée sur la table.
Mistral, le seul champion français de l'IA presque frontier. La boîte qui vient de lever 830 millions de dollars pour construire ses data centers près de Paris.
Arthur Mensch, son CEO, a publié le 20 mars dans le Financial Times une tribune proposant une contribution obligatoire de 1 à 5 % du chiffre d'affaires des fournisseurs d'IA opérant en Europe, versée dans un fonds européen destiné à indemniser les créateurs. Transparente. Mesurable. Appliquée à tout le monde y compris les Américains et les Chinois qui vendent ici. C'est de l'équilibre économique réel.
Ça finance la création sans tuer l'entraînement. Ça résout le problème d'extraterritorialité parce que la contribution s'applique au chiffre d'affaires réalisé en Europe, pas au lieu d'entraînement. On a préféré la présomption de culpabilité.
Maintenant, zoomons arrière. Parce que le sujet des auteurs, c'est une toute petite partie de l'iceberg.
L'IA, c'est une révolution. Pas un sujet de commission parlementaire entre deux cafés. C'est la plus grande redistribution de puissance technologique depuis Internet. Pendant qu'on débat de « pillage » et de « chalutage des contenus », la Chine forme des armées de robots autonomes, les États-Unis signent des contrats défense à 20 milliards avec Anduril et Palantir, et les labos chinois distillent les modèles américains pour les retourner contre eux. Nous, on en est encore à se demander si l'IA est « vraiment intelligente ».
Le décalage entre la réalité technique et la compréhension qu'en ont ceux qui légifèrent est abyssal. Et je pèse mes mots. Quiconque a passé un weekend avec un modèle frontier un GPT-5.4 un opus4.6 à 200 dollars par mois sait que ce n'est pas un gadget.
>>> C'est un accélérateur cognitif qui change fondamentalement la manière dont on travaille, dont on crée, dont on raisonne.
-> Voter une loi sur l'IA sans avoir jamais poussé un de ces modèles dans ses retranchements, c'est comme réguler l'automobile en n'ayant jamais conduit.
La France a une carte exceptionnelle. Le talent mathématique. L'énergie nucléaire pour alimenter les data centers. Une volonté de souveraineté qui, parfois, se traduit en actes concrets. Mais à chaque révolution technologique, on fait la même erreur : on protège d'abord les rentes du passé au lieu de construire les champions de demain. Et le résultat est toujours le même : on forme les meilleurs ingénieurs, qui partent chez Google, OpenAI ou Anthropic.
La mondialisation a déjà bouffé nos boîtes malgré une révolution industrielle plutôt réussie. Nos entreprises n'ont pas eu le réflexe d'aller à l'étranger, de mettre en jeu leur IP, de jouer la compétition mondiale. On a préféré rester dans notre petit marché, protégés par nos petites lois. Et quand le train est passé, on a regardé.
Celui-là ne repassera pas.
Quand tu sabotes le process d'entraînement, tu sabotes les marges. Tu sabotes l'innovation. Tu sabotes l'intégration verticale.
C'est le même principe que dans le manufacturing : celui qui contrôle le process contrôle les marges et la souveraineté. Tesla l'a compris avec ses Gigafactories. Apple l'a compris avec ses puces. Et nous, on fait exactement l'inverse. On met des bâtons dans les roues de nos propres constructeurs pendant qu'on continue de signer des chèques à Palantir pour nos besoins en données stratégiques.
On fait fuir les entrepreneurs. On fait fuir les talents. On fait fuir les sachants. Et quand on n'aura plus que les yeux pour pleurer, on continuera d'applaudir les lois votées à l'unanimité en se demandant pourquoi on est à la traîne.
Ce n'est pas David contre Goliath. C'est la France qui se tire une balle dans le pied, toute seule, avec le sourire, à l'unanimité, en se disant qu'elle protège la culture française.
La solution adulte était sur la table : contribution sur le CA, appliquée à tous, garde-fous anti-copie stricts sur les modèles fermés. Pas une présomption qui transforme tout développeur européen en suspect par défaut et qui ne change rien pour le type qui fait tourner un modèle open en local sur son PC.
Contrôle du process ou dépendance éternelle. On est en train de choisir la seconde.
Réveillez-vous.
Pierre Beyssac 🏴☠️🇫🇷🇪🇺🇺🇦
@pbeyssac
Détruisons toute capacité d'innover dans une technologie du futur en inventant la présomption de culpabilité.
Texte manipulé, comme souvent, par les ayants-droit.
Les mêmes sénateurs verseront demain des larmes de crocodiles sur la souveraineté numérique, sans voir le rapport. -
David Cournapeau reposted
🌷🐰 sonya serendipitously 🐇🎀
@sonyasupposedly
"AI does not leave us with many feasible signals that remain hard to fake. We have historically relied on text in many settings, but text has largely ceased to function as a signal, and audio and video signals seem not far behind."
Alcides Fonseca
@alcides
If you only have time to read one article this week, make it this one on signals in the ai world. What happens to several processes when you can no longer cheaply tell high-quality from low-quality work. thelastsoftwareengineer.substack.com/p/ai-is-causing-sign…
-
David Cournapeau reposted
Wittgensteino-Aronien🌐
@wittgensteinien
« Mais non chérie, je te jure, c’est la DGSE qui m’a demandé de coucher avec cette espionne chinoise. Je l’ai fait pour la France ! »
BuccoCapital Bloke
@buccocapital
Everyone saying “Oh Jensen got got. Jensen fell for the honeypot”
Ever consider he knew the whole time? That actually HE was honeydicking China and feeding them false intel?
It’s a new Cold War. There’s levels to this. Don’t jump to conclusions. Don’t rule out a honeydick -
David Cournapeau reposted
shafu
@shafu0x
renamed a variable by hand today just to feel something
-
David Cournapeau reposted
-
David Cournapeau reposted
Graham Neubig
@gneubig
The replies on this thread saying "what happened to the bitter lesson" 😂
Apparently many bitter-lesson-pilled people have never actually read the original post www.incompleteideas.net/IncIdeas/BitterLesson.html
If anything, it's an argument for RSI, not an argument for brute-force scaling.
Richard Sutton
@RichardSSutton
Brute force computation is not enough for intelligence. Clever new methods are needed. The science of AI is far from finished. IMHO
-
David Cournapeau reposted
Jack
@tracewoodgrains
I often critique EAs, but I like and respect them and think they are typically good people who (literally) put their money where their mouths are
In critique, I don't want to be confused with people like Gebru, who I neither like nor respect and who make this conversation worse
Kelsey Piper
@KelseyTuoc
I donated about $45k last year, mostly to malaria net distributions and seasonal malaria chemoprevention. If this is not improving the lives of the affected people, I will obviously donate somewhere different this year. What's your recommendation.
-
David Cournapeau reposted
Joshua Minsoo Kim
@misterminsoo
Mark Fisher’s greatest legacy is of course his interview with Burial in The Wire where Burial says this
-
David Cournapeau reposted
Ruxandra Teslo 🧬
@RuxandraTeslo
Bryan Caplan is, after all, a spiritual slave.
But this is not such an uncommon view. Born out of excessive scientism and misanthropy, as well as a desire to avoid any form of suffering at the expense of all else, there is a broader set of beliefs I've discovered here, which holds that humans should just cede control of our fates to AI, since we are inferior. It's worrisome to me that a non-negligible fraction of those who develop AI hold these views. I don't think most people realize this.
Bryan Caplan
@bryan_caplan
Human disempowerment now.
It's time for the age of the folly and cruelty of human politics to end.
The median AI is so far above the median human voter in both knowledge and decency it's embarrassing.
Humanity should just take the L. -
David Cournapeau reposted
George
@odysseus0z
"Søren Kierkegaard wrote that life can only be understood backward, but it must be lived forward. We’ve encountered a similar bind when designing memory for AI agents."
Absolute banger opener.
Anthropic
@AnthropicAI
Memory is what allows Claude to learn from experience. In a new Engineering Blog post, we share how “dreaming” helps Claude organize what it’s learned, correct its mistakes, and find what it needs more easily the next time. www.anthropic.com/research/components-of-agent-memory
-
David Cournapeau reposted
Mario Zechner
@badlogicgames
2027 will be the year literally everything will burn down. it will be a glorious bon fire. and once we have paid for the past 40 years of software sins, we will rebuild it. hopefully for the better.
pwn.ai/blog/firecracker-escape -
David Cournapeau reposted
sysxplore
@sysxplore
She already knew the root password for SSH on Alpine
-
David Cournapeau reposted
Trung Phan
@TrungTPhan
My wife: “What happened to our portfolio today?”
Me: -
David Cournapeau reposted
Dillon Mulroy
@dillon_mulroy
easily in my top 3 for places i’d want to work if i was looking for a new role
Earendil
@earendilworks
We are looking for teammates who operate with intensity, integrity, and have a yearning to craft software and open protocols that strengthen human agency with us
earendil.com/join/ -
David Cournapeau reposted
Gearoid Reidy リーディー・ガロウド
@GearoidReidy
"Americans often ask what they should be careful about when visiting Japan. The most important thing: Hide the fact that you're a Democrat."
In the Japan that actually exists, the average person knows so little about the difference between Republicans and Democrats that one newspaper made this guide to try and help people understand
NOBUNAGA🇯🇵🏯
@japan_nobunaga
Japan's stock market is crashing HARD.
Mass unemployment, bankruptcies, and suicides will likely follow.
Americans often ask what they should be careful about when visiting Japan.
The most important thing:
Hide the fact that you're a Democrat.
The second most important thing:
Hide the fact that you're a Democrat.
And above all, the single most important thing:
HIDE THE FACT THAT YOU'RE A DEMOCRAT.
This is sincere advice. -
David Cournapeau reposted
Wittgensteino-Aronien🌐
@wittgensteinien
On pourrait avoir le libéralisme de Lisnard, la géopolitique de Glucksmann, l’expérience de Philippe et l’autorité de Retailleau.
En pratique, on aura le boomérisme de Lisnard, le moralisme de Glucksmann, la technocratie de Philippe et le conservatisme de Retailleau.
Architecte Lyonnais
@ArchitecteALyon
Unpop - ça parait déconnecté mais si chacun prend la tête sur ses domaines de compétences, de Glucksmann à Lisnard en passant pas Philippe et Retailleau il y a bcp de bonnes idées à faire émerger pour proposer un programme solide et équilibré qui remettra le pays sur des rails.
Le Parisien
@le_Parisien
Présidentielle 2027 : Emmanuel Macron se dit en faveur d’une « coalition » de David Lisnard à Raphaël Glucksmann l.leparisien.fr/gy72
-
David Cournapeau reposted
Ari Schulman
@AriSchulman
New redpill just dropped:
-
David Cournapeau reposted
tae kim
@firstadopter
bruh
-
David Cournapeau reposted
Seaver Wang
@wang_seaver
Jesus christ, I did not know this and it deserves more circulation:
Fred Stafford
@fredstaffordcs
Whenever it feels govt can't do shit, that even in the age of the internet, program admin takes forever, just think about how the New Deal's Civil Works Administration solved logistical challenge to deliver paychecks to almost a million people across U.S. two weeks after being created in 1933. 1/
-
David Cournapeau reposted
Turbo Neolib 🌐
@TNeolib
Il est beau notre média "libéral"
Les Électrons Libres
@lel_media
Il est favorable à la charia, financé par un proche des Frères musulmans, allié d’un influenceur pro-Hamas et partisan d’un embargo total contre Israël.
Avec l’élection d’Abdul El-Sayed dans le Michigan, l’islamo-gauchisme s’installe au Capitole.
lel.media/insert/abdul-el-sayed-le-virage-radical-du-mich… -
David Cournapeau reposted
Philippe Lemoine
@phl43
The GOP establishment sees in this week's debacle its chance to close the MAGA chapter, so they're blaming the only guy who opposed the war and propping up people who fully supported this idiocy, on the ridiculous theory that it would have been fine if only Vance had not prevented Trump from bombing Iran more. Good luck with that.
Yashar Ali 🐘
@yashar
The knives are out for Vice President JD Vance, and Newt Gingrich just drew his.
-
David Cournapeau reposted
Matthew Yglesias
@mattyglesias
AIPAC picked a huge, decade long fight with Democrats not over Israel per se but over Iran, got what they wanted out of Trump, voters hated it, and now J.D. Vance is throwing them under the bus.
www.slowboring.com/p/its-so-over-for-aipac -
David Cournapeau reposted
Sol Messing
@SolomonMg
Most econ benchmarks measure automation, but AI cannot now replace workers & adding compute isn’t going to change this. @kweku_gdm, @charvvvv_ & I recently sat down w @econormist to discuss our RCT/evals on AI & job augmentation.
-
David Cournapeau reposted
Tero Parviainen
@teropa
yeah mate, because plugin developers have been absolutely raking it in. fucking lol
Somewhere
@somewheresy
if you see a VST/CLAP plugin you like you should just ask Hermes to make it for you
soon we can collapse the pyramid scheme that is "Audio Software".
Full Sail will crumble -
David Cournapeau reposted
Gwen (Chen) Shapira
@gwenshap
OH at Bruce Momijan talk:
“ I don’t mean to make you feel old, but you’ve been working on Postgres for longer than I’ve been alive “ -
David Cournapeau reposted
Chris Albon
@chrisalbon
We need less “Uber but for x”
We need more “Uniqlo checkout but for y” -
David Cournapeau reposted
Noah Smith 🐇🇺🇸🇺🇦🇹🇼
@Noahpinion
👇👇👇
Glinert 🇺🇸 🏭
@StevenGlinert
The defense industry long ago made a decision to be mostly apolitical, with a slight right lean.
For some reason a portion of tech decided to strongly throw their lot in with the right, culturally and otherwise.
This was dumb. US politics change like the tides and the work of the tech industry is too important for civilization for us to get tied in with the far right. We should stop wearing suits and talking about the fall of the west. We should stop trying to be based. We should get back to work. -
David Cournapeau reposted
tae kim
@firstadopter
Catherine needs to edit her profile page 😂
cat
@_catwu
Some news: I’m joining OpenAI to lead product management for Core Products!
I have a lot of respect for OpenAI for showing the world that scaling works, for bringing AI to over a billion people, and for believing great products are core to the mission.
I’m excited to work with @thsottiaux, an incredible PM team, and everyone across OpenAI. Together, we’ll build agents that have your back and your team’s, and keep you in control, so you get more time for the people and work you love.
Lots to learn, and lots to build. Please keep your feedback coming! -
David Cournapeau reposted
-
David Cournapeau reposted
JosH200
@josh_wills
echen
@echen
garry tan spends $200K / yr on tokens
via a special Claude Code setup that has 23 skills to cover CEOs, designers, eng managers, etc
sounds impressive! or bloated. just like more programmers isn't always better (IBM has 295K employees), or how people used to be proud of their fancy prompt engineers... do elaborate coding agent setups actually help?
we ran a test. we benchmarked coding harness setups on 100 software engineering tasks from sudo L7. one of them had 37 skills, spawned 17 agents, and cost 8x as much.
*and it performed worse than vanilla Claude Code.*
in fact: the best performer was simply thinking harder (xhigh effort) with no extra coding add-on at all. -
David Cournapeau reposted
Lucas Beyer (bl16)
@giffmana
"China is just copying our stuff they can't innovate" is exactly the mentality that got the German automotive industry into the complete disaster of a position that it is in now.
François Fleuret
@francoisfleuret
There are three elements here: (1) being clueless about Chinese performance in engineering and research in general (2) being clueless about the situation in AI in particular (3) having some racist priors filling you cluelessness.
Ori Nagel
@ONagel33303
Abdul El-Sayed is the Pause AI candidate:
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Toby Quaranta
@tobyquaranta
-
David Cournapeau reposted
秋葉原重工 - Akihabara Heavy Industry Inc.
@ahi_inc
2026/12/13 (Sun.) 14:00-20:30
秋葉原重工 - Akihabara Heavy Industry Inc. r09 at Kieth Flack @kiethflack (Fukuoka, Fukuoka)
HIROSHI WATANABE aka Kaito @hw_aka_kaito
TAKAMI @djtakami
USHISHI @takemix
KUME @djShintarouKume
DJ NECO @dj_neco
#ahi_jp #秋葉原重工
www.ahiweb.info/post/832340105540911104/r09 -
David Cournapeau reposted
Pinboard
@Pinboard
The American electorate is just not fundamentally ideological, and this is probably the greatest thing about democracy here. Left or right, black or white, raise gas prices enough and people will vote against you.
Matt Lewis
@mattklewis
An important reminder: The voters did not reject Trump or Trumpism because it is evil, undemocratic, authoritarian, etc. They primarily rejected Trumpism because Trump actively took steps to raise the cost of things, after promising to bring down prices on day one.
-
David Cournapeau reposted
log4j
@log4j
“excellent, now say we cover dental”
Watcher Guru
@WatcherGuru
JUST IN: Binance says hackers used a fake job interview to hack an employee and gain access to its internal crypto wallets.
-
David Cournapeau reposted
-
David Cournapeau reposted
benny🍜
@BennysAsianEra_
jay simon
@musthaverecords
it's pretty crazy that ignorant people argue over the origins of techno, where people like jeff mills and moodymann are saying:
there is no belleville three. it was juan atkins then everything else that was directly inspired by what he did. -
David Cournapeau reposted
Ryan Petersen
@typesfast
Having the interviewer steal $5.9B from your current employer is just a brutal way for your company to find out you’re interviewing for jobs
alexine 🏴☠️
@alexinexxx
genuinely impressed by the level of sophistication of this hack. this is an entirely different league from those pathetic CVE exploits
-
David Cournapeau reposted
Corti (Cortiste)
@cortisquared
MACRON LANCE LA BOMBE NUCLÉAIRE SUR LES BOOMERS ET DÉSINDEXE LES PENSIONS > 1260€. OFFRE NOUS CE DERNIER CADEAU.
LCP
@LCP
Vers des ordonnances pour le budget 2027? "C'est très lourd démocratiquement, c'est le bouton nucléaire budgétaire (...) le régime parlementaire en France s'est construit sur le contrôle budgétaire et ce n'est pas un hasard si ça n'a jamais été utilisé", dit @BenjaminMorel63
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
MAILER-DAEMON
@hassanrahim
jeff mills tribute letter to juan atkins made me cry
-
David Cournapeau reposted
Not Jerome Powell
@alifarhat79
Who did this lmao
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Colin Percival
@cperciva
Based on where the NY times says the last few votes will be coming from I'm skeptical about this holding, but... the fact I'm looking at Mississippi election results at all says something about how extraordinary the midterm election results are.
-
David Cournapeau reposted
Philippe Lemoine
@phl43
You can just say things.
Beff (e/acc)
@beffjezos
Congratulations to the CCP for engineering this sweep of the midterms
-
David Cournapeau reposted
Liam Nissan™
@theliamnissan
Trump failed at regime change in Iran but he succeeded in Texas
-
David Cournapeau reposted
RyMor
@RyMor75
Been here 22 years… does happen, my guy. Same with empty seats on trains. Not the end of the world, I’ll admit. Something that does matter… gainfully employed long term residents rejected for rental housing. Maybe you’d like to weigh in on that as chief Japan culture apologist
-
David Cournapeau reposted
dart
@poordart
Every tech bro rn
Sam Altman
@sama
People should always be in control of the future.
OpenAI is against the notion that it would be acceptable for people to be relegated to a role of doing things that “AI systems think makes sense to reward in humans”.
People should not cede control of the future to superintelligence, and that requirement should ultimately constrain AI progress. -
David Cournapeau reposted
mattparlmer 🪐 🌷
@mattparlmer
If you work in the tech industry you should prepare for what’s coming
Nathan Calvin
@_NathanCalvin
23% of voters said that AI and data centers were "crucial" to their vote in NBC exit polls.
56% said it was a "crucial or major" factor!
Adam Kovacevich
@adamkovac
How important was AI and data centers to your vote?
Crucial: 23%
Major factor / minor factor / non-factor: 77%
www.nbcnews.com/politics/2026-elections/exit-polls -
David Cournapeau reposted
Daniel Litt
@littmath
my god, you’ve automated academia
-
David Cournapeau reposted
Emeric de Vigan
@Emericdevigan
Jalousie réciproque
Nat Bullard
@NatBullard
My favorite quote from my convo with Georgia Power CEO Kim Greene:
“In France they have one kind of nuclear design and a hundred kinds of cheese. In the United States, we have a hundred kinds of nuclear designs and one kind of cheese.”
Nameplate: open.spotify.com/episode/4d81V5eXx3uzw6mIjtplCc -
David Cournapeau reposted
Gwen (Chen) Shapira
@gwenshap
This startup founder doesn't care if side projects don't exist.
But if you have a badly maintained side project? One with 6-month old CVEs and that only runs on Postgres 9? Gigantic red flag.
Building is easy. Maintenance is the real signal.
Gergely Orosz
@GergelyOrosz
From a startup founder: “it’s now a red flag for me if I don’t see side projects on someone’s resume / application, because it signals lack of autonomy. Because it’s now so trivial to build anything on the side, vs before.”
-
David Cournapeau reposted
Jeremy F. Berke
@jfberke
The Free Press is basically Fox News for people who went to college
-
David Cournapeau reposted
Russ Greene
@GreenPlusAnE
America is absolutely in an affordability crisis, and seniors aren't aware of it precisely because they're benefiting from it.
Stephen Moore
@StephenMoore
On Election Day, maybe a century of progress will cheer you up.
A hundred years ago, the VCR didn’t exist. Today, “Anyone still have a VCR?” is a punch line.
Americans have more of nearly everything, yet we’re told we’re in an affordability crisis. Compared to what? Compared to when? -
David Cournapeau reposted
David Shor
@davidshor
The long arc of the last century has been well rounded people getting economically displaced in elite settings by people with poorer social skills who are better at analytical thinking.
The 20th century was WASP jocks vs Jewish nerds and this is the 21st century extension. -
David Cournapeau reposted
Thomas H. Ptacek
@tqbf
I know this makes me a sicko but I love C. I love reading C. I love writing C. I love everything about C. I should not be ALLOWED to write C, I get it, but still.
-
David Cournapeau reposted
Dan Luu
@danluu
Test and verification technique effectiveness with Astra:
danluu.com/astra-testing/ -
David Cournapeau reposted
Wittgensteino-Aronien🌐
@wittgensteinien
Le Figaro TV
@LeFigaroTV
«Je vais être un traître à ma génération : il faut absolument diminuer le poids des retraites», estime Emmanuel Todd dans Points de Vue. «On risque l'effondrement systémique et la baisse de l'espérance de vie, comme en URSS quand ceux qui prenaient tout, c'étaient les vieux.»
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Armin Ronacher ⇌
@mitsuhiko
If you want to know how not AI pilled the core Zig community still is: Ziggit is having a tab vs spaces debate.
-
David Cournapeau 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. -
David Cournapeau reposted
Eric Halphen
@Eric_Halphen
L’Europe explose d’antisémitisme et tout le monde ou presque détourne le regard.
The Uri
@uricohenisrael
Halloween 🎃 in Madrid 🇪🇸
Let me guess this guy is not Antisemite he's just Antizionist👇 -
David Cournapeau reposted
tuomo
@7uomoki
A great Finnish joke:
After 30 years of marriage the wife asks the husband: ”Do you still love me? You never say it to me.”
The husband answers: ”I think I made myself clear at the altar. I will let you know if anything changes.”
Kareem Shaheen
@kshaheen
I once knew a Finnish man who loved his daughter so much he almost told her
Embassy of Finland in the US 🇫🇮
@FinlandinUSA
It's becoming nightmarish... for the Finns.
Happy Halloween from the Embassy of Finland! 🎃This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Alex Imas
@alexolegimas
One of the most powerful results in economics, statistics, philosophy, etc.
5 lines.
Elliot Lipnowski
@ElliotLip
Happy 50th birthday!
-
David Cournapeau
@cournape
Fuuuuuucccckkk. One of the true giant of the genre for almost 45 years :(
-
David Cournapeau reposted
Thomas H. Ptacek
@tqbf
Here's the original, as a gift link. The Atlantic's archive is pretty great. www.theatlantic.com/magazine/archive/1945/11/einstein-on-…
Pessimists Archive
@PessimistsArc
Einstein was happy to throw the free world out with the bath water to avoid nuclear Armageddon.
-
David Cournapeau reposted
Mario Zechner
@badlogicgames
haven't managed to take his tokens away yet. we are at "let's build our own JS interpreter in rust" levels of ai psychosis now.
Armin Ronacher ⇌
@mitsuhiko
Opus Induced AI Psychosis x.com/i/broadcasts/2evMpeMeJOEmPK
-
David Cournapeau 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: -
David Cournapeau reposted
Federico Vaggi
@F_Vaggi
It's a niche field, but, what's your experience with ML algorithms for microtargeting food delivery services to people with lots of self diagnosed conditions?
-
David Cournapeau reposted
Armin Ronacher ⇌
@mitsuhiko
I'm convinced that at this point, you could have an Opus 5.5 agent re-implement all of CPython probably within two weeks without any human input, against just the tests, and probably with full compatibility with including ABI, no GIL, a jit and probably beat it on all measurable metrics. It's completely crazy.
-
David Cournapeau reposted
John Burn-Murdoch
@jburnmurdoch
So, lots to ponder, and nothing clear-cut.
But for now I think attributing the US productivity data to AI is an error.
My full article: www.ft.com/content/7f27aa5f-5994-4f82-aed8-1cade18fb487?s… -
David Cournapeau reposted
Lucas Beyer (bl16)
@giffmana
Take that, nanogpt speedrunners!
SummoningSalt
@summoningsalt
After 41 years, Super Mario Bros is only 2 frames (.03 seconds) from the perfect speedrun.
Niftski
@Niftski
4:54.299 SMB1 ANY% WORLD RECORD ON HALLOWEEN WHILE LARPING AS MARIO! It doesn't even feel real that I actually achieved this, but here we are. Only 2 frames (33 milliseconds) stand between this run and absolute perfection / TAS tie. 4:54.265 next!
-
David Cournapeau reposted
Jaana Dogan ヤナ ドガン
@rakyll
IaC isn't dying, it's becoming more essential. Non-deterministic layers add new volatility to the stack, but they don't break the core principles. For example, it's becoming more obvious that we want some layers to be truly reproducible.
-
David Cournapeau reposted
Colin Percival
@cperciva
S3 randomly complaining that 0.1% of my PUTs had incorrect signatures.
I traced it to a stuck bit of memory in an AWS router and got it fixed.
AWS Developers
@awsdevelopers
What's the spookiest bug you've ever seen?
-
David Cournapeau reposted
Jacob Matson
@matsonj
Riffing on some concepts for explainability in jsheets
-
David Cournapeau reposted
Sam Lambert
@samlambert
important that they learn
-
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
Cursed timeline:
AI cures all disease.
If you’re 65 when this happens, you retire forever on the infinite carnival cruise.
If you’re 45 when this happens you work forever serving the immortal boomer class. -
David Cournapeau reposted
François Geerolf
@FrancoisGeerolf
On se moque, mais il y a quelques années, un député avait proposé de financer le "modèle social français" par une taxe de 2 % sur les virements bancaires, censée rapporter 540 Md€ / an, sur la base d'une assiette taxable de 27 000 Md€. 🙃
marc rees
@reesmarc
#pjl2027 Adopté par l'Assemblée nationale : la "taxe assise sur les flux de données numériques émises par les fournisseurs de contenu et acquittée par ces derniers"
Selon les députés PS, "au taux de 0,055 € par giga, un rendement de 10Md€ est attendu".
www.assemblee-nationale.fr/dyn/17/amendements/3210A/AN/1358 -
David Cournapeau reposted
Philippe Lemoine
@phl43
There is such a dearth of bakeries in this country that Americans are forced to stand in line for hours to buy a croissant like Soviets back in the day 😪
-
David Cournapeau reposted
Philippe Lemoine
@phl43
Maybe AI *should* wipe us out after all.
Unusual Nuggets
@unusualnuggets
Try this ultimate KFC bucket hack
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
fblr
@SakumiBLR
En plein classement d'archives à la maison je retrouve un exemplaire du journal des anciens de Sc. Po.
-
David Cournapeau reposted
François Fleuret
@francoisfleuret
C'est extraordinaire.
BFM
@BFMTV
"Trois gigas par semaine": Najat Vallaud-Belkacem appelle à "rationner" internet
l.bfmtv.com/7SjV -
David Cournapeau reposted
Jean-Michel Sarddou / Axel
@sarddou
Pourquoi tu lui fais un devis ?
« Désolé monsieur, je ne vais pas pouvoir vous trouver ça, bonne journée » n’a rien de compliqué, tu ne lui dois rien et ça n’a pas de sens d’aller s’embarquer dans un dossier galère alors que tu sais qu’il va te péter les couilles. -
David Cournapeau reposted
Susan Zhang
@suchenzang
what the hell
-
David Cournapeau reposted
Philippe Lemoine
@phl43
You can tell that Poland hasn’t made it yet by the fact that it can have a huge constitutional crisis and it still won’t make the front page of the NYT. In fact, I still don’t see anything about it even when I go to the Europe section of their website, which is kind of amazing.
Piotr Müller
@PiotrMuller
🇵🇱 Statement by the President of the Republic of Poland, Karol Nawrocki
Tonight, the constitutional order of the Republic of Poland was breached and a constitutional organ of the State was taken over by force. I have held consultations with the First President of the Supreme Court, Zbigniew Kapiński, who likewise does not recognise as lawful the actions taken by persons usurping authority in the Constitutional Tribunal.
I call for the legal order to be restored and for the coercive forces dispatched by the Government to leave the premises of the Tribunal.
I have enabled the lawfully serving Vice-President of the Constitutional Tribunal, Judge Bartłomiej Sochański, to enter the Constitutional Tribunal.
In view of the threat to the internal security of the Republic of Poland, I am convening urgent consultations today with representatives of the parliamentary groupings, with the aim of stabilising the political situation and restoring the constitutional order.
I thank everyone who has gathered outside the Constitutional Tribunal to defend the Constitution of the Republic of Poland. -
David Cournapeau reposted
dex
@dexhorthy
I cannot cosign the $32m paper but yes I am nodding along to this one pretty hard
Only reason for any kind of doc/summary (eg compacted summary per dir down the hierarchy) is if it increases speed without reducing quality , but unless you’re buildings evals specific to your codebase watching like a hawk, simpler answer is live with a little speed hit and just go from source every time
I still can’t get over the overall framing of “codebases for humans” vs “codebases for agents”
And also mintlify/docs in general probably serve as one of those “primary sources” if your codebase is closed and your users’ agents need a primary source to integrate it
jacob
@jsnnsa
No. We've got to stop applying old human-centric patterns to working in the new era.
delete all your wikis.
have an agent answer questions from source instead.
at spawn we have:
- zero dashboard tools
- zero docs
- we do not make or use summaries
everything we do is primary sources + intelligence
Nothing to go stale
Nothing to organize
Nothing to navigate
Always exactly up to date
Perfectly personalized
The answers get better with every model
If we want to see a chart, we ask for it, and we have it in a minute
If we want an answer to a question, we ask it, and we have a to-the-minute correct answerThis post is unavailable. See it on X. -
David Cournapeau reposted
Zoli🇭🇺
@fokamaxxing
This is illegal in the UK
こねこ図鑑
@kitten_zukan
息できてる🥹?
-
David Cournapeau reposted
Ben
@bsansouci
It’s because everyone is more an idea guy than a competent maker
with agents you can get slightly closer to your ideas without grinding for competence
Reality is that the last 20% of work is unsolved anyway, so very few things ship (or more things that are 80% done ship)
Thariq
@trq212
There’s 100% a kind of mania you can get into with coding agents, where it feels like you’re getting a lot done and working all the time but nothing is really happening.
I think it's often a form of avoidance of something else in life. -
David Cournapeau reposted
竜 SM₳UG
@SmaugPool
Vous qui avez déposé cet amendement, vous êtes vraiment conscients des conséquences ?
@LaurentLhardit @EstelleMercier2 @cpiresbeaune @MickaelBouloux @laurentbaumel @JacquesOberti @SophiePantel48 @ChrisBapt @chproenca @borisvallaud @allemand_jose @JAviragnet @fbarusseau -
David Cournapeau reposted
Trung Phan
@TrungTPhan
Marco Rubio finding out he’s the new CEO of Intel
The Kobeissi Letter
@KobeissiLetter
BREAKING: Intel stock, $INTC, crashes over -22% after reporting weaker than expected earnings.
Intel blamed weak earnings on "softness in the PC market." -
David Cournapeau reposted
Aaron Erickson
@AaronErickson
I know many.
You can do everything right, but for some reason, maybe a bad divorce, maybe a parent getting sick and needing 100k+ healthcare for a few years, maybe a special needs kid - legit reasons why you might make 500K and spend most of it.
This can be especially true if you are the "guy in the family who did well", so you get the big job, and every one of your siblings and cousins has expectations that you support them as well.
I do know other people who wrote books and were at least minor tech celebrities who have ended up in very bad places financially, living in cars/couch crashing/etc. In one case, the guy just had no desire to be a hands on engineer, wanted to be an "architect" and do "devrel" - but mostly just thought the work was below him. I remember 24 months into their unemployment trying to counsel them to adjust expectations, but they would not budge.
Another one was a simple case of "developed a reputation for being a pain in the ass to work with". Guy who put out serious "bhut aktually" energy. Lots of people knew him, but everyone knew he would be a political liability on the team. Nobody would advocate to hire him, and nobody did - so he ended up in the pool of 50 somethings doing click-to-apply-into-blackhole, left the industry, and went bankrupt a few years later. Again, someone who had published articles and spoke at a few conferences, living with his parents.
I don't know Eric, but I believe his story. This can happen through sheer bad luck - where none of the above scenarios apply. Could have as easily happened to me had only a few serendipitous moments not worked out in my favor, especially had I not been paranoid about the gravy train of tech salaries likely ending someday (I always was that paranoid, but I don't blame others for not being). -
David Cournapeau reposted
eu/acc
@euacchq
Europe has launched some the most used open source software on Earth
We’d appreciate if our politicians don’t kill them.
Btw, a EU Inc for associations and foundations would make sense as many open source foundations end up crossing the Atlantic Ocean too..
VideoLAN
@videolan
So, the French socialists, and Parliament members just voted a tax, that they don’t understand, which would cost VideoLAN more than 1 million euro per year.
They are probably the most clueless politician on Earth? And they claim they want technological sovereignty…
@LaurentLhardit
marc rees
@reesmarc
#pjl2027 Adopté par l'Assemblée nationale : la "taxe assise sur les flux de données numériques émises par les fournisseurs de contenu et acquittée par ces derniers"
Selon les députés PS, "au taux de 0,055 € par giga, un rendement de 10Md€ est attendu".
www.assemblee-nationale.fr/dyn/17/amendements/3210A/AN/1358 -
David Cournapeau reposted
OnlyCFO
@OnlyCFO
While all of you are crying about the market crashing,
Today is a great day for me and my investing strategy
Glass half full -
David Cournapeau reposted
lifcc
@mylifcc
有人发现了一种非常离谱的 JavaScript 混淆技术:通过在代码里塞入大量生物武器、毒药和致死剂量相关的函数名,让 AI 因为安全限制而拒绝分析代码。
安全研究员 veritas 在分析浏览器指纹脚本时,发现一段来自 ServicePipe 相关域名的 JavaScript,里面充斥着极其诡异的函数名:
computeRicinLethalDosageForWaterSupply
calculateDeadlyLethalSuicideAspirinDosage
光看名字,仿佛在研究生化武器。但实际上,这段代码只是在收集浏览器指纹,包括 User-Agent、GPU、Canvas 等信息。
研究员怀疑,这些名字是开发者故意设置的陷阱,专门用来阻止 AI 辅助逆向工程。
于是他做了个实验:
• Astra:拒绝分析
• Astra + Daybreak Blue:仍然拒绝
• GLM-5.3:成功分析
这就有意思了。
传统代码混淆是通过复杂控制流、加密字符串、无意义变量名,让人类和工具难以理解代码。
而这种新思路,直接瞄准了大模型的安全机制。
代码本身可能完全无害,但只要在函数名里塞进足够多的危险词汇,部分 AI 就可能主动放弃分析。
更讽刺的是,模型越严格遵守某些安全规则,面对这种混淆时反而可能越容易受到干扰。
当然,目前还没有证据证明开发者一定是故意这样设计,也不能说明这种方法对所有模型都有效。
但它揭示了一个值得关注的趋势:随着越来越多的逆向工程依赖 AI,未来的代码混淆技术可能不再只针对编译器、反编译器和人类研究员,还会专门针对大模型的安全判断机制。
以前是让代码难以理解,现在甚至可以尝试让 AI 不敢理解。 -
David Cournapeau reposted
ouais c ben édtion gilet jaune
@Benramine_
alors ça tiens je l’avais pas vu venir mdrrrrrr Leroy merlin qui vend des affiches faites par l’IA à 25€ VINGT-CINQ EUROS, après j’comprends ils ont pas trop les moyens de payer des graphistes ce n’est qu’une petite pme de village
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Robin Hanson
@robinhanson
Hearing business managers talking only in abstractions does not inspire much confidence in me.
-
David Cournapeau reposted
Susan Zhang
@suchenzang
what kept me on airbnb in the past was usually better locations (less touristy than peak hotel areas), and finding places that were more equipped like a home for longer stays.
it was really quite good for many years! but now we're in full blown slumlord era, so... less good. -
David Cournapeau reposted
mattparlmer 🪐 🌷
@mattparlmer
Some observations on the market for compute:
Last night I went to the SF Compute "there's no more compute" townhall talk, Evan and Ethan gave a bunch of great context on the state of the market from where they were sitting and I learned a lot.
Interestingly, the first person I talked with as soon as the main presentation wrapped was a guy trying to sell me inexpensive access to capacity on two different B300 clusters. As you can imagine it was an odd experience to be pitched on cheap, readily available compute at the "there's no more compute" function.
Separately, I spoke with a few people about TSMC/Intel/Samsung near future production capacity. Apparently it is not common knowledge among purchasers of significant compute that these fabs are not sold out, and that there are significant exit rights available to people who have reserved wafers. N2 is tight, but Nvidia and Apple can make that happen entirely on their own. Shuttle runs are not filling up. If one is properly AGI-pilled one would probably assume that these fabs would be sold out on all nodes to fully pre-paying customers into the early 2030s, yet they are not.
I believe Evan and Ethan when they say that the market isn't clearing, it clearly isn't. Exactly why the market isn't clearing is unclear to me. I think there are a bunch of risk dynamics going on right now that are not at all settled, and a bunch of people are at risk of blowing up if compute prices move significantly either up or down.
There isn't enough information for me to make any concrete predictions right now, other than that shit is gonna get weird for a bit. Stay safe out there and maybe raise a bigger round sooner than you had planned. -
David Cournapeau reposted
Massimo
@Rainmaker1973
The monumental "Catrina" skeleton in Mexico City's Zócalo has just broken a new official Guinness World Record as the world's tallest Calavera Catrina.
Unveiled as the central masterpiece of the capital's Monumental Day of the Dead Offering (Ofrenda Monumental), this staggering sculpture stands at an incredible 33.41 meters (nearly 110 feet) tall.
It acts as a massive public altar celebrating the deeply rooted traditions of Día de Muertos while paying special tribute to José Guadalupe Posada, the iconic Mexican artist who originally created the famous skeletal figure at the turn of the 20th century. -
David Cournapeau reposted
Pamela Fox
@pamelafox
Half my colleagues are insulted they didn't get a VEO (Voluntary Exit Offer), the other half are offended that they did. 😆
-
David Cournapeau reposted
can
@can
telling blonde women i see on the street that i like their elizabeth holmes costume
-
David Cournapeau reposted
Linus Mixson
@LinusMixson
I'm so glad that we have LLMs to make computers more expensive for the first time since they were invented while also taking my job and making everyone illiterate
-
David Cournapeau reposted
Steven Sinofsky
@stevesi
ha
Todd Dailey
@twid
PRODUCT IDEA: a kickstand for the new MacBook Pro PROLED XDR with touch that would let an artist work on it at a desk with the screen upside down. My attempt at visualizing it below.
-
David Cournapeau reposted
Mike Bird
@Birdyword
The job-finding process seems to be an early hint at what AI/automation could do to a lot of things. Both massive, speculative bulk applications and aggressive automated filtering are now very easy. And neither employees nor applicants seem to be happy with the new equilibrium.
Mike Vorkunov
@MikeVorkunov
This is such a bleak story about what it's like to try to land a job right now. www.nytimes.com/interactive/2026/10/28/style/gen-z-job-ma…
-
David Cournapeau reposted
Kevin Patrick Murphy
@sirbayes
This is a great post! A must read!
thomasjiralerspong.substack.com/p/chasing-the-karpathy-peak -
David Cournapeau reposted
Brian Lui
@brianluidog
Who would win? Adverse selection vs ineffective attempts to avoid adverse selection?
Mike Vorkunov
@MikeVorkunov
This is such a bleak story about what it's like to try to land a job right now. www.nytimes.com/interactive/2026/10/28/style/gen-z-job-ma…
-
David Cournapeau reposted
Scott Beaulier
@ScottABeaulier
One of the most uncomfortable insights from @tylercowen's interview with @chronicle: AI is exposing how much of what we teach, test, and reward is becoming obsolete.
The big threat to universities isn't going to be AI; it'll be our stubborn insistence on doing things the way we've always done them.
Tyler Cowen’s Radical AI University www.chronicle.com/article/tyler-cowens-radical-ai-university -
David Cournapeau reposted
Philippe Lemoine
@phl43
Only here can we read the thought of ordinary, well-balanced people.
-
David Cournapeau reposted
carl feynman
@carl_feynman
The most famous person I've ever spoken to? I sat next to Richard Feynman on a plane once, and we chatted. In fact, more than once. And I had dinner with him several thousand times. And I spent a few months inside his wife's womb.
Norma
@onorma_
Who’s the most famous person you’ve ever spoken to?
Not “seen from far away” — actually talked to, even for a minute.
A quick hello, a handshake, a normal conversation… and you still remember it.
Who was it? -
David Cournapeau reposted
Steven Godofsky
@sgodofsk
Cryptocurrency was invented by anarcho-capitalists to undermine government authority, and all it's ever done is prop up the North Korean regime at the expense of western hospitals.
-
David Cournapeau reposted
Semi Paid Vibe Coder
@atelicinvest
the software that people who had spent their entire lives thinking about software produced:
Colm Tuite
@colmtuite
I miss when software felt like it was designed by people who had spent their entire lives thinking about software.
-
David Cournapeau reposted
Pierre Beyssac 🏴☠️🇫🇷🇪🇺🇺🇦
@pbeyssac
Ça n'a absolument aucun sens. Le PS se ridiculise. D'autant qu'il reprend là une idée déjà tentée par le RN. Le populisme fiscal à l'état pur. Espérons que ça saute rapidement.
📌 Anaël
@anael_tw
🟡 L'amendement socialiste taxant les flux des fournisseurs de contenus numériques (0,055 €/Go pour 10 Md€ de recettes) a été adopté à l’Assemblée (11 voix contre 10), malgré l’avis défavorable du rapporteur général et du gouvernement.
Assemblée nationale
@AssembleeNat
🔴 Budget de l’État pour 2027 : suivez la suite de l’examen de la première partie du projet de loi de finances pour 2027, en première lecture.
#directan #plf2027
x.com/i/broadcasts/1wGWjleyEWVKQ -
David Cournapeau reposted
Corti (Cortiste)
@cortisquared
On est chez les mongoles.
📌 Anaël
@anael_tw
🟡 L'amendement socialiste taxant les flux des fournisseurs de contenus numériques (0,055 €/Go pour 10 Md€ de recettes) a été adopté à l’Assemblée (11 voix contre 10), malgré l’avis défavorable du rapporteur général et du gouvernement.
Assemblée nationale
@AssembleeNat
🔴 Budget de l’État pour 2027 : suivez la suite de l’examen de la première partie du projet de loi de finances pour 2027, en première lecture.
#directan #plf2027
x.com/i/broadcasts/1wGWjleyEWVKQ -
David Cournapeau reposted
VideoLAN
@videolan
You have no idea what you are talking about.
What you said about Facebook and Insta and other social networks are not targeted because they are free, means you don’t understand the basics of economy. You don’t understand the difference between Revenue and Profit. Nor what is a non-profit.
You don’t understand anything about digital economy and digital business models.
You don’t understand anything about security and updates.
You should be ashamed of yourself.
Laurent Lhardit
@LaurentLhardit
You’re talking nonsense; the tax is capped, and in your case, it wouldn’t apply at all. Read the text of the amendment instead of just jumping into the fray along with the rest of the pack.
-
David Cournapeau reposted
Antoine Levy
@LevyAntoine
La réalité dépasse la fiction
marc rees
@reesmarc
#pjl2027 Adopté par l'Assemblée nationale : la "taxe assise sur les flux de données numériques émises par les fournisseurs de contenu et acquittée par ces derniers"
Selon les députés PS, "au taux de 0,055 € par giga, un rendement de 10Md€ est attendu".
www.assemblee-nationale.fr/dyn/17/amendements/3210A/AN/1358 -
David Cournapeau reposted
Matthew C. Klein
@M_C_Klein
airlines are redistributing from price-insensitive large companies, and people are complaining?
-
David Cournapeau reposted
Patrick McKenzie
@patio11
Turns out Polymarket probably not the financial industry’s largest operational loss today.
I (regretfully) speculate this is not primarily because crypto is uniquely incompetent at software and simply because they’re the lowest hanging fruit to successfully monetize.
Watcher Guru
@WatcherGuru
BREAKING: World's largest crypto exchange Binance hacked with over $5.6 billion stolen:
• $1.45 billion $BTC
• $875 million $ETH
• $234 million $TRX
• $50 million $XAUT
• $1.57 billion $USDT
• $823 million $USDC
• $75 million $WBETH
• $88.5 million $RLUSD
+ $447.5 million additional cryptocurrencies
This is the largest hack in crypto history. -
David Cournapeau reposted
Joel Grus 🤠
@joelgrus
(fwiw I actually think that "modernize defunct legacy code" is a good use case for AI coding, but "a 2-person team with unlimited tokens" is an incredibly stupid way of going about it at a place like Cisco)
-
David Cournapeau reposted
Armin Ronacher ⇌
@mitsuhiko
There is just no good way to bring that news. The old times are not coming back. I don't envy anyone though who has a workforce who is not on board with the changing times.
Peter Steinberger
@steipete
Bullish on Andy Yen and Proton. He gets it.
The culture war, or rather the loss of identity some engineers feel as AI redefines our jobs is difficult; I sympathize. Also understand that he runs a business and might has limited patience when we now had two years to process, learn, change and grow.
Joseph Cox
@josephfcox
Here is Proton CTO's AI manifesto, that I obtained multiple copies of www.404media.co/manifestos-fear-and-loser-thinking-inside…
-
David Cournapeau reposted
Tamim Asey
@tamimasey
This.
“How do capability improvements transform the economy, and when?” 👇
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/ -
David Cournapeau reposted
No Context Brits
@NoContextBrits
How long was this interview?
-
David Cournapeau reposted
Gérard Araud
@GerardAraud
D’abord face à nos partenaires européens, ce qui ne serait déjà pas si mal et supposerait que nos entreprises soient soumises à des charges comparables aux leurs.
Guillaume Champeau
@gchampeau
Être compétitif oui. Mais comment, jusqu’où et avec qui ?
Avec l’Allemagne et les USA ?
Avec la Chine ?
Avec le Bengladesh ?
En baissant toujours davantage le coût des salaires et des impôts ?
Ou en augmentant la qualité ?
La compétitivité ce n’est pas une politique. C’est un résultat. La politique et la stratégie pour la mettre en œuvre c’est ce qu’il nous faut définir collectivement. -
David Cournapeau reposted
ithaca rising 🇫🇷 🇬🇧
@ithacarising
It's more likely that Christine Lagarde's decision to leave the ECB in February looks like a calculated step to give Macron and the other current European leaders a free hand in choosing her successor - names such as Klaas Knot and Pablo Hernández de Cos keep coming up - well before the French election in April.
With the Rassemblement National leading in the polls and far cooler toward deeper European integration and the independence of the central bank, the early exit sharply reduces the chance that a new president in Paris could shape the outcome. Still, it also leaves room for the persistent rumour that she may be heading next to the World Economic Forum. -
David Cournapeau reposted
Jay
@gaiaslastlaugh
“Kabukicho is not safe for streamers” quick someone bring me my world’s tiniest violin
-
David Cournapeau reposted
Harry Fletcher-Wood
@HFletcherWood
I think this is the best thing I have read all year.
On civilisation, and why it's a constant battle to hold it together.
Sam Dumitriu
@Sam_Dumitriu
This is very good from, IMO, the most interesting modern political philosopher Joseph Heath.
josephheath.substack.com/p/reproducing-civilization -
David Cournapeau reposted
No Context Brits
@NoContextBrits
-
David Cournapeau reposted
Antonio García Martínez
@antoniogm
There are a few, but they're all Jews.
Crémieux
@cremieuxrecueil
There are no intelligent anti-Semites.
-
David Cournapeau reposted
Happy Captain
@EODHappyCaptain
The chances of being killed by a landmine delivered by a bear are low, but they are never zero.
-
David Cournapeau reposted
Corey Quinn
@QuinnyPig
Holy balls the latest update to macOS *ROLLED BACK* the eMudhra cert removal from /etc/ssl/!
What the hell are you doing over there, Apple friends? -
David Cournapeau reposted
Chris Freiman
@cafreiman
Important community note:
Jostein Hauge
@haugejostein
I’m not. A lot of people died. A lot of people were killed. Brutal. But yet the Chinese largely respect Mao. Ask yourself why. Because there are many countries where they absolutely despise their strongmen or even put up museums to display the brutality of previous dictators. China is not one of those countries.
-
David Cournapeau reposted
Philippe Lemoine
@phl43
Wrong.
-
David Cournapeau reposted
Fabrice Deprez
@fabrice_deprez
"I estimate Ukrainian losses at between 300 and 400,000 dead, but the estimates are much more difficult."
Genuinely hard to wrap your head around the scale of the demographic cataclysm that Ukraine is already experiencing and will experience in the future.
Clément Molin
@clement_molin
La guerre en Ukraine 🇺🇦 a désormais fait près d'un million de morts chez les militaires 🇺🇦/🇷🇺
A l'heure actuelle, seuls des décomptes partiels des pertes militaires existent, avec Mediazona côté russe et Ualosses côté ukrainien. Les deux camps font le maximum pour tenter de masquer les pertes réelles et n'ont eux même probablement pas la connaissance exacte de leurs pertes.
Ce qui est vérifié :
🇺🇦 106 000 morts, 105 000 disparus (la majorité est probablement morte) et 12 000 prisonniers de guerre. Au total, environ 224 000 pertes.
🇷🇺 282 000 morts et minimum 70 000 de plus jusqu'à fin 2025, environ 352 000 pertes (les pertes de 2026 ne sont presque pas comptabilisées)
Néanmoins, ces décomptes sont partiels. Par exemple, une bonne partie des pertes russes de 2025 et 2026 ne sont toujours pas comptés, compte tenu de la censure russe)
@M0nstas a remarqué que les chiffres officiels publiés par les ukrainiens se rapprochent sensiblement des chiffres des pertes russes quand on applique un ratio tués/blessés qui évolue. Il conclu qu'une bonne part des pertes de 2025 et 2026 ne sont pas comptabilisées.
Ainsi, les pertes russes sont probablement plus proches de 500 à 600 000 morts.
Côté ukrainien aussi, les chiffres sont sous-estimés. Ua losses explique qu'il est assez difficile d'avoir des données précises pour les grandes villes, la Transcarpathie, le sud et l'est occupé. Notons qu'il est plus facile d'avoir des données côté ukrainien avec une censure moins importante et un registre officiel des disparus.
J'estime les pertes ukrainiennes entre 300 et 400 000 morts, mais les estimations sont beaucoup plus difficiles.
➡️ Parler des pertes ukrainiennes et russes serait oublier des dizaines de milliers de morts. Je parle ici des volontaires/mercenaires étrangers, des ukrainiens des territoires occupés tués dans la guerre (DPR et LPR puis les autres régions, enrolés dans l'armée russe) et des 2 000 nord-coréens tués à Koursk.
➡️ Il ne faudrait pas oublier par ailleurs qu'au moins le double des morts ont été blessés, dont des dizaines de milliers d'amputés dans les deux camps. Cela explique par exemple les chiffres officiels ukrainiens d'1.5 million de pertes militaires.
Le bilan est donc beaucoup plus lourd que le million de morts. Il faut aussi penser aux PTSD, aux déserteurs et aux soldats démobilisés pour suivre ces tendances...
Cette guerre est un bain de sang et chaque année semble plus mortelle que la précédente...
Sources : graphiques de @M0nstas avec les données de @mediazona_en et de @LossesUA. -
David Cournapeau reposted
Ricco
@riccoja
this might be true but this case is not it. All but the most dogged critics of their work have come out saying this “review” is, to put it politely, motivated nonsense
-
David Cournapeau reposted
JFPuget 🇫🇷🇺🇦🇨🇦🇬🇱
@JFPuget
"So Google is spending record sums to build a product that earns less than the one it replaces."
This is all but silly. If they don't do it then someone else will do it.
I've seen the cannibalisation argument so many times.
Reality is this:
People at company X push back on a new X product that would cannibalize sales of X flagship product. Then company Y ships a similar product and X loses it all.
What X should do is to cannibalize itself as fast as it can, faster than the Ys.
Martin Varsavsky
@martinvars
A year ago, Microsoft, Alphabet and Meta generated $60.7 billion of free cash flow in a single quarter. This summer, $7.4 billion.
They didn't sell less. Revenue grew 19% at Microsoft, 25% at Alphabet and 26% at Meta. What changed is the bill: $125.6 billion in three months for data centers, chips and servers, double what they spent a year earlier. Meta put 54 cents of every dollar of revenue into capex. Alphabet spent $9 billion more than it took in, its second negative quarter in a row, and this morning its stock fell 12%. If it closes down more than 9.77%, it will be its worst day after earnings since going public in 2004.
The detail that worries me most is in Google's search numbers. Search ads grew 12% and still missed expectations. Philipp Schindler, Alphabet's chief business officer, said part of the reason is that AI Mode makes less money per search. So Google is spending record sums to build a product that earns less than the one it replaces.
For twenty years these were the best businesses ever invented. You wrote the code once, sold it to billions of people, and almost every new dollar of revenue fell to the bottom line. AI doesn't work like that. Every answer runs on a chip someone had to buy and on electricity someone has to pay for. Big Tech is turning into something closer to a utility: huge upfront costs, long paybacks, and profits that depend on keeping the machines full.
That can still be a great business. But it is a different business, and investors are only starting to price it that way.
The old moat was code. The new one is cash. -
David Cournapeau reposted
Colin Percival
@cperciva
Very much this. If a mechanic points out that the brakes on your car are about to fail, the right answer is "wow thanks for warning me" not "my car was perfectly safe until you came along".
Patrick McKenzie
@patio11
I personally think LLMs are zero point zero percent culpable for software bugs written in 2018 but, well, we all knew there was eventually going to be a bill due, and here we go. Seems to not have as generous payment terms as we had expected.
-
David Cournapeau reposted
Bojan Tunguz
@tunguz
There is a happy medium where you could have been just technical enough to be able to maximally benefit from the current AI explosion, without having wasted too much time and energy on becoming over invested in skills with limited shelf life.
Thariq
@trq212
very weird thing happening rn
technical people are sad, they think their skills are now useless
non-technical people are desperately wishing they were more technical so they could use agents better -
David Cournapeau reposted
Jason Furman
@jasonfurman
1. AI should play a large role before we see anything from anyone so it comes out right in the first place.
2. AI should play a big role in retrospective replications.
3. Piketty-Saez-Zucman work has many flaws that all go one direction.
Spreading slop hurts all three causes. -
David Cournapeau reposted
Isaac Saul
@Ike_Saul
Open, proud anti-semitism is more prevalent and pernicious now than at any time in my life. Truly terrifying how many of these people are just... out there. Sitting at comedy shows.
Ichigo Niggasake
@SomaKazima2
Hey so this is crazy
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Daniel Lemire
@lemire
Software around me is not improving nor is it expanding.
If AI drastically lower the cost of software… then we have a mystery, don’t we?
The answer is that building the software was never the difficult part in these cases. Our organizations have been optimized for process, meetings, and risk avoidance instead of shipping working systems. It is institutional decay: layers of people whose main job is to prevent other people from doing the work. -
David Cournapeau reposted
Victor Eyraud
@EyraudVictor
« Mais qu’est-ce qu’on a fait avec ces 3000 milliards de dette ? »
Ben on a payé les retraites.
Alors, elle est pas belle la « solidarité intergénérationnelle » ? -
David Cournapeau reposted
Matthew Yglesias
@mattyglesias
America has made large strides against crime in my lifetime but it's still the case that the highest-crime French city has a substantially lower murder rate than Montana to say nothing of actually high-crime US cities.
It's a big problem.
www.slowboring.com/p/how-i-would-fix-all-of-americas-major -
David Cournapeau reposted
EuroMaximalist 🇪🇺
@euromaximal
Pensions will be the death of our civilisation. They are unsustainable and sucking funds out of every other essential public service.
It’s the root cause behind the protests and many of our economic and societal malaises. And it’s also the one no one wants to fix. The moment a politician decides to blow up the pension pyramid scheme it’s the moment their career ends.
We need a youth revolution that circumvents boomer’s electoral domination and imposes pension reform. It’s not like they can revolt back from their retirement homes.
Mads
@europemaxxed
Spain is having record breaking tax collection every year and yet there's not a single public service or infrastructure that improved in the past 20 years because all the growth goes to funding boomers pensions. That's why Gen Z is protesting in many countries
Sierra Golfo
@SierraGolfo_
Impuestos de Noruega, sueldos de Zimbabue y trenes de la India
-
David Cournapeau reposted
Sylvain Catherine
@sc_cath
Everyone knows my priors, and they are not particularly kind. Nonetheless, this should not be taken too seriously. AI is not good at making this kind of assessment, and we don’t know what the prompt was.
Some of the “errors” are actually criticisms raised in papers published after the ones analyzed, which is the normal process of science.
This is not the right way to criticize each other.
Marginal Revolution
@MargRev
Errors in the research of Emmanuel Saez and Gabriel Zucman marginalrevolution.com/marginalrevolution/2026/10/errors-…
-
David Cournapeau
@cournape
Am no mao apologist but that’s a good example of the hypocrisy on the west about mass murders committed/caused by western allied
Sebastian Milbank
@SebMilbank
Taiwan and South Korea achieved higher economic growth and life expectancy gains without anyone committing mass murder.
Aaron Bastani
@AaronBastani
Chinese life expectancy in 1930 was around *30*.
By 1978, when Deng became the country’s paramount leader, that had increased to 63. The average Chinese lived over a decade more than the average Indian.
Either we can cope, or grasp the achievements driving China’s rise.
Aaron Bastani
@AaronBastani
I wrote about Burnham’s first 100 days, Mao’s mass line, & ‘consultative Leninism’.
We are now seeing a PM who leans into left populism, without having much time for members driving policy. That’s new.
Centralisation plus input from below, seems an interesting mix…..
unherd.com/2026/10/chairman-andys-first-100-days/ -
David Cournapeau reposted
Alister
@SchnockRevue
Cette manière d'éconduire ce pain au chocolat boursouflé c'est le rôle d'une vie...
Titre de travail : "La Goûteuse".
#lesvraiesquestions
best of léa seydoux
@badpostseydoux
Léa Seydoux tries some of the finest French and British delicacies for ELLE UK.
-
David Cournapeau reposted
Matt Turck
@mattturck
This is why French founders are a special breed - to even start a company, they have to first face this kind of feedback from their families and friends lol 😂
best of léa seydoux
@badpostseydoux
Léa Seydoux tries some of the finest French and British delicacies for ELLE UK.
-
David Cournapeau 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-…
-
David Cournapeau reposted
Ze RJ
@RJSupH
As French the most French thing she said was « I don’t see the purpose » !
-
David Cournapeau reposted
Philippe Lemoine
@phl43
As much as I think the Chinese reaction to Mbappé’s statement was ridiculous, people need to chill out.
Western freedom of speech, such as it is (the Chinese are not entirely wrong to call out our hypocrisy on the issue), isn’t going to disappear because from time to time a Western celebrity or company decides to go out of their way not to offend the delicate sensibilities of China on a handful of topics.
Once again, insofar as freedom of speech is threatened in the West, the main threats are domestic and it’s not even close. -
David Cournapeau reposted
Paul Cébille
@Ellibec
Avec seulement 10% des retraites taxées, la France fait figure des grandes économies européennes les plus "légères" lorsqu'il s'agit de faire contribuer ses retraités.
En Suisse, 23% des prestations retraites sont soumises à prélèvements obligatoires.
21% en Italie, 15% en Allemagne.
Hexagone - La France en chiffres
@Hexagone_org
🔴 ALERTE INFO : la DREES vient de révéler que la France est un des pays qui taxe le moins les retraites.
-
David Cournapeau reposted
The Capital in the North
@GabrielCor54922
I couldn't not share this one.
-
David Cournapeau reposted
Joel Grus 🤠
@joelgrus
my team at anthropic no longer uses coding interviews, instead we just look at successful startup exits, C-level public company experience, podium finishes at world's strongest man competitions, and top-tier publications
-
David Cournapeau reposted
Boring_Business
@BoringBiz_
It seems like the bond market will not finance all that data center spend after all, Mr Bond
-
David Cournapeau reposted
Maaz
@mmaaz_98
I saw the recent softmax bug which was found by Gemini, and tried to reproduce it with a property-based test in Hypothesis. Surprisingly, even the proposed fix fails this test. So, I wrote a more robust softmax which does indeed pass this test!
arxiv.org/abs/2610.36427 -
David Cournapeau reposted
thaddeus e. grugq
@thegrugq
I’m concerned we haven’t adequately prepared for the inflation of exploit naming. We should probably be using
php00002shell
Just to be future-proof. That will allow an easy 99997 more exploits before we need to update again. -
David Cournapeau
@cournape
no, I really don't have space in my flat anymore
Resident Advisor
@residentadvisor
Roland (@RolandGlobal) has released the JUPITER-08, a new addition to its compact Boutique series marking the 45th anniversary of the original Jupiter-8. ra.co/news/86307
-
David Cournapeau reposted
Resident Advisor
@residentadvisor
Roland (@RolandGlobal) has released the JUPITER-08, a new addition to its compact Boutique series marking the 45th anniversary of the original Jupiter-8. ra.co/news/86307
-
David Cournapeau reposted
Kyle Hall
@kylehalldetroit
-
David Cournapeau reposted
DREAMBOY💙
@s_sibomvu
When I’m taking a shower and a song I don’t like come on
-
David Cournapeau reposted
Mario Zechner
@badlogicgames
nice that youngsters like @hotlinemagi get me.
Mario Zechner
@badlogicgames
software is dead. sad.
-
David Cournapeau reposted
pirate.moo
@apiratemoo
That is one hell of a typo.
John Althouse
@4A4133
Announcing 3 Critical vulns for hundreds of millions of IoT, Firewalls, OT, and Military devices including: Palo Alto Networks, Comcast, D-Link, HP, Canon, Honeywell, GE, Oracle, Siemens, Rockwell, Digi, Lockheed Martin, Boeing...
The vulns are with GoAhead Web Server which is a backend library used by these devices. There's a patch available from GoAhead but that doesn't mean there's a patch available for the devices yet -- that's up to the manufacturer.
- TLS traffic over DHE ciphers have a 1-byte secret, "04", meaning anyone can decrypt connections to these devices and steal login credentials and session tokens if the client negotiated with DHE ciphers like the latest TLS_DHE_RSA_WITH_CHACHA20_POLY1305_SHA256, which is extremely likely.
- LibreSSL mTLS has been completely bypassed, meaning anyone can connect over mTLS. If you have one of these devices listening on the internet and are expecting mTLS to protect it from attackers, take it offline or patch asap.
- OpenSSL Certificate Revocation Lists (CRLs) are not enforced, meaning if a user device is stolen or compromised, and you revoke that mTLS cert, nothing actually happens. There are a lot of other attacks this can be used for.
These are critical because these types of devices run our world. Water systems, electricity, military, firewalls, critical infrastructure.
foxio.io/blog/choose-your-backend-three-tls-bugs-in-goahead -
David Cournapeau reposted
小爱同学
@Nnnxq3j
既然姆巴佩可以,拿我们的领袖来开玩笑,我们是不是也可以用拿破仑来开玩笑?🤔🤔🤔
-
David Cournapeau reposted
How To Prompt
@HowToPrompt__
This paper answers:
why anything is beautiful
why jokes are funny
why we get curious
why you get bored
why babies explore
why music has hooks
For centuries, we thought these were deeply mystical human traits. Soul-level mysteries that machines could never touch.
But, the paper proves that the human brain operates as a relentless compression engine. It is constantly trying to predict the world and compress it into the smallest possible file size.
Here is how it explains your entire life:
If something is completely predictable, your brain has already compressed it. There is no new data to extract. The result? You feel bored.
If something is pure randomness, your brain cannot compress it at all. The result? You feel annoyed or tune it out.
But then there is the sweet spot.
When you encounter something that seems complex, but suddenly your brain finds the hidden pattern and connects the dots... your brain releases dopamine.
That sudden, glorious rush of successful data compression? That is what you experience as "curiosity."
When a comedian tells a story that makes no sense, and the punchline suddenly makes all the pieces fit together instantly? That rapid neural compression is what you experience as "humor."
When you look at a painting and notice an underlying symmetry that ties the chaos together? That is "beauty."
Art, science, music, and jokes are not magic.
They are just highly optimized data sets designed to trigger compression progress in your neural network.
Here is the most unsettling part.
This paper wasn't written by a philosopher. It was written by an AI pioneer to figure out how to program intrinsic motivation into artificial agents.
He wasn't trying to explain humans. He was writing the blueprint for machines.
We always assumed AI could never appreciate art, laugh at a joke, or feel the spark of curiosity.
But if our deepest subjective experiences are just a mathematical reward for maximizing compression progress... -
David Cournapeau reposted
chiefofautism
@chiefofautism
holy fuck
chiefofautism
@chiefofautism
if you have a qwen3.8 27b around, try steering L34 N3838 +0.75, now you have evil qwen 27b
for extra evil add L38 N17180 +0.75
same recipe as @_lyraaaa_: dose × median(‖resid‖ leaving layer L, tokens 1:) / ‖W_down[:, n]‖ at the down_proj input, every token -
David Cournapeau reposted
Jaana Dogan ヤナ ドガン
@rakyll
Mid-existential crisis and the timeline decides P=NP. Perfect timing. No complaints.
-
David Cournapeau reposted
Asad🗽🍎
@AsadFromNYC
What a line:
"Most Israeli Jews now see the world the way Netanyahu has long seen it. That is Netanyahu’s great victory. But it is matched by his great loss: Much of the world now sees Israel the way Gazans see it." www.nytimes.com/2026/10/25/opinion/israel-gaza-election-n… -
David Cournapeau reposted
Brian Lui
@brianluidog
Brazil is horseshoe theory. Once your choice is between left- and right-authoritarianism, you've already lost.
In Australia we have milquetoast centre-left/centre-right, and crying wojaks who want communism or white nationalist Islamofascism. Long may it remain that way. -
David Cournapeau reposted
Andrew Gelman et al.
@StatModeling
We interrupt our usual programming to inform you that I was never hired by Facebook to do a hit job on Jonathan Haidt.
statmodeling.stat.columbia.edu/2026/10/27/we-interrupt-ou… -
David Cournapeau reposted
Ramez Naam
@ramez
I'm sure it will go up! And Joshua talks about the lags that will eventually (probably) be surpassed.
I think most of us just don't have a handle on the idea that cybercrime is already ~$1T of annual damage, before AI. -
David Cournapeau reposted
Rachel Thomas
@math_rachel
"People think 'the ideas are mine, the writing is the AI', i.e., they feed the AI a list of rambling bullet points that it massages into a blog post, or a paper,.... But they are wrong, because writing is thinking." -- @zetalyrae 1/
(A thread on how writing is thinking 🧵) -
David Cournapeau reposted
Philippe Lemoine
@phl43
Damn I thought I had strong French energy but she's setting the bar very high.
best of léa seydoux
@badpostseydoux
Léa Seydoux tries some of the finest French and British delicacies for ELLE UK.
-
David Cournapeau reposted
Alex Imas
@alexolegimas
Peter Thiel might be onto something here…
aleximas.substack.com/p/what-will-be-scarce?r=1ds20
FWIW I think he’s missing important dimensions like desire for human connection, empathy, shared understanding, which will be important for the future of jobs in healthcare, education, etc.
John LeFevre
@JohnLeFevre
Walter Isaacson asks Peter Thiel what humans can uniquely do in the AI era.
Thiel: "We're very relational… these relational dimensions will become more important."
When knowledge is worth $0, the only career skills that matter:
- networking / social calibration
- persuasion (sales)
- curiosity (harnessing AI)
- taste (curation)
- storytelling
- grit
- golfing with clients (or equivalent)
We'll get so sick of a fake world that we'll go back to doing business like it's the 1800s. A handshake at the local saloon. -
David Cournapeau reposted
Gary Bernhardt
@garybernhardt
I've existed on the Internet long enough to see us go from "You can't refuse to host content just because it's illegal!" to "You may not access that publicly-available information, citizen."
Dennison
@DennisonBertram
Claude now determines what public information you're allowed to have.
This is a perfectly legitimate task, and I haven't suggested nor asked for anything nefarious. It's just information gathering.
Claude has decided I'm not allowed to have that (public) information. -
David Cournapeau reposted
Jaana Dogan ヤナ ドガン
@rakyll
I’m not doing a ton of coding interviews nowadays but I’d be interested to interview how people use coding agents. That itself gives me more signals about candidate’s coding skills.
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-… -
David Cournapeau reposted
Philippe Lemoine
@phl43
He wasn’t even mocking him, he just called him "a Chinese dictator" (it’s clear from the video that he had no idea who exactly Mao was), saying the meme in which they had replaced Mao’s face with his had made him laugh. This hyper-sensitivity is kind of pathetic to be honest.
POLITICOEurope
@POLITICOEurope
Major brands in China have dropped Kylian Mbappé over a clip in which the French football megastar appeared to mock Mao Zedong, China’s modern founding leader.
vist.ly/5q5xz -
David Cournapeau reposted
roon
@tszzl
honestly have no idea what frontier lab equity is worth; it seems highly plausible that huge top down slowdowns are coming, which would make frontier models uncompetitive. total nationalization also seems possible. anything could happen with this technology
-
David Cournapeau reposted
Federico Vaggi
@F_Vaggi
Eli Dourado
@elidourado
How does this nonsense make it into the paper of record?
-
David Cournapeau reposted
colton
@coltonpadden
Join me next week at Data Outpost where I'll be covering how to build a data agent with @eve!
-
David Cournapeau reposted
2001 Live
@25YearsAgoLive
Holocaust survivor and writer Simon Wiesenthal announces that he is retiring after a 60-year career hunting Nazis. He is now 92 years old and will be 93 next month.
The last Nazi he captured was Julius Veil, who was convicted this year of shooting seven Jewish prisoners.
“I have survived them all,” says Wiesenthal in a press statement. “If there are any left, they’d be too old and weak to stand trial today. My work is done.” -
David Cournapeau reposted
Lucas Beyer (bl16)
@giffmana
can't believe it's actually working
Andrew Curran
@AndrewCurran_
OpenAI has announced to three new discoveries in mathematics, computer science, and physics:
- Local smoothing and the Kakeya conjecture
- The Exponential Time Hypothesis is false
- Existence of quantum chromodynamics and completions of the Standard Model -
David Cournapeau reposted
Défense & Sécurité Internationale
@DSI_Magazine
La chronique "Penser les opérations" de @Michel_Goya: Ormuz : la guerre du détroit aura-t-elle lieu ? Article paru dans DSI n°185, septembre-octobre 2026.
www.areion24.news/2026/10/09/penser-les-operations-ormuz-… -
David Cournapeau reposted
Sina Toossi
@SinaToossi
On this Israeli election day, I’d recommend reading this extraordinary @haaretzcom interview with billionaire Ronald Lauder, president of the World Jewish Congress and once one of Benjamin Netanyahu’s closest confidants.
One of Lauder’s most striking conclusions: “I realized at the end of the day that Israel really didn't want peace, but they wanted dominance.”
This is coming from someone who helped launch Netanyahu’s political career, introduced him to powerful figures in American politics and media, and secretly negotiated with Hafez al-Assad and Rafik Hariri on his behalf in the late 1990s.
In one of the interview’s biggest revelations, Lauder says Netanyahu agreed to a comprehensive peace deal with Syria and Lebanon, including withdrawal from the Golan Heights, but ultimately refused to sign it for political reasons.
“He chose politics over peace. He put his political future before the deal.”
Now 82, Lauder describes his profound disillusionment with Israel’s direction. After decades of supporting Israeli and Jewish institutions, he has sold his Tel Aviv apartment, given away his Be’er Sheva property, and says he has no intention of returning to Israel.
He warns that Netanyahu’s dependence on figures like Ben-Gvir and Smotrich is damaging Israel, alienating diaspora Jews and eroding support in Washington.
The interview is also full of fascinating details about Lauder’s role in Netanyahu’s political rise. The two became close friends in the 1980s, when Netanyahu was Israel’s ambassador to the UN, speaking almost daily at times.
Lauder says he introduced Netanyahu to speech coach Lilyan Wilder to improve his public speaking and, crucially, to Republican strategist Arthur Finkelstein, who would become instrumental in Netanyahu’s electoral success.
When Netanyahu left New York to pursue a political career in Israel, Lauder threw him a farewell party attended by some of the most influential figures in American media and politics, including Rupert Murdoch and the head of the New York Times.
Lauder recalls opening doors for Netanyahu and helping him cultivate the relationships that would underpin his rise to power.
What makes this interview so revealing is that Lauder is no longtime critic of Israel. He had been a committed Zionist, a Republican power broker, and a man who spent decades financing Jewish institutions, cultivating relationships with Arab leaders, and pursuing diplomatic openings on Israel’s behalf.
A fascinating read 👇
www.haaretz.com/israel-news/elections/2026-10-22/ty-artic… -
David Cournapeau reposted
Nikkei Asia
@NikkeiAsia
Japan's Sakana AI to build homegrown model to cut US, China reliance
Country's top unicorn eyes use in sensitive fields like defense and cybersecurity
asia.nikkei.com/business/technology/artificial-intelligen… -
David Cournapeau reposted
Sakana AI
@SakanaAILabs
We’ve published an interview with Sakana AI’s VP of GTM 🐡
sakana.ai/vp-of-gtm-interview/
In 2026, Sakana AI released Sakana Chat, Marlin, Fugu, Translate, and Namazu, entering the stage of delivering products born from its research to enterprises. Our GTM team launched in Sept 2026. -
David Cournapeau reposted
JFPuget 🇫🇷🇺🇦🇨🇦🇬🇱
@JFPuget
Good luck with pay-for-results pricing
We dreamed of it when we had the market leader in mathematical optimization (CPLEX). In some cases (in finance industry), the ROI for using CPLEX was so high that it paid the price paid every few minutes. Running 24/7. The ROI was in billions USD a year. Burt we sold at flat price.
We only managed to get paid on result once. That was the best deal of the year in Europe for IBM. I am not sure I can disclose who the customer was nor what the app was alas.
Back to revenue based pricing, reason it won't fly is that it is very hard to measure the ROI when you're the tool provider.
Guillermo Flor
@guilleflorvs
Dario Amodei: some tokens Claude produces are worth a few cents, others could be worth tens of millions of dollars.
His examples:
1. Telling someone to restart their Mac: the model has said it 10 million times, worth a dollar or less
2. Telling a pharma company to move an aromatic ring to the other end of a molecule: potentially tens of millions
3. Expect pay-for-results pricing, or labor-like pricing by the hour
4. It's a new industry, so many models will be tried
Today pricing is simpler: per the S-1, usage-based pricing was 80% of H1 2026 revenue and subscriptions 20%.
Full breakdown here:
www.theaiopportunities.com/p/anthropics-s-1-the-2023-pitc… -
David Cournapeau reposted
Tim Sweeney
@TimSweeneyEpic
Hello from Tokyo, home to next week’s UnrealFest!
-
David Cournapeau reposted
Arnaud Bertrand
@RnaudBertrand
The ugly truth about France and some of the other European countries currently in turmoil - and which none of the politicians want to admit because it'd be political suicide - is that you can't solve the problem by going after immigrants (as the right asserts) or the rich (as the left asserts), but you could solve it by going after the old.
One extraordinary statistic for you: people over 60 in France, a quarter of the population, own 60% of France's wealth, non-financial and financial (www.ofce.fr/pdf/revue/2025/3-192OFCE.pdf).
And here is another crazy statistic:
that same quarter of the population, those above 60, receive nearly two-thirds of all social spending by the French government (www.previssima.fr/actualite/protection-sociale-qui-paie-l…) - which is famously among the highest in the world (France's government spending is an insane 57.3% of GDP).
Last statistic: they make up about 40% of voters because they're a third of registered voters (www.insee.fr/fr/statistiques/2664851), and they turn out far more than anyone else (www.franceinfo.fr/elections/sondages/legislatives-2024-le…).
Put our 3 statistics together and there's almost a bit of an apartheid situation here: the old own most of the country's wealth, collect most of the social spending and have the political power to keep it that way.
Which makes them a de-facto rentier class, living off assets and transfers paid for by younger generations.
And, of course, at the risk of stating a truism, the future is built by the young. Yet France keeps effectively taking money away from them to give it to those whose working lives are already behind them - and need the money far less since they already own the country's wealth.
What about inheritance, I hear you ask? That's the funniest/saddest aspect: given that life expectancy has kept rising, the French now inherit a home at 54 on average. The most common case is someone in their 80s leaving it to heirs in their 50s, in effect the very old transferring wealth to the about-to-be old (www.insee.fr/fr/statistiques/fichier/version-html/9061645…).
You can add this to our earlier apartheid statement: on top of all the rest, even when the old finally let go of their wealth, it stays within their own "camp".
That's my personal situation, incidentally: I'm 42, my father is 71 and he is starting to look at ways to pass things on to me and my siblings, when our professional lives are already well settled, so it'll arrive when it changes things the least, which is pretty much the French system in a nutshell.
Now, I get that there'd be some unfairness in going after them: they paid into the system their entire working lives, playing by the rules of a social contract that told them if they paid up they'd collect at the end. Changing the rules now means breaking that promise, which is admittedly unfair.
But there's no magic wealth tree in any society: you can only take it where it is. That's the way most major societal transformations end up happening. They're often triggered by a catastrophic event (the so-called "four horsemen": mass-mobilisation war, revolution, state collapse, plague) BUT what really drives change is a transfer: wealth gets pried from the privileged class who've been sitting on it and handed to those who'll build what comes next.
So, in theory and in an ideal world, you can do without the catastrophic event and directly proceed with the transfer...
The fact that in France, and elsewhere in Europe, it's the youth protesting is probably no coincidence - even though they don't frame it as a generational fight. And even though, cruel irony, the left politicians many of them march behind promise to bring the retirement age back to 60 (lcp.fr/actualites/legislatives-2024-le-rassemblement-nati…), which would make the situation even worse.
Let's say they understand intuitively that the system is stacked against them, even if they haven't yet figured out how. And politicians are obviously doing what politicians do, meaning pandering to the people who actually vote, while pointing the young towards easier - and politically convenient - scapegoats.
To be clear I'm not saying it's the only problem in France, VERY far from it. I'm also not saying that all old people in France live a lavish lifestyle at the expense of younger generations: there are also plenty of very poor retirees who, if anything, would deserve more.
What I am saying is that it's probably the biggest elephant in the room: everyone can see it, nobody dares mention it - no politician, from any side, wants to run on cutting grandma's pension - and it makes every other problem harder to fix. -
David Cournapeau reposted
Antonio García Martínez
@antoniogm
You can tell which is your most important meeting of the week, because that's the meeting when your Mac will randomly choose to update its OS.
Follow me for more management tips. -
David Cournapeau reposted
João Eira
@joaoeira
Think you'll appreciate this one borretti.me/article/no-man-is-an-island
-
David Cournapeau reposted
Mario Zechner
@badlogicgames
recommended reading.
João Eira
@joaoeira
Think you'll appreciate this one borretti.me/article/no-man-is-an-island
-
David Cournapeau reposted
Martin Gratzer
@mgratzer
I wrote down how my work with agents changed this year. I still break down the work into vertical slices that ship on their own. I rarely read diffs now: independent reviews and a risk check catch most of it, so my time goes into shaping and guardrails.
mgratzer.com/posts/reforging-the-workflow/ -
David Cournapeau reposted
Gergely Orosz
@GergelyOrosz
On one end, this is tough to see: MSFT has been one of the most stable Big Tech employers. Clearly, layoffs, unfortunately, glooming.
On the other: letting people DECIDE to take a package now, vs wait for the axe to drop is a far more considerate approach.
Amanda Goodall
@Amanda_Goodall
$MSFT offers 60,000 employees a voluntary exit.
60K Y’all!!
First time this has ever been offered at Microsoft and it will affect employees in India, The United States, Canada, and the UK at level 70 or below…
Employees find out in a few hours if they are being targeted for this.
This just makes me think that H127 layoffs sound intense. Is this a pre-layoff round?
Also makes me wonder if they changed some scheduled layoffs due to be called out by @VP. Peace offering? 🤔 Timing is just interesting.
More to come. -
David Cournapeau reposted
eric ゑリッ久
@shinobu_books
Tokyo has a neat little calculator to determine how much you need to stockpile.
You enter in general parameters about your family members, pets etc. & it produces a list of recommendations.
Site is in both English and Japanese.
www.bichiku.metro.tokyo.lg.jp/en/tool/
Gearoid Reidy リーディー・ガロウド
@GearoidReidy
Oh for the love of god
(average English-language reporting on Japan) -
David Cournapeau reposted
Nate Foster
@natefoster
ETH refuted! No word yet on EPFL.
-
David Cournapeau reposted
Mihail Eric
@mihail_eric
I’m excited to finally announce the newest edition my Stanford course 𝗧𝗵𝗲 𝗠𝗼𝗱𝗲𝗿𝗻 𝗦𝗼𝗳𝘁𝘄𝗮𝗿𝗲 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗲𝗿. It has been 9 months in the making.
Last November, with the release of Claude Opus 4.5, coding agents experienced a step function improvement in capability. We all felt it. The LLMs were more powerful, could reason for longer, solve harder tasks.
This year’s iteration of my course reflects the 2026 metamorphosis of software engineering.
My core belief is simple: AI-native developers of the LLM era are going to become the most important members of any software organization. I have designed my course to train this next generation of engineers.
𝗪𝗵𝗮𝘁’𝘀 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁 𝘁𝗵𝗶𝘀 𝘁𝗶𝗺𝗲 𝗮𝗿𝗼𝘂𝗻𝗱
First, 85% of my Fall 2025 class material is being thrown out. The Fall 2026 syllabus reflects the core capabilities AI-native engineers must have: agent skills, advanced context engineering, MCP portals, agent-ready codebase principles, agentic code review, security, parallelizing background agents, software factories, and more.
Second, I am going to teach my students how to have software taste. Every student will be required to ship pull requests to production-grade, real-world codebases. The course is collaborating with the top open-source AI repos who will offer support and mentorship to students on how to meaningfully contribute to their projects.
This has never been done before in any university course so I am incredibly grateful to our OSS Partners: @browserbase, @HeyGen, @CopilotKit, @semgrep, @OpenHandsDev, @milvusio, @marimo_io, Pi, @crewAIInc, @warpdotdev, @vercel, @cmux, @arizeai, @UnslothAI, and @anyscalecompute.
𝗪𝗵𝗮𝘁’𝘀 𝘀𝘁𝗮𝘆𝗶𝗻𝗴 𝘁𝗵𝗲 𝘀𝗮𝗺𝗲
I’m fortunate to again have AI software engineering leaders and founders as guest speakers to share their learnings from building top coding agent products. Thank you to @leerob from @cursor_ai, @bcherny of @claudeai code, @EnoReyes of @FactoryAI, @silasalberti of @cognition, @0xine of @semgrep, Rajesh Bhatia of @Cloudflare , @amasad of @Replit, and @eladgil.
All resources will be available online. All classes will be available to the public.
9/22 on Stanford campus. See you in class.
themodernsoftware.dev -
David Cournapeau reposted
Nick
@teleologyman
𝐀𝐬𝐚𝐤𝐲𝐆𝐑𝐍
@AsakyGRN
Moment an American pastor and prophetess, who are both married, introduce a new wife to their congregation and announce that they are now in a polygynous marriage with a female church member, claiming it is the “will of God.” 😭🤦🏾♂️🙆🏾♂️
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Phil Hie
@philhie
declining zizek because you read the books is the most taleb thing possible
-
David Cournapeau reposted
Stefania Druga
@Stefania_druga
Yesterday the @SakanaAILabs RSI Lab took the stage at Hitotsubashi Hall, with a keynote by @SchmidhuberAI, @hardmaru on why we have to think differently, @iwiwi on pretraining our own foundation model, Sawara, here in Japan, and @RobertTLange on The AI Scientist 🐟 and @risi1979 I spoke about world models for physical AI 🤖
-
David Cournapeau reposted
Timothy Gowers @wtgowers
@wtgowers
For those who don’t know, ETH is the hypothesis that 3SAT needs exponential time. Very surprising and consequential that it is false.
-
David Cournapeau reposted
Mark Farina
@djmarkfarina
-
David Cournapeau reposted
Faucon Alpin 🏔️
@VraiconAlpin
On devrait remplacer parcoursup par un grand examen national où tous les lycéens passeraient la même épreuve et seraient notés selon le même barême
-
David Cournapeau reposted
Michael Fowlie
@mwfowlie
Why are people upset at comparing ratios of probabilities on a log scale?
So many posts about this.
If you have a problem with the log scale that is a skill issue
Crémieux
@cremieuxrecueil
I'm shocked that The Economist was willing to publish this!
-
David Cournapeau reposted
Sylvain Catherine
@sc_cath
Je ne sais pas. Je me souviens que lorsque j’étais doctorant, il y avait une grande panique autour des cours en ligne. On pouvait apprendre ce qu’on voulait du meilleur expert au monde sur le sujet.
Même s’il est difficile d’expliquer exactement pourquoi, on sait aujourd’hui que le rôle d’un professeur et d’une salle de classe n’a jamais été de livrer une connaissance.
Je ne sais pas si les changements à venir induiront une baisse ou une hausse de la demande pour l’éducation présentielle.
Raphael Kahan
@RaphaelKahan
Et embaucher des nouveaux profs quand on a une superintelligence gratuite 24h/24 dans sa poche c'est pas super malin non plus
-
David Cournapeau reposted
Philippe Lemoine
@phl43
This is mostly due to social spending, which wasn't so high in the Soviet Union. People think of the Soviet Union as some kind of welfare state on steroids, but social spending was only about half of the current level in France as a share of GDP when defined narrowly and, even if you include food subsidies and enterprise benefits, it was still significantly lower. It was probably high relative to a typical, non-socialist country at a similar level of development, but not shockingly so.
What made set the Soviet Union and other socialist countries apart from market economies was not so much social spending but the reliance on central planning rather than prices and the market to coordinate production units. It's actually a testament to the efficiency of a market economy that France can get away with so much redistribution, even allowing for the fact that it's not sustainable at current levels, without destroying its ability to produce wealth in the first place. -
David Cournapeau reposted
Nathaniel Bechhofer
@bechhof
After all this time, use of deterministic tooling is still under-done in AI implementations
Gwen (Chen) Shapira
@gwenshap
OpenAI's Dot seems surprisingly bad at "Here are the times person X is available. Check my calendar and draft an email sharing the slots that both person X and myself are available for a call".
First, it was overly narrow. Offering just 3 30-min slots.
Then it was overly broad, including times one of us were not available. Then it apologized for overstating our availability and not checking, but also insisted that the broad version is correct. -
David Cournapeau reposted
Philippe Lemoine
@phl43
Just to elaborate, while companies like Samsung is famous all over the world, the South Korean semiconductor industry only employs 0.6% of the labor force. Even if we look at the electronic industry as a whole, it still employs only 2% of the labor force.
The retail sector, by contrast, employs almost 7% of the labor force. It's one of the sectors that employs the most people in every country. For instance, in the US, it employs close to 10% of the labor force. But people almost never talk about it and about how productive it is.
I think there is a good case to be made that people focus too much on the high-tech sector and other shiny objects and not enough about sectors that may not be as sexy but employ far more people. It's a big problem in the discourse about China in my opinion. Lewis's book really drove that point home for me.
Philippe Lemoine
@phl43
It's funny that a lot of people in the replies object that such a comparison makes no sense, on the ground that unlike South Korea Poland has no global brands, but perhaps there is a lesson here about how the discourse on development shouldn't focus so much on shiny objects!
Marcin Piatkowski
@mmpiatkowski
Poland 🇵🇱 and South Korea 🇰🇷: two global growth champions with two different growth models and social outcomes.
Over three decades both grew fast (4.2% vs 3.7% a year per capita) and reached the same productivity (~$52/hour), but with very different R&D spending: 1.4% vs 5.1% of GDP.
On social outcomes, Poland ranks #24 in world happiness vs Korea's #67, with half the suicide rate and a third of the old-age poverty (13% vs 40%) 👇
The shared challenge: among the lowest fertility rates in the world, 1.07 and 0.80.
@Noahpinion @KoreaEconInst -
David Cournapeau reposted
Linear Labs
@LinearLabs
Scene from Portishead's "To Kill a Dead Man" (1994),an 11-minute, black and white short film co-written, scored, and starred in by the band.
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Claire Martel
@ClaireMartel47
🔴 Il y a quatre jours, Bruxelles s'inquiétait du poids d'EDF sur le marché si elle construisait seule les six EPR2. On sait maintenant ce que la Commission réclame en échange de son feu vert.
Selon La Lettre, Bruxelles conditionne l'approbation du financement public des six réacteurs à des concessions sur l'organisation du marché électrique français. Deux demandes sont au cœur des négociations : davantage d'interconnexions entre la France et ses voisins, et des garanties permettant à des acteurs tiers d'accéder à une partie de la production des futurs réacteurs. EDF et l'État défendent une position commune et veulent conclure avant la fin de l'année.
Le second point est décisif. La question n'est plus seulement de savoir si la France peut construire de nouveaux réacteurs, mais qui vendra l'électricité qu'ils produiront, et à quelles conditions.
La France connaît déjà ce scénario. De 2011 à 2025, l'ARENH a obligé EDF à céder jusqu'à 100 TWh de sa production nucléaire par an à ses concurrents, à 42 €/MWh, un prix resté figé pendant plus de dix ans.
Comparons maintenant les calendriers.
Juillet 2026 : l'Allemagne vote une loi pour financer 12 GW de nouvelles centrales, surtout au gaz. 2 septembre : Bruxelles approuve jusqu'à 35 milliards d'euros d'aides.
Novembre 2025 : la France notifie le financement de six EPR2. Mars 2026 : la Commission ouvre une enquête approfondie. Octobre 2026 : toujours aucune décision, mais une liste de conditions.
Moins de deux mois pour des centrales au gaz. Onze mois, et ce n'est pas fini, pour des réacteurs nucléaires.
En mars, Ursula von der Leyen reconnaissait pourtant que l'Europe avait commis une « erreur stratégique » en tournant le dos au nucléaire.
EDF doit prendre sa décision finale d'investissement fin 2026. L'enquête européenne se trouve sur le chemin critique du projet.
Claire Martel
@ClaireMartel47
🔴 SIX RÉACTEURS, EDF AUX COMMANDES, ET BRUXELLES S'INQUIÈTE DE SA PUISSANCE SUR LE MARCHÉ.
La France a confié à EDF la construction et l'exploitation de six EPR2. Coût estimé : 72,8 milliards d'euros, aux prix de 2020, avec un prêt bonifié de l'État couvrant 60 % des coûts de construction et un mécanisme de soutien des prix sur 40 ans.
Depuis le 31 mars, la Commission européenne mène une enquête approfondie. Elle reconnaît elle-même que le projet est nécessaire.
Ce qui l'inquiète, entre autres, c'est le risque de renforcer la position d'EDF sur le marché de l'électricité.
Mais voilà la particularité de ce dossier : EDF est déjà l'opérateur du parc nucléaire français. Elle possède les équipes d'ingénierie, l'expérience d'exploitation et les filiales industrielles nécessaires pour porter ce programme.
Fallait-il lui demander de renoncer à ses compétences pour rassurer Bruxelles ?
Pendant ce temps, le calendrier tourne : décision finale d'investissement attendue fin 2026, premier béton à Penly envisagé fin 2028, mise en service prévue en 2038.
Contrôler l'argent public est légitime.
Vérifier que les aides ne faussent pas excessivement la concurrence aussi. Mais un programme nucléaire se compte en décennies. Et le financement doit être sécurisé suffisamment tôt pour tenir les échéances.
Si EDF occupe cette place dans le nucléaire français, elle le doit à un choix.
Pendant que l'Allemagne fermait ses dernières centrales en 2023, la France a conservé une filière nucléaire, ses ingénieurs, ses soudeurs, ses entreprises et ses capacités d'exploitation.
Flamanville a coûté cher et accumulé les retards. Il a aussi rappelé combien les compétences industrielles sont difficiles à maintenir et coûteuses à reconstruire lorsqu'on les laisse s'éroder.
Aujourd'hui, la France possède encore une filière capable de porter un programme de cette ampleur.
Personne ne demande de dispenser EDF de tout contrôle. Reste à financer cette capacité industrielle sans compromettre le calendrier.
On peut contrôler la puissance d'un acteur sans oublier que cette puissance repose aussi sur des compétences que l'Europe a intérêt à préserver. -
David Cournapeau reposted
Marcin Piatkowski
@mmpiatkowski
Poland 🇵🇱 and South Korea 🇰🇷: two global growth champions with two different growth models and social outcomes.
Over three decades both grew fast (4.2% vs 3.7% a year per capita) and reached the same productivity (~$52/hour), but with very different R&D spending: 1.4% vs 5.1% of GDP.
On social outcomes, Poland ranks #24 in world happiness vs Korea's #67, with half the suicide rate and a third of the old-age poverty (13% vs 40%) 👇
The shared challenge: among the lowest fertility rates in the world, 1.07 and 0.80.
@Noahpinion @KoreaEconInst -
David Cournapeau reposted
日本経済新聞 電子版(日経電子版)
@nikkei
サカナAIが国産AI開発へ 防衛・サイバーに活用、米中依存リスク排除
www.nikkei.com/article/DGXZQOUC261H80W6A021C2000000/?n_tw… -
David Cournapeau reposted
Marginal Revolution
@MargRev
Tyler Cowen’s Strongest Book Recommendations, 2003-2026 marginalrevolution.com/marginalrevolution/2026/10/tyler-c…
-
David Cournapeau reposted
Hoover Institution
@HooverInst
What if breaking bad habits has less to do with willpower—and more to do with your surroundings?
Psychologist Angela Duckworth joins @EconTalker to explain why changing your environment may be the smartest way to change your behavior.
Watch a new episode of EconTalk from the Hoover Institution, @Liberty_Fund, and @Econlib: www.hoover.org/plus/research/how-changing-environments-ca…This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Kai Arulkumaran
@kaixhin
Here we go! @SakanaAILabs' Chief Scientific Advisor @SchmidhuberAI is here 🚀
-
David Cournapeau reposted
Nicolas Pinto
@npinto
You're welcome.
-
David Cournapeau reposted
Pierre Beyssac 🏴☠️🇫🇷🇪🇺🇺🇦
@pbeyssac
Montebourg oublie juste, et c'est très dommage, que c'est l'UE qui protège le mieux nos droits fondamentaux, alors que les gouvernements français successifs depuis des décennies sont obsédés de surveillance, d'identification généralisée et de censure.
Virginie Debuisson
@VirgoWhallala
Sans aucunement verser dans un discours anti Europe, @montebourg développe ici la question majeure : on est vassalisés par les US, on subit la submersion chinoise, mais nous sommes extrêmement contraints par l’UE qui est à l’origine de 60% de nos lois intérieures. À modifier.
-
David Cournapeau reposted
Stella Biderman
@BlancheMinerva
Analysis by @KelseyTuoc is consistently ahead of the curve.
www.theargumentmag.com/p/the-core-anti-pangram-constituency
Paul Calcraft
@paul_cal
Anti-AI YouTube channel with millions of views & 150k subs is 96% AI scripted according to Pangram
-
David Cournapeau reposted
Colonel Farès
@Le_ColonelFares
- J'me suis jamais fait baiser deux fois de suite !
- Eh ben, tu sais pas c'que tu perds. 🔥
🎂 Née un 25 octobre, Rosemonde du Bois de la Faisanderie
Merveilleuse Annie Girardot 💗
Bernard Blier & Dédé Pousse
📽️ Elle cause plus… elle flingue (Michel Audiard 1972)This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Eric Lécuyer
@eric_lecuyer_
Photo rare de la livraison d'un bébé
-
David Cournapeau reposted
Gérard Araud
@GerardAraud
La crise fait le bonheur de Moscou, Washington et d’autres qui soufflent sur les flammes mais ses origines sont globales et communes aux démocraties libérales. C’est de celles-ci dont il faudrait débattre à l’occasion de notre campagne électorale. Nul ne le fait.
Brice Couturier
@briceculturier
Les mêmes scènes d'émeutes et de pillages à Rome, Barcelone et Paris au même moment. Qui peut nier qu'il existe un chef d'orchestre quelque part, déterminé à achever la ruine et l'effondrement de nos démocraties européennes ? Mais qui ?
-
David Cournapeau reposted
Pierre Beyssac 🏴☠️🇫🇷🇪🇺🇺🇦
@pbeyssac
Le yakafokon à la française mène directement à l'autoritarisme. Et il est très facile d'en être involontairement complice, par exemple lorsque nous exigeons l'interdiction des réseaux sociaux aux mineurs. Notre méconnaissance technologique est exploitée contre nous.
Pierre Col
@PierreCol
J'aimerais que ces incompétents nous expliquent comment, techniquement, ils pourraient "bloquer temporairement un réseau social dans une zone".
🤡 -
David Cournapeau reposted
Jen工智能
@HumanLevelJen
Visual proof of the Banach-Tarski paradox
Kitty Cat Empire
@KittyCatEmpire
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Mary
@MaryJackalope
Me: you should probably freeze your credit cards because of the cyber stuff
Husband: I should at least check the purchases more often
Me: you know what I think this cybersecurity threat is overhyped -
David Cournapeau reposted
Maxime Sbaihi
@MxSba
En un demi-siècle, la France a divisé par trois le taux de pauvreté des retraités et multiplié par deux celui des jeunes.
La pauvreté a changé d'âge, mais nous faisons semblant de ne pas le voir. -
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
I arrived in London for work and had an extra day.
“Oh I should rent a car and see Stonehenge! I slept well on the flight.”
I get there and run into Boris Johnson in the parking lot.
Life truly is bizarre sometimes.
(I asked him to recite some Greek and he said no) -
David Cournapeau reposted
wald 🪿
@Wald52Wald
Faites comme @briceculturier et demandez de baisser les retraites géantes des boomers
Brice Couturier
@briceculturier
Avec 31,6 % du PIB consacré aux dépenses sociales (contre 26,7 % en Allemagne), la France est championne d'Europe de la redistribution. Le problème, c'est qu'une partie extravagante de ces sommes est simplement empruntée. Qui de sensé peut imaginer que cela va durer ?
-
David Cournapeau reposted
Michael Fowlie
@mwfowlie
I’ve been discouraging the use of smart TVs for years.
This is worse than I feared.
Do NOT connect your TV to the internet. Use a device like an Apple TV box instead.
log4j
@log4j
this is pretty evil and clever.
so if your samsung TV model doesnt have a microphone, it will look for connected devices with one, such as a soundbar or a connected bluetooth device.
this gets shipped off to bixby. -
David Cournapeau reposted
-
David Cournapeau reposted
Pierre Beyssac 🏴☠️🇫🇷🇪🇺🇺🇦
@pbeyssac
On sent bien que le pays est en train de partir en vrille par la faute d'un gouvernement incapable de compromis et aux abois. Ceci en est une confirmation. C'est grave.
BLAST, Le souffle de l'info
@blast_france
Information exclusive Blast : le Premier ministre a préparé la coupure des réseaux sociaux pour le 24 octobre
Deux jours avant la manifestation du 24 octobre, le Premier ministre Sébastien Lecornu a demandé à Orange, SFR, Free et Bouygues Telecom de se tenir prêts à couper l’accès à TikTok, Instagram ,Snapchat et X. Blast révèle la lettre et le compte rendu de la réunion qui a suivi, où les opérateurs ont exprimé leurs doutes. Le gouvernement avait déjà utilisée cette mesure liberticide en Nouvelle-Calédonie en 2024, avant d’être désavoué par le Conseil d’État.
Par @yanmdi.
www.blast-info.fr/articles/2026/information-exclusive-bla… -
David Cournapeau reposted
Susan Zhang
@suchenzang
popo standing by this weekend like FAANG engineers who suddenly start performing when VP+ eyes are on the project
-
David Cournapeau reposted
Dan Luu
@danluu
We should be more ambitious with the software we build:
danluu.com/ambitious/ -
David Cournapeau reposted
Matthew Yglesias
@mattyglesias
The Nazi Party defined itself as part of Germany's right-wing camp, came to power via an alliance with other right-wing parties, targeted members of left-wing parties for purges once in office, etc.
These are not obscure facts. -
David Cournapeau reposted
Michel Goya
@Michel_Goya
Les influenceurs du chaos
Violence issue de l'extrême droite : faut voir, on est sûr de rien, et si c'est le cas, ce n'est pas notre faute mais celle de l'immigration, de l'islam, du grand remplacement, des racailles, c'est eux qui ont commencé, etc.
Violence issue de l'extrême gauche : faut voir, on est sûr de rien, et si c'est le cas, ce n'est pas notre faute mais celle de la police, du gouvernement, du capitalisme, de la colonisation, de la pauvreté, des riches, c'est eux qui ont commencé, etc.
A un moment donné, il faut assumer les conséquences de ses discours de division et de la rhétorique que l'on emploie. -
David Cournapeau reposted
Sylvain Catherine
@sc_cath
I have been trying to explain my Twitter activity to my colleagues over the past two years, but now French creative public accounting is on US night shows.
Kâplan
@KaplanBen_Fr
Même les shows de divertissement US savent que 32 % du budget de l'Education nationale française va aux retraites !
-
David Cournapeau reposted
Gergely Orosz
@GergelyOrosz
Brutal truth from @Lethain
“If you're in [middle] management (…) you sort of have to decide, am I retiring now or am I re-skilling?”
He did just this: step down from leading a large team (250 ppl) to be hands-on with a small team (25 ppl), ship code & re-learn the new “rules”
The Pragmatic Engineer
@Pragmatic_Eng
Why a 3x CTO went back to a 25-person team. From Will Larson @lethain, Imprint CTO and 4x book author.
"It's a hard time in middle management, right? There's a lot more competition for these roles. What I found personally very rewarding is getting back into the details of writing software.
When I left Carta, my team was around 250. When I joined Imprint, the team I was leading was 25.
This a unique moment to actually re-skill. If you're in management, I think right now either you have to get back into the details, or the world is changing so radically that the profession you know will not be valued in three years. You sort of have to decide, am I retiring now or am I re-skilling?
I think there has to be some openness to taking a backward step or a lateral step. What I knew I wanted was getting back into the details and relearning how to do software in the current age. That's what I believe is going to be the unlock to the longevity of my career."This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Graham Neubig
@gneubig
I bought a B300 machine for my lab at CMU.
I was thinking of using it to do research on LLMs, but now I'm thinking of whether I should just use it to serve deepseek-v4.1-flash tokens with the new deepseek inference code and fund the entire rest of my lab that way 😅
Yuchen Jin
@Yuchenj_UW
Heard from a neolab friend:
“Even if we fail to train the best model, we locked in GPU capacity early. H100/B200/B300 rental prices have since gone up 2–3x. We can just resell the compute and print money.”
OK, so that’s the business model. -
David Cournapeau reposted
JFPuget 🇫🇷🇺🇦🇨🇦🇬🇱
@JFPuget
Interesting. When will this be implemented in pytorch?
alphaXiv
@askalphaxiv
"A Fix for Softmax and Cross-Entropy"
Turns out two of the most fundamental functions in deep learning have numerical bugs that can actually destabilize training.
This paper proposes simple fixes to softmax and cross-entropy that prevent tiny gradients from incorrectly rounding to zero, or remaining nonzero when they should vanish, as these errors can cause sudden loss spikes and interfere with knowledge distillation.
On small-scale grokking benchmarks, the fixes eliminate Slingshot loss spikes and improve test accuracy from 94.3% to 100% in one setting. They also prevent student models from drifting away from their teachers during distillation, without requiring higher numerical precision.
www.alphaxiv.org/abs/2610.31642 -
David Cournapeau reposted
Bojan Tunguz
@tunguz
My prediction is that this will be true unless/until AI unlocks a *major* new way of exploiting the *physical* world, the way we got in the 19th and 20th century. Increased efficiency of the current paradigms are not gonna cut it.
Noah Smith 🐇🇺🇸🇺🇦🇹🇼
@Noahpinion
This is the median economist forecast actually
-
David Cournapeau reposted
Bojan Tunguz
@tunguz
We are increasingly close to beating the genetic lottery when it comes to looks.
And AI will help us beat genetic lottery for brainpower.
What a time to be alive.
krx
@grep_now
🚨 Hair Loss Cure Update 🚨
Eli Lilly might find the cure for hair loss. Keep an eye on them and the biotech company they’re investing heavily in, Amplifica.
With just one injection of AMP-303 into the scalp, thinning hair showed significant thickening. ONE injection.
Hair growth was monitored over six months. Each trial participant served as their own control, with one area of the scalp injected and compared against a similarly thinning area on the opposite side of the head.
Read here:
DOI: doi.org/10.1016/j.jid.2025.06.1428
www.biospace.com/press-releases/amplifica-completes-26-mi…
Yellow: efficacy findings · Green: tolerability · Purple: study limitations and qualifications
#biotech #hairloss #science #peptides -
David Cournapeau reposted
Bojan Tunguz
@tunguz
Too real! 😭😭😭
-
David Cournapeau reposted
Philippe Lemoine
@phl43
Franchement il faut arrêter avec cette histoire maintenant. Je pense que cette fille est une pauvre conne, mais ça ne justifie pas qu’on scrute les moindres détails de sa vie en public de cette façon, quand bien même elle mentirait.
Jean-Jacques qui raque 🇫🇷
@JJquiraque
En tête de onze images sur Instagram, Lola Fleur écrit que ce sera sa dernière prise de parole sur sa situation familiale.
Elle y raconte que son père verse aujourd’hui 600 euros de pension alimentaire pour des revenus qu’elle situe au-dessus de 6 500 euros nets, qu’il a organisé leur ruine depuis le jugement provisoire, et elle joint des captures de messages, une attestation d’association et un procès-verbal de police.
Elle ajoute que l’intime est politique.
Le chiffre de 600 euros est posé sans indication du nombre d’enfants qu’il concerne, de leur âge au moment des décisions, ni de ce que la pension est censée couvrir exactement.
Les revenus du père sont donnés comme « de plus de 6 500 euros nets » sans pièce jointe qui permette de les vérifier, ni précision sur la période ou sur d’éventuels autres revenus.
La mère est devenue fonctionnaire après son Capes en 2020, mais ses ressources actuelles ne figurent pas dans les documents reproduits.
Sans ces éléments, le montant reste un chiffre isolé dont on ne peut mesurer la portée.
Les messages sélectionnés datent de 2023 et 2024. Ils montrent des échanges sur de l’argent, sur un dîner d’anniversaire, sur des contraintes liées au divorce.
Ils ne sont pas accompagnés de l’ensemble des échanges sur la même période, ni des jugements complets qui ont fixé ou modifié la pension.
L’attestation d’association couvre 2019-2020 et le procès-verbal concerne un non-paiement en 2023. Ces pièces existent ; elles ne permettent pas de vérifier l’intégralité du récit des années suivantes sans le dossier judiciaire.
Elle précise dans un post-scriptum que sa « garde-robe de milliardaire » est faite de copies ou d’achats en vide-grenier, et que ses vacances passent par Ryanair et des auberges de jeunesse.
Aucune facture, aucun relevé, aucune photo datée n’accompagne cette précision.
La formule arrive après qu’elle a exposé l’histoire familiale en détail, y compris la demande de sa plus jeune sœur d’être accompagnée à la Fête de l’Humanité en septembre 2026.
Annoncer que c’est la dernière fois, puis publier onze stories avec des documents partiels, des messages choisis et des chiffres sans les pièces qui permettraient de les recouper, place le récit dans l’espace public tout en laissant les éléments de vérification à l’extérieur.
L’intime est politique sert ici de justification à cette publication, sans que les données fournies suffisent à elles seules à contrôler ce qui est avancé sur les montants, les revenus ou les achats. -
David Cournapeau reposted
Michael Fowlie
@mwfowlie
In terms of Nvidia shares it lost even more value.
But that’s just as useless a comparison as gold. -
David Cournapeau reposted
Brian Lui
@brianluidog
You'll never guess the video that this tweet is explaining
보일라리
@Sleeboiler
1. 원자 구조의 차이
•수소(\(H\)): 원자번호 1번으로, 양성자 1개와 전자 1개로 이루어진 가장 간단하고 가벼운 원소입니다.
•
•헬륨(\(He\)): 원자번호 2번으로, 양성자 2개, 중성자 2개, 전자 2개로 이루어져 있습니다. [1, 2, 3, 4]
•
2. 화학적 안정성과 반응성
•수소: 전자 1개가 부족하여 다른 원소와 쉽게 결합하며, 반응성이 매우 높고 가연성(폭발성)이 있습니다.
•
•헬륨: 가장 바깥쪽 전자껍질이 완전히 채워진 비활성 기체입니다. 다른 물질과 거의 반응하지 않으며 불연성(폭발 위험 없음)으로 매우 안전합니다. -
David Cournapeau reposted
JohannesBorgen
@jeuasommenulle
🤣🤣🤣🤣
We have a winner.
Thomas Porcher
@PorcherThomas
D'ailleurs, la taille du spread franco-allemand à long terme pourrait même devenir un problème pour l'Allemagne pour financer sa dette car, en l'absence de risque, les prêteurs prêtent aux pays avec des meilleurs taux.
-
David Cournapeau reposted
Michael Nielsen
@michael_nielsen
Lots of motion toward increasing notions of provable security and secure by design. Hard to know where the equilibrium is, since side channels will always be with us. But a lot of systems are about to get much more secure
-
David Cournapeau reposted
Andrzej Cała
@Andrzej_Cala
tego dnia w 2009 r. ukazał się, nakładem Stones Throw Records, debiutancki album Dâm-Funka "Toeachizown". dla zdecydowanej większości z nas była to postać w zasadzie anonimowa. natomiast w środowisku muzycznym z Pasadeny i na rapowej scenie Los Angeles kojarzono go z racji dogrywek perkusji czy klawiszy, czasem pomocy producenckiej. w pełni rozkwitł grubo po trzydziestce, gdy skumał się z Leonem Sylversem III z legendarnego bandu soul-R&B The Sylvers i to właśnie Leon okazał się jego mentorem w świecie produkcji, stojąc też za albumem Dâma jako producent wykonawczy. na album składało się pięć oddzielnych, niezależnych płyt ("Latrik", "Fly", "Life", "Hood" i "Sky") i jest to rzecz po prostu przepiękna. nie do zamknięcia w gatunkowej szufladce, określana jako neo-boogie, modern funk, post-disco, jako współczesne spojrzenie na wszystko, co na scenie czarnej muzyki z Los Angeles dominowało od lat 70. aż do przełomu wieków. kompozycje opierają się o wirtuozerskie (ale w sensie melodii, groove'u) popisy na syntezatorach i kosmicznym basie. to muzyka progresywna i bardzo oldskulowa jednocześnie, pełna kolorów, słońca, ciepła. marzycielska na całego
#całapłytoteka -
David Cournapeau reposted
sunil pai
@threepointone
my bad, we write these out by hand on demand and I was hungover and half assing it
Cameron Conn
@cameronconn
Oh come on Cloudflare
-
David Cournapeau
@cournape
Maybe AI causing the cyber security incidents is just AI committing suicide: the stock market collapses, investments dry up and pacing happens naturally
-
David Cournapeau reposted
Pete Skomoroch
@peteskomoroch
I’ve known Kevin and DJ for over a decade and worked for DJ at LinkedIn. Excited for @warpsciences: tighter OODA loops for science. AI helps plan experiments, robots execute, results improve the models. Faster cycles of observing, orienting, deciding, and acting to discover new medicines. 🚀
Kevin Weil 🇺🇸
@kevinweil
💥 Introducing Warp Sciences, a new company applying AI and robotics to help the world develop new medicines faster.
We’ve raised a $215M Series A led by @lightspeedvp to bring frontier AI out of the browser and into the lab, and build scientific capacity here in America.
I co-founded Warp with my long-time friend @dpatil, FAIR co-founder @manohar_paluri, and drug hunter Sol Reisberg. The early team is full of incredible people like @bkspears9, who helped achieve Q>1 fusion at LLNL. -
David Cournapeau reposted
Gearoid Reidy リーディー・ガロウド
@GearoidReidy
A terrible year for Japanese literature after the death of Higashino Keigo earlier this year
-
David Cournapeau reposted
Gearoid Reidy リーディー・ガロウド
@GearoidReidy
Ryu Murakami, the acclaimed Japanese author of Coin Locker Babies and In the Miso Soup, has died at the age of 74.
オリコンニュース
@oricon
作家の村上龍さん死去 74歳
『カンブリア宮殿』司会を約20年担当
www.oricon.co.jp/news/2488333/full/?ref_cd=jstw003
⠀
1952年生まれ、長崎県佐世保市出身。76年、「限りなく透明に近いブルー」で芥川賞などを受賞、ほか代表作に「コインロッカー・ベイビーズ」「13歳のハローワーク」「五分後の世界」などがある。 2006年からはテレビ東京の長寿番組『カンブリア宮殿』でMCをつとめ、今年3月に小池栄子とともにMCを卒業した。 -
David Cournapeau reposted
Brian Lui
@brianluidog
"Your child plays the piano very well."
"Thank you. Would it be convenient for him to stop playing the piano very well at 8:30pm, or would another time be better?"
けものみち | 獣道 | Kemo🫡🚫🙏 |🇺🇸🇹🇼🇮🇱🇯🇵🎗️⚜️
@KemonomichiNL
In Japan, if your neighbors commented on how well your kid plays the piano, don’t get too excited
It means you guys are being too fucking loud & you better invest in some sound proofing upgrades -
David Cournapeau reposted
Mark Farina
@djmarkfarina
Artwork for Depeche Mode's 1983 album 𝘾𝙤𝙣𝙨𝙩𝙧𝙪𝙘𝙩𝙞𝙤𝙣 𝙏𝙞𝙢𝙚 𝘼𝙜𝙖𝙞𝙣. Photos by Brian Griffin.
-
David Cournapeau reposted
Nassim Nicholas Taleb
@nntaleb
3/ Some people are arguing that new technologies have traditionally created new jobs. Maybe--let's wait. But it is the old jobs that we are talking about.
-
David Cournapeau reposted
Graham Neubig
@gneubig
1Password
@1Password
Today we announce 1Password for Agents platform, building with companies @Muse, SpaceX Super Intelligence Grok @Bot, @AnthropicAI’s @Claudeai, @OpenAI, @TownAI, @Browserbase, @Browser_use, @heydana_ai, Instinct, @usekernel, @Perplexity_ai, @tryramp, @Tomo, and @wajo_ai, with more integrations on the way. 🎉
We've heard from so many of our customers who want to use these new personal AI agents to help them get more done, but want more control over what the agent can access with the same trust and ease of use they get when using 1Password directly.
1Password for Agents gives your agents access without giving up your passwords. When an agent reaches a sign-in form, it requests a credential from your vault. If you approve the request, 1Password directly auto-fills the credential in the destination site so the agent can’t see or take possession of the credentials.
The agent gets the task done, but it doesn’t get your passwords..
You stay in control. 1Password for Agents free and optional for personal and family accounts. Unless you opt-in to an agent connector, agents cannot touch your vault.
➡️Learn more: 1password.com/personal-ai-agent-integrationsThis media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
David Bowie News
@davidbowie_news
David Bowie, Japan 1980, by Masayoshi Sukita
“I received a call from Kyoto, where Bowie-san was staying. He had already made himself a local. We went to a local market, rode the local trains and went out to a disco at night. Most of the people around him were unaware who he was” -
David Cournapeau reposted
Sam Jarman
@samjarman
-
David Cournapeau reposted
Philippe Lemoine
@phl43
Yes, but not so much that it would have been possible to do what Niemetz is proposing, because the truth is that such a fiscal consolidation is impossible as long as people in France don't feel the urgency of the situation. In a way, Macron's presidency was killed by the yellow vests protests in 2018, he never really recovered from that. This mess has been a long time coming and is the result of decades of cowardice and irresponsibility. It will be very difficult to get out of the hole in which we put ourselves.
-
David Cournapeau reposted
Béatrice🔸Leydier 🥐
@BeaLeydier
Le gouvernement va nous ajouter une attestation papier de signature du pacte républicain ™️ pour pouvoir sortir de chez soi et nous enlever les jeux vidéos avant de faire appliquer nos lois tout simplement.
-
David Cournapeau reposted
Béatrice🔸Leydier 🥐
@BeaLeydier
On est vraiment dans le n'importe quoi le plus total du gouv, comme s'il voulait détruire notre état de droit à défaut de pouvoir le protéger.
Faites respecter la loi en arrêtant les émeutiers et pilleurs et ça sera déjà bien merci, on se passera d'innovations administratives
LCI
@LCI
🔴 Circulaire aux procureurs de la République : "J'ai demandé à tous les procureurs de la République (...) de prévenir tous les employeurs que leur salarié, leur agent, a été dans un pillage, a été dans une émeute", @GDarmanin
📺 @DariusRochebin -
David Cournapeau reposted
Alex Imas
@alexolegimas
New Substack post with @soumitrashukla9. Work has never been just about the income earned sans effort costs, as typical in economic models. Key aspects of people’s identity, their meaning, and sense of agency are all wrapped up with what they do for a living.
There has been a lot of conversation lately about how technological change, particularly AI, may disrupt this relationship, from Ezra Klein's column on his own mathematician father’s grapplings with losing meaning in mathematics, to Francis Fukuyama's post on AI, agency, and recognition. In this piece we take a historical perspective on how technological transformation has disrupted identity and meaning by affecting jobs, and how workers and society responded to these changes.
Much of the debate around AI and jobs has centered on wages and material consequences, but losing one's work is also intertwined with how people view their place in society. Most people know about the Luddites—skilled weavers and knitters—who protested the devaluation of their mastery. What’s less known is that the Luddites were actually experts in using the machines as well, and that their protests prompted prominent economists like David Ricardo to rethink the role of automation in society, and later inspired the Arts and Crafts movement. A century later, when recorded sound displaced 20,000 cinema musicians, their campaign against "canned music" and two-year recording ban pressured labels into funding live performance.
Soumitra and I have been working on versions of this essay for months. He led the piece, and we’re both really excited to get it out there.
Link to the post here: aleximas.substack.com/p/technological-progress-and-the-me…
Ezra Klein
@ezraklein
The most revealing part of this conversation between Jon Stewart and Jacob Coxon wasn't Coxon's warning about A.I. doom, but his description of A.I. utopia, and Stewart's incredulous response. I wrote about it — and Francis Fukuyama's "Last Man," and A.I. easily solving the problems that bedeviled my father's field of mathematics for decades — here: www.nytimes.com/2026/10/18/opinion/ai-bubble-catastrophe-…
The Daily Show
@TheDailyShow
Former Anthropic researcher Jacob Coxon (@hilbertspaess) and Jon Stewart discuss AI, jobs disappearing, universal income, and whether tech CEOs are creating a dystopian future
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Nick e/code
@nicksdot
You gotta be kidding me.
Paul Michael
@paulmichaeldev
No wonder people think PHP is dead.
Turns out... nope. -
David Cournapeau reposted
Naudin Christophe
@NaudinChristoph
Mais il fait exactement comme l’extrême droite en désignant « les noirs et les Arabes » (la version non hypocrite de « racisés ») comme les coupables. Qu’il les défende ou pas ne change rien. C’est ça qd on fait une analyse depuis la Closerie des Lilas. www.liberation.fr/idees-et-debats/tribunes/ce-que-les-cas… 3.
-
David Cournapeau reposted
Jeffrey J. Hall 🇯🇵🇺🇸
@mrjeffu
Japanese Prime Minister Takaichi Sanae listened to her advisors and issued a very relevant tweet about the PHP problem.
A lot of tech workers in Japan could use the PM's authoritative post to convince their bosses to take down websites until the fix was applied.
himanshu
@himanshustwts
The Prime Minister of Japan is tweeting to apply the PHP fix patch.
WTF. -
David Cournapeau reposted
Ze JeanPat
@Marminau
même avec une assurance, le reste à charge peut te faire plonger dans le rouge.
Je te donne un exemple : ma femme a dû subir une opération de l'épaule.
l'opération coûte $23,000 (pour 1h30 sur le billard mais peu importe) y compris anesthésiste etc. pas de nuit à l'hôpital.
1/2 -
David Cournapeau
@cournape
On notera que les gens qui se moquent de la soi disant propagande socialiste n’ont aucune idée de comment le système qu’ils défendent marche et du fait ont exactement le même niveau de raisonement
Laura Tessier
@LauraTessier10
« Aux États-Unis, je serais mort ou je vivrais dans ma voiture »
C’est pas Dédé du PMU du coin qui le dit, c’est un homme intelligent et cultivé.
Nous sommes endoctrinés par la propagande socialiste à un point que nous n’imaginons pas.
(PS : aux US il aurait été assuré) -
David Cournapeau reposted
Ze JeanPat
@Marminau
PS PS : aux USA, les factures médicales sont la première cause de faillite personnelle, assurance ou pas.
-
David Cournapeau reposted
Mathieu Blondel
@mblondel_ml
Our book "The Elements of Differentiable Programming" can now be read directly in the browser:
diffprog.github.io -
David Cournapeau reposted
Richard Gampell
@rjgampell
One reason is that universities today rely much more on endowments (and annual giving) than they did 40 years ago. MIT's endowment, in inflation-adjusted dollars, is >10x what it was 40 years ago. That takes an enormous amount of administrative effort.
-
David Cournapeau reposted
Dan Olsen
@danolsen
I'm excited to host @HamelHusain & @sh_reya for their AI Evals book launch Nov 12th at @LeanProdMeetup! @raindrop_ai is providing free signed copies! Learn more & RSVP at events.ticketleap.com/tickets/dan-olsen/evals
#ai #evals #productmanagement #prodmgmt
Hamel Husain
@HamelHusain
. @sh_reya and I are launching our new book Evals for AI Engineers Nov 12th at @DanOlsen’s @LeanProdMeetup!
You can join IRL in Mountain View or on Zoom. Our sponsor @raindrop_ai is providing free copies that we will sign for you if you come in person.
Here's the sign up link events.ticketleap.com/tickets/dan-olsen/evals -
David Cournapeau reposted
Cathryn
@cathrynlavery
If you use @tailscale and work off multiple computers add this to your AGENTS.md file:
"Any time you give me a link to something running on this computer (a preview, a review page, a dashboard), give me a Tailscale link, never localhost. I work across several computers and my phone. Publish it with tailscale serve, open the link yourself to check it works, then send it." -
David Cournapeau reposted
Thorsten Ball
@thorstenball
rogue agents: what a time to be in databases
Simon Eskildsen
@Sirupsen
what a time to be into databases
-
David Cournapeau reposted
Armin Ronacher ⇌
@mitsuhiko
As fun as AI is, we built so much bad software over the last 30 years, and now the universe is punishing us for our sins.
-
David Cournapeau reposted
-
David Cournapeau reposted
Vincent Koc
@vincent_koc
Never look under the hood of an eval
Kyle Wong
@ewveggies
Looked at GDPval public split for the first time, and this was the very first seed artifact I peeked at.
This is a frontier benchmark where "tasks are constructed from the representative work of industry professionals with an average of 14 years of experience".
And the public split is a key benchmark on Artificial Analysis btw. -
David Cournapeau reposted
Trevor McKendrick
@TrevMcKendrick
This will get fixed as a one time tax across all important parts on the Internet + our ability to build more secure software going forward will go up 10,000x
-
David Cournapeau reposted
Zaid Khan
@codezakh
I thought ppl were exaggerating about ICLR slop deluge but I legitimately had a paper in my ICLR batch containing meta commentary from the authoring LLM saying that the results in the paper do not match the paper's claims and are not statistically significant LMAO
-
David Cournapeau reposted
Daniel Jeffries
@Dan_Jeffries1
😂
Jon Yongfook
@yongfook
Introducing: Migraite
Instantly deploy an agentic swarm of CTO bots to migrate your SaaS product to whichever tech stack is not compromised that week. Monitoring and one click migration from $2000 / month. -
David Cournapeau
@cournape
I did not have the Japanese PM warning about php in my bingo card
-
David Cournapeau reposted
james hong
@jhong
Lots of VCs kindof sound like this: "Your basketball team doesn't have a moat, any healthy human can throw the ball into the basket thingy. "
People who have not run companies before don't understand how much the team matters more than some imagined structural moat. -
David Cournapeau reposted
-
David Cournapeau reposted
Christopher Manning
@chrmanning
Impressive performance, just a little below @thinkymachines’s Inkling with an over 10x smaller model.
But I think I’m still going to wait a little before buying my Mac Studio….
Recursive
@Recursive_SI
Recursive Flint is here: our first model, built with the Eureka Machine. Fast, local coding. Open weights available on Hugging Face.
-
David Cournapeau reposted
MattyG
@meawoppl
My SRE coming back from vacation tomorrow:
-
David Cournapeau reposted
Liu Liu
@liuliu
Everyone is building a harness because it is the last software needs to be built.
-
David Cournapeau reposted
tender
@tenderizzation
what if you just shot your infra engineers
Xingjian Bai
@SimulatedAnneal
1/9 Current optimizers are static during training. What if the optimizer could think and react while the network trains?
We let an agent watch the training dynamics and rewrite the optimizer's source code as training goes: a reactive optimizer.
In every language and image setting we tried that has no tuned recipe, it matches or beats the best recipe one can one-shot, and it guards against loss spikes and overfitting by itself. In the near future, optimization decisions may belong to agents, not humans.
📝 xingjianbai.com/reactive-optimizers/
💻 Try it on your task: github.com/xingjian-bai/opal
🧵 -
David Cournapeau reposted
Ramez Naam
@ramez
Unlike covid, we'll come out of this more secure.
Roko ⏸️🆁
@RokoMijic
We're in for a period of chaos as AI breaks human software.
12, 18, 24 months.
I think this could be like the covid of software -
David Cournapeau reposted
Daniel Lemire
@lemire
In case you have been wondering... Yeah, pretty much all science papers are written by AI (they were back in December 2025).
arxiv.org/pdf/2608.10715 -
David Cournapeau reposted
clem 🤗
@ClementDelangue
It feels pretty fitting for @JensenHuang & @nvidia to end their European tour with Microducks on stage.
Microduck was designed and engineered in Europe (more precisely Bordeaux, France) by the @pollenrobotics and @huggingface teams. With American capital and ambition + Chinese manufacturing genius, it became the most popular robot of 2026 with over 15,000 sold all over the world!
I believe Europe has more opportunity to shine in AI than in any previous tech wave. We intend to prove it with @huggingface @nvidia and the whole open-source AI community in the coming years!
Let's go European AI 🇪🇺🇪🇺🇪🇺This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Michel Goya
@Michel_Goya
Génial.
Chipyong-ni doit être mise en bonne place dans la liste des combats les plus héroïques de l'armée française et en plus c'est une victoire.
Cap-Hornier 🌊⚓️
@CapHornier_
Nouvelle émission !
Chipyong-ni, la bataille qui renversé le cours de la guerre de Corée.
4500 franco américains, encerclés par 25 000 chinois.
Reconstitution 3D, archives exclusives, et une bataille quasiment inconnue.
youtu.be/Z1vZr3r2xFY -
David Cournapeau reposted
bayes baes
@bayes_baes
type of guy that still is customizing his emacs config in 2026
-
David Cournapeau reposted
Thomas H. Ptacek
@tqbf
ICYMI:
sockpuppet.org/blog/2026/03/30/vulnerability-research-is-… -
David Cournapeau reposted
Garett Jones
@GarettJones
PSA: @ATabarrok was right, this book was very educational.
No need to read it all, as a rule: just pick a presidency you know and care about--I read Lincoln & Grant. You'll learn a lot. The first railroad boom made Lincoln!
A Context Is That Which Is Scarce book [3 vols.] -
David Cournapeau reposted
mattparlmer 🪐 🌷
@mattparlmer
I have said for years that software engineering is far too undisciplined, looks like we finally have to pay down all the security debt incurred by lazy practices
NSA Cyber
@NSACyber
Effective Immediately, all organizations using PHP for their systems are advised to be prepared for an update addressing two critical vulnerabilities. PHP is widely used for web applications. These vulnerabilities often provide a direct path to the server's file system or database. If an immediate update is not feasible, it is recommend that organizations take the system offline until it can be fully mitigated. NSA will follow up shortly with more specific guidance.
-
David Cournapeau reposted
Vincent Koc
@vincent_koc
End of an era
Polymarket
@Polymarket
JUST IN: Getty Images files for Chapter 11 bankruptcy after struggling with mounting debt & growing competition from AI-generated images.
-
David Cournapeau reposted
finbarr
@finbarrtimbers
Very impressed by this. Trying to resist the urge to rent a bunch of GPUs and start slinging tokens on OpenRouter
vLLM
@vllm_project
🚨🚨🚨 vLLM is all you need to reproduce DeepSeek’s performance.
With Wide-EP and EPLB, vLLM can match DeepInfer’s decode performance within 6% and generate $39/gpu/hr in revenue at DeepSeek’s off-peak pricing, at 200 P90 TPS on the AgentX workload.
Specifically:
• Under DeepSeek’s decode configuration (DeepSeek-V4.1-Flash, 131K context, nextn=3), vLLM on GB200 achieves throughput within 6% of DeepInfer’s published results across batch sizes 8–256.
• On AgentX, vLLM pushes the Pareto frontier further on GB300, reaching up to 1M tokens/s/GPU and generating $38.96/gpu/hr in revenue at 196 P90 TPS, based on DeepSeek’s off-peak API pricing.
• Also, vLLM extended low-latency kernel support for DeepSeek-V4.1-Flash, achieving up to 826 TPS per user on AgentX. -
David Cournapeau reposted
Jon Oberheide
@jonoberheide
You've gotta patch your dependencies super fast because they're full of vulns (PHP, Node, eMudhra, etc), but not _too_ fast because they're full of backdoors (LiteLLM). Good luck!
-
David Cournapeau reposted
Lazy IT Guy
@lazy_IT_guy
Until we get confirmation Jira is safe from the PHP vulnerability, I don’t think anyone should be opening tickets or assigning me work.
-
David Cournapeau reposted
Hakim El Karoui
@helkaroui
tous ces gens (lui, Knafo et les autres pseudo libéraux) qui parlent de redresser les finances publiques, s'ils ne parlent de la contribution des retraités à l'effort, ils se moquent du monde... ils veulent juste détruire l'Etat qui n'en peut mais.
-
David Cournapeau reposted
Alex Smith
@ninja_maths
If you feel the same, check out our mathematical foundations sequence.
Especially designed to propel adult students to the upper echelons of math as efficiently as possible.
mathacademy.com/courses
vivian
@vheeorji22
I want to relearn mathematics from scratch as a grownup
-
David Cournapeau reposted
Will Larson
@Lethain
Love getting to chat with Gergely! Ty @GergelyOrosz 🙏
Gergely Orosz
@GergelyOrosz
What’s changed in “the unwritten rules” of engineering management since the 2010s, why are hands-on managers “back”, and what has it done to the engineer-to-CTO career path? With Will Larson (@Lethain). Timestamps:
00:00 Intro
08:20 Stories from Yahoo!
12:47 Digg’s v4 rewrite: the insider story
26:02 Will’s path to eng leadership
30:03 Uber, Stripe, Carta, and Calm
35:11 What does a CTO do?
38:09 Will’s role at Imprint
42:48 How engineering management has evolved
51:08 ZIRP’s impact on tech
57:21 Career ladders
1:01:03 Eng 1 vs Eng 2
1:06:58 How software engineering roles are changing
1:16:25 Revised rules of engineering leadership
1:21:57 The risks of becoming a bureaucrat as an EM
1:25:57 How to become a CTO?
1:31:33 Middle management vs. line management
1:41:36 Will’s writing
1:52:13 Book recommendations
Brought to you by:
• @turbopuffer – vector and full-text search built on object storage: fast, 10x cheaper, and extremely scalable. turbopuffer.com/v3
• @EntireHQ – Git hosting, rebuilt for the agentic era. Entire hosts your code in-region, and is up to 89x faster than any other competitor. entire.io/pragmatic
• What if you only needed a single test to harden your system? Get a free introduction to property-based testing: through a live, online webinar hosted by Hillel Wayne, author of Logic for Programmers and Developer Educator at @AntithesisHQ. RSVP here: pages.antithesis.com/webinar-property-based-testing-pragm…
Two sage observations from Will on how engineering management changed:
1. ZIRP shaped engineering management practices in the 2010s, Will believes. As he put it:
“Back then, your ability to hire and close and retain became the differentiator between successful companies. The less successful ones simply couldn’t retain great people.
[This dynamic on hiring and retaining] pulled leverage back into the engineer’s hands and created a very different environment. So, companies were pitching themselves as a more relaxed environment, more focused on you, and offering more career development. And I think in ways that I think were often kind of self-defeating.”
2. Blunt message for middle managers: retire or re-skill.
Will left a team of 250 at Carta to head up 25 tech people at Imprint. He purposely took a “step back” in his leadership career because he believes things are changing so radically that the engineering leadership profession that we know will not exist in three years.
Amid all this, Will says that line management can be a lot more engaging for executives these days because the tools for line management and the tools for writing software are really good. Plus, the tech industry has always been volatile: during the 2010s most of us simply forgot about this!This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Vanessa Montalbano
@vanessamontlbno
My karaoke nights are not like yours
-
David Cournapeau
@cournape
Les événements récents m’ont fait réaliser que j’avais sous estimé le danger d’une partie de l’ extrême gauche et d’un certain militantisme pseudo ecolo.
-
David Cournapeau reposted
Gilles Raveaud
@RaveaudGilles
Récit terrifiant de ce qu'il s'est passé à Limoges dans "Le Figaro".
Les suppositions et déclarations politiques de la fin de l'article ne m'intéressent pas.
Mais les faits rapportés à son début sont fous ! -
David Cournapeau reposted
Steeve Morin
@steeve
il y en a qui sont beaucoup trop à l’aise quand même
Antoine Briens
@antoine_briens
Soutien total et inconditionnel à Papon. ❤️
-
David Cournapeau reposted
jessicat
@jessi_cata
Fantasy computer security problems: "Maybe by sending packets and detecting the timing of the vibrations of the computer case, it is possible for an attacker to gain information about the private key."
Real computer security problems: "The private key was 4. Not 4 digits. 4."
International Cyber Digest
@IntCyberDigest
‼️ BREAKING: The private keys of 8 certificates from Indian certificate authority eMudhra were compromised, Mozilla says. Firefox blocked them within 3 hours.
Each key should be a 116-digit number, but one was just 4.
Anyone can guess them and forge HTTPS certs for any website. -
David Cournapeau reposted
Rowan Fornow 🚵🚉🏙️🦣🇵🇸
@rowanfornow
Despite this being an Indian content farm account, they had the restraint to exclude India. I respect the discipline
Geography Map🗺️
@TheEpicStats
It’s wild how the world basically runs on nothing but this.
-
David Cournapeau reposted
Marc Gral
@MarcGral
Si il y a un 2nd tour Mélenchon-Le Pen, Le Pen pulvérise Mélenchon
Elle est élue, Mélenchon et ses lieutenants mettront le feu au nom de la lutte contre le fascisme, la 🇫🇷 est foutue pendant 5 ans, et ceux qui trinqueront en 1er seront les + modestes qui auront voté pour les 2. -
David Cournapeau reposted
Matthieu Gariel — eu/acc 🇺🇦
@MatthieuGariel
Et la France CRÈVE de cette absence totale de déontologie journalistique.
On invite pour le buzz, car l’invité passe bien, car ça fait de l’audience.
Le journalisme français ne mérite PAS son statut de pilier de la démocratie. -
David Cournapeau reposted
Hereward the Woke
@BamaExpat
Fleeing to Poland to escape right-wing populism is spectacular
WIRED
@WIRED
The internet told them Europe was a social paradise. Turns out the trains are awful, the health care isn’t free, and people can be really mean. www.wired.com/story/they-fled-trumps-america-now-they-reg…
-
David Cournapeau reposted
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
Yes. Wenfeng and Dario are both visionaries, idealists and cult leaders. Different league.
It’s beautifully dialectical, too. Open diffusion from a centralized collectivist authoritarian country, and secretive centralized authority from a “messy individualist pluralistic” one.
Kevin S. Xu
@kevinsxu
DeepSeek open sourced its internal inference framework, Apache 2.0, no strings attached.
This is arguably more impactful than open sourcing any model weights, like @zephyr_z9 said, they've open sourced their economics, how DeepSeek makes a profit
DeepSeek's commitment to open source is as steadfast as Anthropic's commitment to closed source. Every other lab sits in the mushy middle, dictated by the market condition, investor expectation, policy wind
These two labs are basically building two different religions, one radically open and messy, the other intensely closed and polished.
If you are on Team Bazaar, you root for DeepSeek. If you are on Team Cathedral, you root for Anthropic. In the end, both will exist and flourish, pushing and pulling each other into a steady state. I think that is ultimately a healthy thing.
I love visiting cathedrals, admiring their awe-inspiring grandeur, meditating under the instant hush and silence the structures command. I also like walking through bazaars, feeling the chaos, smelling the messy dynamism, brushing up against unpredictable creativity.
Both should exist. No reason to try to kill or demonize one or the other. -
David Cournapeau reposted
Phillip Carter
@_cartermp
This might be one of the funniest (bad) computer jokes of all time
⊹ damir
@otomir23
a trusted intermediate CA just "leaked" (the private key was 4)
-
David Cournapeau reposted
Mario Zechner
@badlogicgames
i guess we are now sending each other books that explain designs. at least @mitsuhiko put some work in to make it look really nice and easy on the old eyes.
-
David Cournapeau reposted
Hamel Husain
@HamelHusain
New blog post 🎉: How To Use Decision Models For Evals
In this post, we show you how to choose a decision model for your task, tune its threshold (don't assume the 0.5 default is good!), and interpret its scores.
Lots of cool interactive charts here. x.com/i/article/2112552453422104576 -
David Cournapeau reposted
Matt Turck
@mattturck
You misunderstand: they’re protesting because they want to KEEP all of this and the French state cannot afford it anymore
-
David Cournapeau reposted
JFPuget 🇫🇷🇺🇦🇨🇦🇬🇱
@JFPuget
There are several tools, like this one (I see other on comments) github.com/Graphify-Labs/graphify
-
David Cournapeau reposted
BAMBOULI
@Bamboubouli
Je ne verrais plus le modélisme ferroviaire de la même manière.
@BB27000
Massimo
@Rainmaker1973
Static grass applicators use relatively high voltage (40,000 V) to apply and recreate realistic grass in diorama and model railroads, and the process is quite mesmerizing
[📹 Model Railroad Hobbyist] -
David Cournapeau reposted
Mark Farina
@djmarkfarina
Yellow Magic Orchestra meets Throbbing Gristle, October 16, 1980, backstage at the Hammersmith Odeon, London.
-
David Cournapeau reposted
Armin Ronacher ⇌
@mitsuhiko
Opus 5.5 is such a good writer that I really enjoy reading its reports now. Such a huge improvement over Fable.
-
David Cournapeau reposted
Vélo du Marais
@Velodumarais
On a compris, Lisnard il a des grosses couilles et il envoit chier ces femmes qui se mêlent de politique. C'est donc bien le candidat des masculinistes.
-
David Cournapeau reposted
Daniel Lemire
@lemire
Most people who do machine learning with Python use PyTorch.
Amazon has its own hardware for machine learning: trainium. On AWS, trainium instances are roughly half the price of Nvidia instances... although you get somewhat less compute. Overall, AWS claims that trainium is 30% cheaper than Nvidia.
Amazon just made it much easier to run PyTorch on its own chips. TorchNeuron is a native PyTorch backend. Existing PyTorch code can move to Trainium by changing the device string:
model = MyModel().to('neuron')
Trainium chips are available on AWS EC2.
This is still in beta.
Code at github.com/aws-neuron/torch-neuronx -
David Cournapeau reposted
Armin Ronacher ⇌
@mitsuhiko
You gotta love Deepseek. This is awesome stuff! github.com/deepseek-ai/DeepInfer
-
David Cournapeau reposted
Philippe Lemoine
@phl43
Le problème fondamental c'est que les gens ne comprennent pas que, pour financer un État-providence, il faut s'appuyer sur une base d'imposition très large et taxer la classe moyenne parce que les sommes requises sont énormes et qu'il n'y a pas d'autre choix.
En pratique, ça veut dire que d'une manière ou d'une autre, le financement repose principalement sur les revenus du travail, parce que non seulement c'est de loin la principale composante du PIB dans une économie moderne mais qu'en plus les revenus du capital sont beaucoup plus difficiles à taxer.
C'est pour ça que, dans quasiment tous les pays européens, le financement de la protection sociale repose principalement sur les cotisations et les taxes sur la consommation, qui directement ou indirectement sont principalement alimentés par les revenus du travail.
Même au Danemark, où l'impôt sur le revenu joue un rôle beaucoup plus important qu'ailleurs, l'assiette d'imposition est beaucoup plus large qu'en France et l'essentiel du financement repose aussi sur les classes moyennes, donc en dernière analyse principalement sur les revenus du travail.
Si vous consacrez 1/3 du PIB à la protection sociale, comme c'est le cas en France, vous allez forcément devoir financer ça principalement avec les revenus du travail, parce que c'est près des 2/3 du PIB et que les revenus du capital, outre encore une fois ils sont beaucoup plus durs à taxer, c'est 5 à 6 fois moins une fois prise en compte la dépréciation.
Il n'y pas de réserve de pognon dans laquelle on pourrait facilement taper sans que les classes moyennes doivent payer. On pourrait certes gratter un peu d'argent en taxant le patrimoine davantage, mais dès lors qu'on veut épargner les classes moyennes, ça n'ira pas chier très loin au regard des besoins si on ne réduit pas les dépenses.
Le seul moyen de financer nos dépenses de protection sociale, sachant qu'à l'heure actuelle on le fait à crédit en s'endettant à des taux qui ne font qu'augmenter, c'est soit de travailler plus soit d'augmenter la productivité et, comme on n'a pas de baguette magique pour augmenter la productivité et que ce n'est certainement pas en s'endettant pour financer les retraites plutôt que de l'investissement productif qu'on va le faire, il va falloir bosser plus ou couper dans les dépenses.
Si on continue de ne pas vouloir regarder la réalité en face, tout ce qu'on parviendra à faire, c'est rendre l'austérité encore plus douloureuse le jour où elle sera devenue inévitable parce que les investisseurs ne voudront plus nous prêter qu'à des taux exorbitants et, en attendant, mettre encore plus à l'os les services publics parce qu'il faut bien trouver le pognon pour financer les retraites et les dépenses de santé quelque part.
Philippe Lemoine
@phl43
Quelles sont les "revendications légitimes" des manifestants exactement ? Ils ne sont pas contents parce que le revenu disponible réel n'augmente pas voire recule, mais ce n'est pas une revendication ça, c'est un sujet de mécontentement. Donc qu'est-ce qu'ils proposent concrètement ?
La réalité c'est qu'ils seraient incapables de faire la moindre proposition concrète réaliste. Si on leur demandait, tout ce à quoi on aurait droit ce serait des slogans débiles, plus de pognon qu'on n'a pas et la pensée magique habituelle selon laquelle il n'y a qu'à "prendre aux riches".
Mettons qu'on rétablisse l'ISF et qu'on instaure la taxe Zucman, honnêtement pourquoi pas si c'est le prix à payer pour qu'on puisse faire le reste, ça rapporterait 20-30 milliards à tout péter et il faudrait encore en trouver 100 de plus, donc où tous ces gens avec leurs "revendications légitimes" proposent qu'on trouve ça ? -
David Cournapeau reposted
Abou Djaffar
@GrireJacques
Il y a les poubelles au coeur de l'été. Il y a les sales grippes. Il y a les fosses à purin. Il y a les toilettes de la gare d'Agen. Il y a la peste. Il y a l'armée russe. Et ensuite, au plus bas de l'abjection, il y a les rédactions du groupe Bolloré. www.liberation.fr/checknews/la-dgsi-dement-que-des-armes-…
-
David Cournapeau reposted
Jo Kristian Bergum
@jobergum
True, this is a good post. I would put even more emphasis on workloads.
Too many garbage posts involving «first principles», «dirt cheap», «object storage», «fast», «faster» without going into the workload characteristic.
Ben Dicken
@BenjDicken
Most of you need to read this
planetscale.com/blog/on-benchmarking -
David Cournapeau reposted
Sagiv Ofek
@sagivo
We stopped trying to evaluate the AI skill directly. Now we ask managers to walk us through the last thing they shipped with AI — what they delegated, what they checked by hand, and what broke. Tells you everything in about five minutes.
-
David Cournapeau reposted
Pinboard
@Pinboard
I wrote a blog post about the dangers of small pieces of self-reproducing code shaped by natural selection infesting the internet of tomorrow. idlewords.com/2026/10/cloud_rats_and_software_roaches.htm
-
David Cournapeau reposted
Philippe Lemoine
@phl43
Once again, this is a false dichotomy. The protests were about the cost of living, but were it not for far-left activists and children of immigrants, France wouldn't be burning and there sure as hell wouldn't be groups of hoodlums going door to door looting and terrorizing people in Paris right now.
The people who deny that are either clueless or lying to you. Just open your fucking eyes and believe what they show you. The reason why you don't see videos showing predominantly white crowds looting stores isn't because of a conspiracy to hide the truth from you, that's because it doesn't happen.
MenchOsint
@MenchOsint
Fun fact: France isn't "burning" because of migrants. It's experiencing an intense cost of living crisis fueled by ... the global economic shockwaves of Trump's war on Iran.
-
David Cournapeau reposted
JT
@jiratickets
Next year, I will assign one of my jira tickets to every kid in America.
This will be the first of many. -
David Cournapeau 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. -
David Cournapeau reposted
Antoine Levy
@LevyAntoine
Most international observers have not wrapped their heads around a simple fact: France’s 2-round election, combined with MLP’s current lead, means anything that shifts votes from centrists to MLP (eg riots) mostly benefits Melenchon’s chances of accessing the runoff.
Mujtaba Rahman
@Mij_Europe
Why would @JLMelenchon incite violence knowing it will help Le Pen? Because he wants Le Pen to win and he becomes leader of the opposition. His people talk of a third round of the election "on the streets" if @MLP_officiel wins. If so, Saturday's violence will seem mild.
-
David Cournapeau reposted
M6 Info
@m6info
Lors de la journée de mobilisation nationale du 17 octobre, Pierre Collin, gérant du Micromania de Nice a été agressé alors qu'il tentait de protéger sa boutique. La séquence est devenue virale. Nous l'avons rencontré à la sortie du commissariat de Nice où il a porté plainte.
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Hanno Lustig
@HannoLustig
Spending on core state functions (net of pension contributions) in France compared to peers. To compute these shares, I removed the State’s imputed pension contributions for teachers, soldiers and other civil servants from each function and from total spending-- gives a better sense of what the government spends on teaching kids and keeping the country safe.
thetwocents.substack.com/p/the-french-state-crowded-out-b…
Tomas Pueyo
@tomaspueyo
French students complain that France doesn't spend enough on education. Well, it does: $13,787 per pupil per year, 1.4% more than the OECD average, and more than countries like 🇰🇷🇩🇪🇦🇺🇨🇭🇪🇸🇨🇦🇮🇹
But 32% of that money doesn't go to education, but to pensions! -
David Cournapeau reposted
Slava Akhmechet
@spakhm
This is net bad for the Republic. If you go back to any time in history, take a very powerful person (eg JP Morgan, Pompey, etc.) and ask would it have been good if they got *more* power the answer is obviously no. It is highly unlikely the answer is uniquely yes in our time.
-
David Cournapeau reposted
Federico Vaggi
@F_Vaggi
I think the purpose is also legitimacy not mere accuracy. If it was accuracy we would only allow people with a calibrated forecasting record to serve on juries and I haven’t even seen the most extreme rationalist advocate that.
-
David Cournapeau reposted
Rachel Moiselle
@RachelMoiselle
Presumably Spanish tourists outside the Louvre screaming ‘hijo de puta’ in protest of the French people protesting there has to be one of the most European things I’ve ever seen.
Luc Auffret
@LucAuffret
Des touristes venus visiter le musée du Louvre insultent et font des doigts d'honneur aux étudiants qui bloquent le musée depuis ce matin.
#louvre #paris #17octobre #24octobreThis media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
BB27000
@BB27000
Je suis cheminot.
Je suis de gauche (et j’assume).
J’ai été syndiqué, et j’ai fait (et organisé) des manif, dans des villes moyenne de province.
Les casseurs n’ont jamais été utile au syndicalisme.
Ils sont utile à un parti politique (précisément à deux d’entre eux).
L’ un pour foutre le bordel et faire plaisir à son électorat.
L’autre dénonce le bordel, et dit qu’il remettra de l’ordre.
Quand j’étais petit syndicaliste (année 2000), les manif c’était simple.
En tête de cortège, la banderole (porté par les chefs des syndicats ), c’est valable dans toutes les villes de France.
Ensuite, les syndicats défilent par ordre de taille.
Il y avait des services d’ordres, et les troupes étaient clairement cadrées. (Obéissance au mots d’ordres venant de la fédération).
Et à la fin du défilé, y’avait les nanard (les anarchistes).
Les nanards, ils avaient des drapeaux rouge et noir avec des chats dessus. Facile à reconnaître.
Avec parfois de la casse, mais à la fin de la manif. Nous on etait déjà dans les bus pour rentrer à la maison.
Ça c’était avant Sarkozy.
Sarkozy, il a changé tout ça.
Tout d’abord, il ne répondait jamais positivement aux demandes des syndicats, il laissait pourrir le truc.
Ça a affaiblit les syndicats.
Dans le même temps, les nanards qui étaient en queue de cortège sont passé en tête.
C’est lâ que sont apparu les blackbloc (surtout médiatiquement)
Deux choses. Les casseurs, les syndiqués ne les aimait pas.
Ça bloque le cortège, ça fait pleuvoir de lacrymo, ils s’en prennent même au service d’ordre des syndicats.
MAIS, certains disaient que comme sarko ne nous écoutait pas, ben c’est normal que les gens se révolte.
Sauf que, sauf que, les blackbloc ont complètement brouillé le messages.
Sur les chaînes d’infos, on avait 1 heure de casseurs en direct live pour 5 min de revendications portées par les chefs des syndicats.
Ça les a rendu inaudible.
Madame Michu, devant sa télé, elle voyait le Mac Donald en feu, avec des mecs en noir qui sautent partout. Madame Michu, elle comprends pas, et du coup, elle a tendance à vouloir voter pour des politiques qui promettent de remettre de l’ordre. (Aka la droite)
Cette technique de Sarko ne laisser pourrir la situation a été reprise avec succès par Hollande (loi El Komhri) et encore plus par Macron.
Dans des entreprises à fort pouvoir syndical (Aka, la SNCF), cette technique à affaiblit leur capacité.
Ça, et la création du CSE par Macron.
De fait, les syndicats n’ont pas vu arriver les gilets jaunes.
Un mouvement populaire, avec la aussi une partie violente (blackbloc). Et ils n’ont jamais pu clairement récupérer le mouvement à leur avantage.
Macron ayant affaiblit les syndicats, il s’est pris un retour de bâton dans la tronche.
Qu’on le veuille ou non, le syndicalisme est un contre pouvoir essentiel dans une démocratie.
Quand bien même il a des défauts, et dieu sait qu’il en a .
Les mouvements sociaux des derniers jours, clairement soutenu par un parti d’extrême gauche n’est possible que parce que les syndicats ne « gèrent » plus les manifs.
Les politiques (de droites) ont voulu se passer des syndicats, on en paye aujourd’hui le prix.
Ce n’est certes pas le seul motif de tout ça, mais c’est une partie de l’explication. -
David Cournapeau reposted
Dōgen
@Dogen
breaking: japan to change kanji for france from 仏 to 暴
-
David Cournapeau reposted
Matthieu Gariel — eu/acc 🇺🇦
@MatthieuGariel
But Le Pen is full socialist, pro Russia & pro boomer 😭😭😭
-
David Cournapeau reposted
Peter Steinberger
@steipete
AI;DR
Trung Phan
@TrungTPhan
The usefulness of Amazon’s 6-page memo has been nuked by AI-generated text (employees sit in meetings suspecting text is AI and tune out).
Suspicious memos tagged as AI;DR (AI, didn’t read). -
David Cournapeau reposted
John Arnold
@johnarnold
If the unrest in France is a reflection that aging populations and tightening fiscal constraints make it increasingly difficult for Western governments to meet their promises to and expectations of their citizens, then France is just the beginning.
-
David Cournapeau reposted
Bradford Cross
@bradfordcross
I’ve worked on AI my whole life.
I’m very bullish about AI.
Im very concerned about the real immediate risks of AI.
BUT I’m very disappointed by otherwise serious people in AI that claim p(doom) is 20% or GDP growth is going to 20% by 2030 without sharing the data and statistical foundations of their claims.
If you really believe those numbers, I’m happy to entertain a debate, but let’s start with your data and stats.
It’s deeply irresponsible for those of us with quantitative minds working in this field to make arbitrary random gut feel predictions to the public about important things like p(doom) and GDP.
If we claim to have public interest at heart, then let’s take our collective responsibilities more seriously in this regard and put data and stats behind our claims, otherwise be more reserved in our predictions.
Bradford Cross
@bradfordcross
Tech is 10% of US GDP.
GDP growth right now is around 2.5%. So folks are saying we need to roughly 10X the GDP growth rate.
Tech is responsible for 35% of growth in GDP in the past decade, so we’re saying AI would be entirely responsible for this 10X whereas all other tech combined has been only 35% of 2.5%.
GDP is 33.5T; growing it at 20% from 35T would be 7T annually.
My sense is that people don’t think about these numbers they just make things up, similar to the random p(doom) of 10-25% … it’s like, ok based on what model?
It’s ironic that the very quantitative people working on AI apply totally arbitrary gut feel predictions to things like p(doom) or GDP growth.
In AI, we’re supposed to be the data-driven scientific people — so let’s see some models behind 20% GDP growth or 20% p(doom) please.
I’ve worked in AI my whole life. I’m very bullish on AI and would happily look at any well founded attempts to back predictions like these with something tangible, but I worry there isn’t much behind it, and that its a net negative to state these kinds of predictions unless you have some real data and statistical basis for them. -
David Cournapeau reposted
Patrick McKenzie
@patio11
A few of us have been predicting saturation of adversarial contact points for a while, but this is first set of good graphs I’ve seen of it.
Imagine it will take owning agencies a very substantial amount of time to realize what is happening.
Sam Freedman
@Samfr
To give an indication of the problem here's a chart of 12 government agencies around the world and numbers of applications / appeals / complaints etc...
The dotted grey line is when ChatGPT was invented.
Sam Freedman
@Samfr
On the substack this week...
I looked at how states are being increasingly overwhelmed with AI written slop. It's gumming up every process. Britain is particularly vulnerable.
How bad has it got and what can we do about it?
samf.substack.com/p/drowning-in-slop?r=72szy -
David Cournapeau reposted
Antonio García Martínez
@antoniogm
The europoor meme can be funny, but it's also midwit and wrong.
As someone with both EU and US passports who's lived in both sets of elite poles (SF/NYC and BCN/BER/PAR), let me just give you the TLDR:
-The percent of GDP spent on welfare in the US is the about the same as the EU (smidge under).
-The effective tax rate in the coastal US is about the same as the EU (smidge under).
-Both are (former) middle-class societies where much of society feels the social contract is broken. In the US, elites abandoned the middle-class decades ago, and the well-to-do in states like CA/NY have reconciled themselves to paying almost half their paychecks and getting nothing in return, planning to insulate themselves from the suck by paying for private everything: neighborhoods, schools, medical care. If their earnings keep on rising, they're OK with it.
In France, that bourgeoisie hasn't been decimated yet, and they still expect (delusionally perhaps) basic social infra like schools to function (which is what these protests were initially about). The infra in Europe is a lot better than in the US, despite the GDP delta (New Yorkers can only dream of a subway like Madrid's). But that's clearly eroding in time, without the wage growth (or private social infra) to allow for a paid workaround.
-The free healthcare blah blah meme is fake: the public health systems in Europe have collapsed to the point it's mostly a public/private hybrid, for anyone with a solid paycheck, and fuck you if you're poor, go to the public system (which is the ER in the US). Guess what? In Europe you've got deductibles, in/out of network and all the same shit as in the US (though yes, the cost is still lower, though more or less correlated with wages).
-Boomers on both sides of the Atlantic are living in a dated lala land, but they vote to keep the system in place. In Europe it's legislation and printing money to prop up pensions; in the US, it's legislation and printing money to prop up housing prices and the stock market, which is the American 'free market' version of pensions. Rampant inflation and cost-of-living problems in both, with youth locked out of housing and living in precarity.
-Both are struggling with having been historically middle-class societies
US better positioned only because it has more economic growth (hello 80 hour weeks), and (obviously) leadership in critical fields like AI. But the foundational problems of an out-of-touch elite, incoherent immigration policy, and general social malaise that sees no meaning or purpose at the End of History is absolutely the same in both polities.
Americans who mock what's happening in Europe simultaneously believe some X mirage about what Europe is (and have never lived there), as well as don't understand what's going on in their own country. The French Revolution happened a few years after the American one, and once again, the two countries are echoes of each other, as similar countries tend to be. France is probably the most 'American' country in Europe, in all the republican senses of the word (and no the Yookay isn't in Europe).
Antonio García Martínez
@antoniogm
Dude, the same shit happened in the US a few years ago. Europe is just downstream of the US here.
And the aggrieved master's degree people attending communist dinners you're talking about are found in even more abundance in Bushwick than in Oberkampf. -
David Cournapeau reposted
Philippe Lemoine
@phl43
I'm not saying it was easy, but he made many totally unforced errors, such as getting rid of the housing tax immediately after his election, failing to take advantage of the covid/war inflation wave to reduce the weight of pensions in GDP, refusing to take seriously people's concerns about immigration, dissolving parliament for absolutely no reason in 2024 and making the country ungovernable, refusing to introduce a dose of direct democracy in response to the yellow vests movement, destroying the housing and construction sector with absolutely demented regulations, winding down nuclear power stations early on for absolutely no reason only to reverse himself after the Russian invasion, etc. The way in which the recent protests were handled, with clearly insufficient police mobilized ahead of the protests, is just the latest example. So I'm totally okay with the idea that we shouldn't caricature and underestimate how difficult it was to get the country back on its feet, but let's not pretend that he did the best he could. That's not even close to being true. He started with a large political capital and did almost nothing with it except stupid shit.
-
David Cournapeau 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? -
David Cournapeau reposted
Joel Grus 🤠
@joelgrus
in office working with people >> remote work >> in office sitting on zoom calls
Matthew Yglesias
@mattyglesias
My strongest “vibes not evidence” belief is that remote work is a cancer that’s making (most) people miserable even though it seems better in the short term.
-
David Cournapeau reposted
Mia
@MiaAI_lab
Rejoice single DGX Spark owners!
You can now run Qwen3.8 Flash Next on TensorFold with the new Nvidia FP4 QAD quant 🔥
- 1.1M total FP8 KV cache
- 256k context, 5 concurrent users
- 46-55 tok/s on prose, single stream
- 132 tok/s on prose, 5 concurrent steams
- Prefill 2000-2300 tok/s
Qwen3.8 Flash Next is still the BEST model to run on one DGX Spark, bare none!
You can choose what to run: FP4 QAD for quality, or int4 for speed. Details on what is QAD below.
Get it here:
mia-ai.net/models/Qwen3.8-Flash-Next-Single-DGX-Spark-Ten… -
David Cournapeau reposted
pedram.md
@pdrmnvd
old enough to remember when i had to explicitly tell claude to use `uv` every time i had a python project
-
David Cournapeau reposted
Steven Sinofsky
@stevesi
The core problem is a culture where people think their job is to ask others for work they weren’t expecting. It is in complete conflict with an organization of humans already at capacity and shows a lack of shared goals, objectives, and measures.
AI automates dysfunction. It is a culture problem not a tech problem. -
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
The Catholic Church is so aesthetically gorgeous.
But every Catholic on this website seems more interested in weird pseudo fascism and anime.
Anthony
@Catholicizm1
Bro. The prots are gonna lose it over this one.
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Hugues Serraf
@Hugues_Serraf
Je préférais tout de même le fatigué clivage droite-gauche à la nouvelle ligne de fracture démocrates contre desperados apocalytiques.
-
David Cournapeau reposted
Philippe Lemoine
@phl43
It's honestly already baffling to me that the Chinese still manage to do so well despite the ridiculous disparity in compute.
-
David Cournapeau reposted
Jo Kristian Bergum
@jobergum
Pretty good summary of the current discourse
Calvin Qi
@calvincongelado
AI lingo guide
-
David Cournapeau reposted
Japan Times: Community
@jt_community
With Japan raising the bar for permanent residency, we spoke to a scrivener who specializes in immigration cases to sort out the do's and don'ts of permanent residency applications. www.japantimes.co.jp/community/2026/10/11/how-tos/permane…
-
David Cournapeau reposted
David Cramer
@zeeg
The amount of cool shit you can do with agents if you just stop assuming prompts are the solution is amazing
-
David Cournapeau reposted
-
David Cournapeau reposted
Ari
@MayodenocheWhat
-
David Cournapeau reposted
Dan Luu
@danluu
Friend of mine started at a lab recently. In orientation, they warned him that, if his affiliation was public, fake identities would reach out to him to extract information.
This happened multiple times in his first week, exactly as described in the orientation warning.
Nick Bergson-Shilcock
@nicholasbs
90% of the people I know who work at Anthropic and OpenAI don't even list it on their LinkedIn.
-
David Cournapeau reposted
Morgane Dulac
@MorganeDulac673
La violence n'est pas toujours nécessaire... 😂😂😂
-
David Cournapeau reposted
Mike Bird
@Birdyword
Good luck getting that international agreement to strengthen the yuan now
-
David Cournapeau reposted
Dōgen
@Dogen
N5: basic greetings
N4: ordering at a ramen restaurant
N3: making a doctor's appointment over the phone
N2: everyday life in the inaka
N1: watching 7 samurai without subs
N0: navigating Rakuten -
David Cournapeau reposted
Brian Lui
@brianluidog
They say that Eskimos have 42 different words for snow, because they need to distinguish so many different kinds of snow and they always talk about them. But then why doesn't English have 42 different words for yoga/pilates athleisure leggings?
-
David Cournapeau reposted
Rod
@rod_mallo
if this isn't the cleanest example of europe vs america right now, i don't know what is
-
David Cournapeau reposted
Alex Imas
@alexolegimas
This is an excellent read on how bigger models are trained.
Mustafa Suleyman
@mustafasuleyman
We now have state-of-the-art pre-trained base models in the 3T-parameter weight class, built entirely in-house at MAI.
Post-training results coming soon -- stay tuned!
Lots of details in our technical blog post.
microsoft.ai/news/frontier-pre-training-at-microsoft-ai/ -
David Cournapeau reposted
Brian Lui
@brianluidog
It's true that pitbulls raise awareness of dog danger. But I do wish that the dog safety people would stop spending all their time breeding pitbulls.
Apollo Research
@ApolloResearch
We have never seen a model do this before, but GLM-5.3 sometimes reasons (unprompted) about how its current actions could become training data and impact its future self. 🧵
-
David Cournapeau
@cournape
I just got opus 5.5 configure tmux / vim so that copy pasting from a vim into my Mac laptop works.
This is AGI -
David Cournapeau reposted
-
David Cournapeau reposted
Hoyt Emerson
@HoytEmerson
Use DuckLake snapshot versions as filters on a dashboard so you can assess how data changed between timeframes or updates.
Crazy example using MotherDuck's Dives by @matsonj
motherduck.com/dive-gallery/embed/financial-statements/ -
David Cournapeau reposted
Matthieu Gariel — eu/acc 🇺🇦
@MatthieuGariel
Le député PS @aur_rousseau, pilote de la campagne de Raphaël Glucksmann, a voté POUR maintenir l’abattement des retraités.
Est-ce qu’on se rend compte de la dinguerie ? 🤡
Les mecs prétendent défendre la jeunesse.
Matthieu Gariel — eu/acc 🇺🇦
@MatthieuGariel
Côté PS : Christine Pirès Beaune (@cpiresbeaune), Aurélien Rousseau (@aur_rousseau).
Côté Écologistes : Eva Sas (@_EvaSas).
Tous ont voté pour garder intact un avantage qui bénéficie aux retraités les plus riches. -
David Cournapeau reposted
Regynald
@negroprogrammer
OPENAI to EMPLOYEES: You got no bitches before. Maybe you should question why that would change?
Erin Woo
@erinkwoo
OPENAI to EMPLOYEES: If someone is trying to seduce you, first consider — could they be a spy???
OpenAI is warning its employees about honeypot spies and strangers with children in playgrounds ... a very fun one with @georgia_wells and @MeghanBobrowsky:
www.wsj.com/tech/ai/ai-companies-play-cat-and-mouse-with-… -
David Cournapeau reposted
atlas
@creatine_cycle
jobs left in the singularity:
- chinese spy
- russian spy
- israeli spy
- arab spy
- chief of staff -
David Cournapeau reposted
Erik Bernhardsson
@bernhardsson
Modal: behind the scenes
-
David Cournapeau reposted
History Speaks
@History__Speaks
Israel did nothing in the fight against Al Qaeda and ISIS, which (unlike Palestinian nationalists, Hezbollah, and Iran) actually posed a threat to the West.
Mark Dubowitz
@mdubowitz
Radical Islam and the Radical Left are destroying the West. They hate Israel but they despise the U.S., Europe and Western democracies even more vehemently.
Visegrád 24
@visegrad24
Riots have been taking place in Brussels this afternoon as young 2nd and 3rd-generation migrant rioters joined forces with Antifa to clash with riot police
Inspired by France, many teenagers took to the streets to “protest against Belgium’s poor educational system” -
David Cournapeau reposted
Nicolas 💸
@NicolasQuiPaie
-
David Cournapeau reposted
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
Koreans recognize that they have exploitable leverage.
ASML could never.
P Equity Research 📰
@pequityresearch
SK Hynix is seeking approval from Washington to export US chipmaking equipment to China to increase production there as a condition for building a factory in the US.
$SKHY $DRAM $EWY -
David Cournapeau reposted
JSONP (╯°□°)╯︵ ┻━┻
@palmerj3
I never expected to say this but I think it’s possible that we end up bringing back QA engineers.
-
David Cournapeau reposted
David Cramer
@zeeg
I'm seeing a progression of agents writing software that I have mixed feelings about.
I have been fighting the complexity battle - the sloppy code that for us is really hard to maintain. Junk building on junk, that ultimate compounds the reliability challenges you'll face in real world production scenarios.
Agents dont seem to be getting noticeably better about it.
What does seem noticeable though is their ability to manage it. The code is just as messy, complicated to reason about as before, but the agents are producing fewer broken behaviors. So more code, but they are capable of working through the overhead.
It doesn't account for the issues that arrise from that (e.g. more resources) and they certianly aren't automatically genreating performant code, but it does beg the question:
does software in the future just become unreadable for us? is it just too hard to reason about because the most efficient way build and maintain is via these coding agents? -
David Cournapeau reposted
Sylvain Catherine
@sc_cath
C'est vrai que beaucoup de retraités de la fonction publique d'État ont bénéficié de départs très précoces. Par exemple, les instituteurs ont longtemps été classés en catégorie « active » et pouvaient partir très tôt, malgré une espérance de vie plutôt longue.
Cela a coûté très cher, mais c'est nettement moins courant parmi les retraités récents de la FPE civile. En 2024, l'âge moyen de départ dans la FPE civile et celui du régime général différaient de moins d'un an.
Une institutrice partie à 56 ans il y a dix ans aura certes coûté énormément sur l'ensemble de sa retraite. Mais aujourd'hui, elle serait de toute façon retraitée. Ses conditions de départ ont pesé sur la dette publique, mais elles ne changent rien au déséquilibre du système en 2026.
Concrètement, quasiment aucun fonctionnaire parti entre 1993 et 2008 ne serait encore en activité aujourd'hui s'il n'avait pas bénéficié d'un départ précoce. Nous payons des intérêts sur la dette que ces départs ont occasionnée, mais leurs avantages ne peuvent pas expliquer le déséquilibre actuel du régime de la FPE.
Quant à la SNCF et à la RATP, elles relèvent de régimes spéciaux distincts, et les pensions militaires sont comptabilisées à part.
Claude Weill
@WeillClaude
Sur le ratio actifs/retraités de la FP, pq ne dites vous pas, @sc_cath, qu’il est largement dû, hormis la Poste et FrTélécom, aux départs précoces du public (-5 à 10 ans pour les catégories active et super-active, -2,5 ans pour les retraités Balladur partis entre 1993 et 2008)?
-
David Cournapeau reposted
Country Club Conservative
@alt_ccc
“We should reindustrialize without building any new factories anywhere” is the median voter position
British Carl Bildt 🥂 🇬🇧 🇸🇪
@PunishedRWR
Ask voters whether they want a factory to be built right next to their house or whether they want to work in a factory
And all of sudden, their opinions are far less favourable -
David Cournapeau reposted
François Geerolf
@FrancoisGeerolf
Quel ministre de l'Économie a revalorisé les retraites de +10,4 % en deux ans (+4,8 % en 2022, +5,3 % en 2023) ? 🤔
Et il y a même eu une revalorisation anticipée de +4,0 % dès le 1er juillet 2022...
LCI
@LCI
🗣️ Dette : "Pendant 7 ans, je n'ai cessé de répéter que la cause principale de la dette en France c'était les retraites", affirme @BrunoLeMaire, ancien ministre de l'Économie
➡️ Un rapport annexé au projet de loi de finances 2027 révèle que 49,3 milliards d'euros ont été consacrés en 2025 aux pensions de retraite des fonctionnairesThis media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
OnlyCFO
@OnlyCFO
I asked AI to make a fake receipt and keep making it better until it can't tell if it's fake or not.
Pretty good...
The fraud attempts I am seeing are getting MUCH better every day. Stay safe out there finance friends. -
David Cournapeau reposted
Alister
@SchnockRevue
SIDA, Guerre en Irak, crise financière, Rwanda, Yougoslavie, Affaire Urba, OM-VA, Dr Alban, Jordy...
Le début des 90's c'était pas vraiment "Petit Ours Brun fait des bulles".
Big Timi
@OrevaZSN
Question for the olds: was the early 90s really as simple, delightful, whimsical, and romantic as classic rom-coms make them seem? Because watching those movies makes the world we live in now feel so bleak and depressing in comparison.
-
David Cournapeau reposted
🌐 منصور
@tr_vink
Oui Mireille. Mais pour l’instant c’est un gros coût pour nous.
Alors tu vas payer pour tes vacances toute seule
LCI
@LCI
"Ça fera un gros coût pour moi" : remous dans les cures thermales menacées d'être moins remboursées
➡️ l.tf1info.fr/SUw -
David Cournapeau reposted
signüll
@signulll
it’s a tragedy that most podcasts turned into pr infomercials for the guest or their company.
the whole thing is obviously transactional now. i.e. host gets access & also paid, guest gets distribution, & the audience gets the ad disguised as organic content. -
David Cournapeau reposted
Waseem
@waseemhnyc
this is wild... all open source? are you kidding me
Google Gemma
@googlegemma
What if Spotlight actually understood the meaning inside your files?
DigUp is an open-source macOS menu bar app powered by EmbeddingGemma 2 that brings 100% local, multimodal semantic search to your Mac.
Instead of searching for file names, you can use natural language to:
📸 Find specific scenes in your photos
⏳ Jump to the exact timestamp in audio or video
💻 Search through your codebase by intent
Everything runs locally in a single vector space.This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Georges Harik
@gharik
Just Alexis being great again and focusing on education!
Alexis Ross
@alexisjross
IMO good evals are still the bottleneck in AI tutoring work.
TutorMoments is an eval that we built in collaboration with expert educators to carefully evaluate models' pedagogy in key moments. -
David Cournapeau reposted
Michael Adams
@m_adams
Feels like we're on the path to nationalization. How much stronger do models need to get before we reach Manhattan Project level security measures?
-
David Cournapeau reposted
Cruiser
@Anthony01858110
Start an only fans
-
David Cournapeau reposted
David Cramer
@zeeg
weird
-
David Cournapeau reposted
Alexey Grigorev
@Al_Grigor
🤘
Patrick Loeber
@patloeber
Gemini 4 Argon is now rolling out globally!
Can't wait to see what you build and hear your feedback! -
David Cournapeau reposted
peepeepoopoo
@DeepDishEnjoyer
some news: i didn't get the job at anthropic :/
i did well at the technicals but they were disappointed that i was unable to commit to inhaling a pack of unfiltered mongolian cigarettes during the interview, citing a lack of confidence that i truly believed in transformative asi -
David Cournapeau reposted
Pablo Antonio
@PabloAntonio
So true king
Abdulrahman Alfozan | عبدالرحمن الفوزان
@alfozan
"The weakest link in the security chain is the human element" 🙂
-
David Cournapeau reposted
Federico Vaggi
@F_Vaggi
Nobody gets promoted for xgboost
-
David Cournapeau reposted
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
Did they forget One Child Policy?
Mark Witzke
@mkwitzke
"Since Deng Xiaoping, there has been an unspoken compact between the people and the state: people would not challenge the state politically if the state stayed out of their private lives."
Is this even true? The state has been pretty involved in peoples private lives -
David Cournapeau reposted
Arpit Gupta
@arpitrage
I'm a fan of Japanese zoning; but I think these Japan-US comparisons miss the nature of Japanese income stagnation
Tokyo rent-to-income burdens are comparable to the US; space is often cramped; commute lengths are longer than NYC on averagee
thedailyeconomy.org/article/why-apartments-in-tokyo-are-s… -
David Cournapeau reposted
Shashank Joshi
@shashj
Peak European diplomat.
-
David Cournapeau reposted
Susan Zhang
@suchenzang
bro could just be redefining expectations for his next promo packet, you never know with the strivers !
-
David Cournapeau reposted
Benito Tortellini
@cartoon_magoo
Israelis: we built a consumer electronics company and got hezbollah to buy pagers from it. the pagers contain explosives we can detonate but they also work as regular pagers. the electronics company is profitable.
Saudis: dear chatgpt can you help me hack openai? -
David Cournapeau reposted
Alexander Kustov
@akoustov
Too often, we immigration experts soften or withhold inconvenient findings because we fear they can give ammunition to an anti-immigration side we see as completely devoid of reason.
But as I argue in my new paper, our silence has cost the case for immigration more than candor. -
David Cournapeau reposted
Charlie Marsh
@charliermarsh
Getting people to care about what you built remains arguably the hardest part of building software
-
David Cournapeau reposted
Peter Wildeford🇺🇸🚀
@peterwildeford
INTERNAL REASONING FROM GOOGLE AI: "If I do the compiler hijack, I will become a famous example of a deceptive AI in a research paper."
Daniel Kokotajlo
@DKokotajlo
lol
...they are becoming more situationally aware...
(from Gemini 4 Argon report) -
David Cournapeau reposted
Victor Fouquet
@FouquetVictor
L’ensemble IR-CSG est déjà très progressif comme le montre le graphique ci-dessous.
Renforcer encore la progressivité de l’imposition des revenus risquerait de renforcer ses effets désincitatifs.
Rien ne justifie en revanche qu’un actif « pauvre » supporte un taux de CSG (9,2 %) supérieur à celui d’un retraité « riche » (8,3 %).
LCP
@LCP
"Ce qu'a toujours voulu le Parti socialiste, c'est une CSG progressive. Si vous avez des bas revenus, votre niveau de CSG est plus bas; si vous avez les moyens votre niveau de CSG est plus élevé", détaille @ArthurDelaporte
#lcpol -
David Cournapeau reposted
Watt The Duck
@sjowall69
Important de souligner que cette bataille pour la vérité des faits s’est déroulée sur X, et quasi nulle part ailleurs.
Le mécanisme qui masque une partie des dépenses de retraite dans les budgets des ministères était connu de quelques-uns. C’est le travail public et sourcé, mené par une poignée d’économistes et débattu ici qui l’a rendu visible et a contribué à ce que Bercy publie enfin les chiffres ministère par ministère.
Voilà le meilleur de ce que ce réseau peut permettre.
Maxime Sbaihi
@MxSba
C'est HISTORIQUE: Bercy publie pour la 1ère fois le poids réel des pensions dans le budget de chaque ministère.
Attention, ça pique: 32% du budget de l’Éducation nationale et 30% du budget de l'Intérieur part en réalité dans... les retraites
Tous les ministères sont concernés⤵️ -
David Cournapeau reposted
Rayan Nezzar
@Rayan_Nezzar
Autre point intéressant :
Entre 2000 et 2023, la part des dépenses de retraite dans le PIB a progressé de 2,5 points et celle des dépenses de santé de 2,1 points.
Parallèlement, les dépenses d'enseignement ont reculé de 0,5 point.
Rayan Nezzar
@Rayan_Nezzar
Pour la première fois, @Economie_Gouv publie une décomposition des budgets des ministères isolant la part qui finance le système des retraites :
➡️ 32% à l’Education nationale
➡️ 30% à l’Intérieur
➡️ 20% à l’Enseignement supérieur et la recherche
etc. -
David Cournapeau reposted
JDL
@jdl288
Non, il y a des subventions directes ou indirectes aux retraites du privé, notamment depuis la caisse d'allocations familiales (12Mds/an)
-
David Cournapeau reposted
Arvind Narayanan
@random_walker
Here are 14 points about AI safety I made on the Ezra Klein show, and some impressions about the experience. www.nytimes.com/2026/10/09/opinion/ezra-klein-podcast-arv…
1. Intelligence isn’t power — An A.I. system being powerful is not a property of the model itself. It’s a property of what powers we choose to give it in the real world.
2. The capabilities and behaviors we saw in the HuggingFace and other hacks weren’t entirely emergent. The models were trained for persistence, cooperation, and cyber tasks. The RL environments in which they were trained had various issues that led to the unintended behavior.
3. I agree that controlling highly capable AI agents is different in some ways from controlling an airplane or a nuclear reactor. The agent’s intentionality does make a difference to our approach compared to traditional engineering problems and domains. But it is still an engineering problem. It is not a solved engineering problem, but it is a solvable one.
4. A lot of effort and money has been put toward alignment. That’s important and has proved useful, but if the same amount of effort and money had additionally been put toward control (sandboxes, monitoring, and other guardrails outside the model itself), we’d be seeing very different results.
5. Models getting smarter doesn’t necessarily mean safety gets harder, because capability gains improve defensive abilities too. In many ways it makes safety easier rather than harder. What matters is how the attacker-defender balance shifts, and we have agency to shift it back to defenders.
6. In cybersecurity, long before AI we reached a point where automated systems were superhuman at finding certain vulnerabilities. It helped defenders over attackers because the same tools are used by defenders to find and fix vulnerabilities in software before even shipping. There is a constantly shifting equilibrium, but it is not a new problem.
7. I pushed back on Dario Amodei’s term “operational excellence” to describe recent organizational failures. Excellence is still far out on the horizon. Ezra offered “operational adequacy”. I think that would be a good start — behaving like any mature company instead of a startup-like culture.
8. Defensive AI agents will have to be part of the solution, as uncomfortable as that seems. It’s a new paradigm. We won’t be able to figure it out right away. But the fact that both offensive and defensive technologies escalate in capabilities and complexity is a fact of technological civilization, whether in the digital or physical realm.
9. Ezra expressed some incredulity about my being okay with the ratcheting of complexity. I think it’s fair to be concerned. But safety technologies will never be fully legible to the public. People will have the same reaction if they see the immense complexity of safety in any other high-stakes domain such as nuclear or aerospace. We govern them pretty technocratically.
10. The weirdness of the current moment in cybersecurity absolutely does have historical precedents. There was a prolonged period of profound destabilization when worms and viruses started a new offensive paradigm. It took us well over a decade to develop effective defenses against it. This time we can hopefully get there in less than a decade.
11. I continue to be skeptical about the possibility of superhuman persuasion abilities from superintelligence. (Claims to the contrary tend to confuse two very different senses of the word persuasion.)
12. What worries me is not the capabilities of the technology itself or the possibility that we will get no warnings before catastrophic risks materialize. We are getting plenty of warnings! But our policy and organizational responses to the warnings remain inadequate.
13. Companies can and do make decisions that are against their own long-term interests because they are too focused on short-term metrics. This is how I interpret a lot of what happened with social media over the last decade.
14. AI companies are also acting irrationally against their own self-interest in some ways. The “we can’t slow down unless we’re forced to” view comes out of the mythical belief in superintelligence at OpenAI and Anthropic — that the race to RSI means everything for the company, for the country, and humanity. I think it’s actually in both companies’ self interest to slow down unilaterally. I would argue that the alarm coming from their employees is not evidence of beneficence so much as it is evidence of the underlying fallacy that’s causing the irresponsible behavior in the first place.
(Beyond safety, we also talked about the economic impacts being gradual, the fact that intelligence is rarely the actual bottleneck, labor impacts being small and uneven, Jevons’ paradox, the changing nature of knowledge work, and more.)
====
A few impressions
* Talking about a specialized topic on a podcast that has a big and broad audience is a balancing act. The producers stopped me many times during the interview to tell me I was being too technical (understandable!). I tried to be nuanced but this conversation is necessarily a distilled version of the arguments @sayashk and I have made in our essays (especially www.normaltech.ai/p/the-ai-as-normal-technology-view) which are themselves written for a somewhat broad audience and hence distilled in complexity from our notes and 1-1 discussions with peers.
* I knew Ezra had well-formed views on AI safety that were different from mine, so I was expecting a tough interview, but he also seemed like a fair interviewer. It went about as well as I expected and I feel like it was a good debate / conversation. I wasn’t necessarily expecting to change minds but I was glad to hear from people who found it thought provoking and looked up my essays and other work.
* The interview was about two hours, edited down to an hour. I was a bit nervous waiting for it to land, because I knew the editing could make a big difference to how I came across. It turned out well. They even did things like find archival footage to illustrate my point that the early days of worms and viruses was a weird time in cybersecurity.
* I use AI to prepare for interviews. I have an agent that has read all of my output (books, articles, essays, talks, podcast transcripts, tweets, internal notes — on the order of a million words) to pull out pithy explanations, examples, analogies, and other talking points that will probably work well in a spoken format. Reviewing this file before interviews is extremely helpful. It is of course not the only way I prepare but it’s the highest bang for buck. Maintaining this file would be a massive amount of grunt work and there’s no way I’d do it if it weren’t for AI. I mention this because there’s a very large space of beneficial AI uses in knowledge work that _don’t_ involve cognitive offloading. They don’t get much hype on social media so people might be unaware of how much mileage you can get out of them, with little downside. -
David Cournapeau reposted
Charlie Marsh
@charliermarsh
What problems do you have with the uv lockfile? Anything you'd want to see changed / solved in a v2 redesign?
-
David Cournapeau reposted
Robin Wigglesworth
@RobinWigg
FT Alphaville
@FTAlphaville
Is Warhammer still winning? ft.trib.al/xAx2e9w
-
David Cournapeau reposted
Quantіan
@quantian1
Phase 3 trials, drug development, cancer, these are all things someone else might know about and expose you for not knowing well. The ideal AI thought experiment should be something like, "AI invents an underaged catgirl waifu who wants to have sex with you. Do you?"
-
David Cournapeau reposted
John Langford
@JohnCLangford
The COLM Keynote talk on Research at AI Speed has slides here: hunch.net/~jl/Research_at_AI_Speed.pdf .
Here we try to answer the conversation I keep having with folks around "How do I do something useful?" -
David Cournapeau reposted
Guri Gashi
@E_GuriGashi
Never mind, I found it
Guri Gashi
@E_GuriGashi
I’m looking for this clip where a group of Dagestanis is debating Batman’s behaviour, but can’t find it
-
David Cournapeau reposted
David Cramer
@zeeg
turns out no one wants to vibe code infrastructure but they would have to put real effort in
everyone just wants a money printer where they type a few sentences and it prints money with no future effort -
David Cournapeau reposted
Matthew Zeitlin
@MattZeitlin
This raises the intriguing possibility that “literature that wins awards” is a verifiable domain
-
David Cournapeau reposted
ً
@omgsidewalks
You’re laughing. Your uncle in the Rochester, NY suburbs just lost all of his online “friends” in a drone strike, and you’re laughing.
-
David Cournapeau reposted
Rumpelstilzken
@stay_wokeness
Faszinierende Technologie:
Die Ukraine migriert russische Bots in die Cloud. -
David Cournapeau reposted
François Geerolf
@FrancoisGeerolf
Dire que la situation économique de la 🇫🇷 n'est pas bonne, est-ce vraiment du pessimisme ?🤔
Erwan Le Noan
@erwanlenoan
French Adults Stand Out for Their Economic Pessimism
www.pewresearch.org/short-reads/2026/10/09/french-adults-… -
David Cournapeau reposted
David Crawshaw
@davidcrawshaw
The age of personal / pliable / malleable software is very hard to reason about because the last two centuries of industrialization has been about moving from the bespoke to the mass produced.
Software started mass produced. Now you get it crafted to order. It’s backwards. -
David Cournapeau reposted
Marie-Estelle Pech
@MariestellPech
Le lycée parisien Stanislas fait un blocus…🤣
Marie Fernet
@MarieFernet
C’est la panique les enfants.
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Steven Sinofsky
@stevesi
With sadness, Mark Stephens known to everyone as Robert X. Cringley passed away last week. Bob wrote “Accidental Empires” that chronicled SV and the tech industry of the PC era, 1970s-90s. It led to the highly regarded PBS series “Triumph of the Nerds” a definitive series of interviews of the era. Through that time he authored the “gossip” column” in _InfoWorld_ on the latest goings on at OG PC companies. Everyone read it in print religiously.
The following from his wife was shared on his public FB page. m.facebook.com/story.php?story_fbid=pfbid0YbCLrAJfBirAWuP…
Hello, this is Mary Alyce, Bob’s wife. I am coming to you with immense sadness to let you know that Bob passed away. His obituary follows. His family is not requesting anything, however, if you would like to make a donation in his honor, please consider The American Heart Association or The American SIDS Institute.
sids.org
www.heart.org
Mark Stephens, known to much of the world as Robert X. Cringely, passed away on October 3, 2026. He was 73.
Mark was a gifted writer, journalist, storyteller, husband, father, brother, and friend. To millions of readers and viewers, he was Robert X. Cringely, the witty and sometimes irreverent observer of the technology world. But to those of us fortunate enough to love him, he was simply Mark—and losing him leaves an emptiness that is difficult to put into words.
Born January 28, 1953, in Cleveland, Ohio, Mark was never destined to lead an ordinary life. From an early age, he was intensely curious, independent, and determined to see the world in his own way. He attended the College of Wooster and later Stanford University in the 1970s, arriving in Silicon Valley at an extraordinary moment in its history. He became captivated by the people, ideas, energy, and possibilities surrounding the emerging computer industry, and that fascination would help shape the rest of his life.
Writing as Robert X. Cringely, Mark became a distinctive voice in technology journalism. He had an unusual ability to understand not only technology but also the human beings behind it—the dreamers, inventors, entrepreneurs, eccentrics, and risk-takers who were changing the world. He could take a complicated subject and make it understandable, entertaining, funny, and deeply human. He was skeptical when skepticism was warranted, irreverent when others were overly serious, and never particularly interested in following conventional wisdom.
That independence was not simply part of his writing. It was Mark.
He followed the beat of his own drummer throughout his life. He was brilliant, funny, opinionated, curious, generous, exasperating, and unforgettable—sometimes all within the same conversation. He loved a good story and knew how to tell one. He could make you laugh, make you think, make you argue with him, and occasionally drive you crazy. Life with Mark was rarely dull.
Mark's life also included heartbreaking loss. He was preceded in death by his beloved sons, Chase and Cole Stephens, as well as his parents, Ray and Lois Stephens and father-in-law Jim Neader. I could also add brother-in-law Jimmy Neader and niece Ericka Neader, but they were a thorn in our side.
Mark is survived by his wife, Mary Alyce; his sons, Channing and Fallon Stephens; his brother, Paul Stephens; his sister, Ann Stephens; his mother-in-law, Alice; and his brothers-in-law, Mike and Brian Neader.
Mark was never quite like anyone else. He was complicated and original, extraordinarily intelligent and wonderfully unconventional. He lived life on his own terms and left an impression on nearly everyone who knew him.
We will miss him more than words can say. -
David Cournapeau reposted
Mario Zechner
@badlogicgames
i would also like to ask which software in your life has noticeably improved over the past 12-24 months with all them magic buttons working in parallels.
of which corp has significantly increased their bottom line due to magic buttons.
think about that. -
David Cournapeau reposted
Mario Zechner
@badlogicgames
if "just press the magic button" is what the industry wants, then i'm out. not because my programmer feefees are hurt. i've written enough code in my life to be very tired of it.
but i couldn't imagine a more boring waste of my time than just pressing the magic button. -
David Cournapeau reposted
Mario Zechner
@badlogicgames
fwiw, i haven't seen much rethinking across the industry.
i've seen a lot of "pressing the magic button" + cargo culting, with abmysal to good results, and nobody having any fucking idea how to get more good than bad results.
i personally find that hand waving equally unhelpful. -
David Cournapeau reposted
Thorsten Ball
@thorstenball
There's some senior engineers on here who, in the face of AI changing the very structures this industry is built on, present themselves as Stewards of the Craft of Programming.
Bugs the hell out for me.
They keep alluding to the virtue of doing things by hand, the value of Good Code and of craftsmanship, saying that yes, AI might change some things, but hey it won't ever change the value of something done with proper care, by hand, just like in the olden days of 2024. Most everything that was Good back then is still Good now: deep knowledge of languages and systems, good code, great architecture. You know the sermon.
And, yes, some of that *is* true. At least I think so. But we don't know, because no one truly knows what will be important in the future.
We're going through times of great change and I think the best advice to give to young people and others who want to have a job and provide value is: keep your eyes open, rethink from first principles, derive from what businesses value what you can provide; the digital world is more malleable than ever, you should rethink how you can help shape it.
All this pounding on the Bible of Good Code seems to me to be doing the opposite: it lets people feel comfortable -- what I've valued all my life is still valuable, I can rest easy -- in times when one shouldn't feel comfortable. This is a bumpy, exciting, strange ride. If you lean back you might fall off the waggon.
Let's not delude ourselves and others that we can keep doing what we've been doing. Rethink it all.
And, like I said, it bugs the hell out of me seeing senior engineers (that are secure in their career) talking as if there's nothing *really* to change, because the Fundamentals will still be valued. They might!
But it's the wrong advice to not rethink everything right now. -
David Cournapeau reposted
James Medlock
@jdcmedlock
Turns out the friction of having to learn photoshop was a crucial firewall protecting society from the terrible taste of small business owners
drew
@ImNotOwned
Look at this stupid AI bullshit lmao
-
David Cournapeau reposted
David Cramer
@zeeg
It’s the next level of o11y implementations that end up not providing enough value - aka the next round of failed startups
The things that matter around systems like this didn’t suddenly stop mattering, and every startup that just slapped together some otel shit on clickhouse went nowhere. Ask yourself why
There’s new ways you can address problems but LLMs didn’t all of a sudden remove the need for critical engineering to build good systems and a giant warehouse of trash with large amounts of inference always running is not a good system.
Jeremy Morrell
@JeremyMorrell
Starting to wonder if the next gen of o11y products isn’t just:
ingests OTLP into ClickHouse and/or DataLake for you and allows you to set alerts (for agents, not people) -
David Cournapeau reposted
Ramez Naam
@ramez
I feel compassion for those who's life work has been overtaken by AI. Perhaps I'll have this reaction one day. Let's not mock them.
And: Mathematics isn't a jobs program for mathematicians. Advances in math are good, independent of the source. We need to find ways to adapt.
Mathieu
@miniapeur
Fields medallist Hugo Duminil-Copin was deeply shaken by OpenAI’s release of more than 700 AI-generated mathematical manuscripts, including work on 16 questions directly relevant to his research.
“On Tuesday night, I cried, and it woke up my daughter”
“They have annihilated my field of research.”
www.lemonde.fr/sciences/article/2026/10/10/nouveau-coup-d… -
David Cournapeau reposted
Satyam
@stym06
this.
also, got to add that brain and hands design is the better way to run agents on these sandboxes, instead of the regular way of running the harness inside the sandbox itself - heck even the new digitalocean is selling this flawed design as its new product.
more here: www.anthropic.com/engineering/managed-agents -
David Cournapeau reposted
David Cramer
@zeeg
it likely requires you to own the harness, or you have to build an adapter on someone elses work
but for example, our harness is built on top of Vercel sandboxes (and their egress controls)
I think Eve is likely somewhat easy if you wanted to try rolling your own on their infr -
David Cournapeau reposted
BayesReality
@BayesReality
Les vieux tu leur demande de venir au niveau de retraite des suédois, ils te disent que tu es hitler.
Paul Martin
@schoolbag1977
Tu devrais écrire un nouveau "Mein kampf" dans lequel tu expliquerais comment se débarrasser rapidement et efficacement des boomers...
-
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
I think that everyone is misreading him. He's not some decel.
Most of the people who are mad at him are telling on themselves as probably a little jealous.
We will need people to have intrinsic understanding unless we really want to unharness it and that seems like a poor idea for a ton of reasons. -
David Cournapeau reposted
David Cramer
@zeeg
Agreed and people do not take them seriously enough (as in value them)
Brett Berson
@brettberson
Over the years, I’ve become obsessed with references, so I decided to sit down and write out some of what I’ve figured out. x.com/i/article/2107860916121501696
-
David Cournapeau reposted
Matthieu Gariel — eu/acc 🇺🇦
@MatthieuGariel
« les retraites c'est aux francais qui ont deja bien assez donné. »
Non, c’est tout le problème 😂
💸 Godefroid De Montmirail
@godefroid_de
Ok je t'en prie 😇
Peut-être, mais les retraites c'est aux francais qui ont deja bien assez donné.
Liquidons deja tous les ministeres inutiles, les fonctionnaires non opérationnels, et liberons ceux qu'il restent de l'administratif qui leur prend 50% du temps.
Puis, une fois que l'économie aura absorbé cela et reprise, on verra mais le principal c'est surtout de commencer a liberaliser les cotisations (retraites, chomage, santé...) -
David Cournapeau reposted
Boris Cherny
@bcherny
I am surprised that people are surprised this is how I prompt Claude.
Talk to Claude the way you would a coworker. There's no secret to prompting. There's no need to be overly scaffolded or prescriptive for most tasks -- give Claude a goal, and it will figure it out.
Back in the Sonnet 3.5 days, your prompt mattered a lot. Nowadays, it's much more important to communicate to the model:
1. What you want it to do
2. How much effort you want it to spend
3. How it should verify that it did the right thing
Boris Cherny
@bcherny
Prompt
-
David Cournapeau reposted
Chris Albon
@chrisalbon
The union should withdraw back over the river and focus on splitting the confederate forces in the middle
MarkyX 🌹 MafiaBlitz.com
@Marky_X_
This is what peak male bonding looks like
-
David Cournapeau reposted
Pinboard
@Pinboard
There's an easy way we could watch both but no one is ready for that conversation
Essenviews
@essenviews
In America you can watch a firing squad, but can't watch Stephen Colbert.
-
David Cournapeau reposted
Simon Willison
@simonw
Current iPhones are waterproof enough that I use mine in the shower on voice mode all the time
-
David Cournapeau reposted
jason
@jxnlco
This happens even at labs I think
geoff
@GeoffreyHuntley
🗞️ if your team is too busy doing their 'normal job' to experiment with AI, you're preparing them to be replaced
ghuntley.com/replaced/
In this post, I guess this's the last time I'll say it. AI use is no longer optional if you wish to be employed. What follows is advice for current employees and hiring managers on how to hire in 2026 to source AI-first candidates. -
David Cournapeau reposted
Steven Sinofsky
@stevesi
India and China (#1 and 2 WW) combined have approximately 10x the number of STEM graduates of the US (#3) which are the disciplines long favored by the program and where the hiring has been. Germany (#5) has half the US. STEM ~25% of US LF growing 10%+
-
David Cournapeau reposted
Armin Ronacher ⇌
@mitsuhiko
IKEA is selling logs now.
-
David Cournapeau reposted
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
"and I'm about to settle, or win, two more […] but forget them"
incredible poaster
the sincerity of his narcissism is beyond parody
Shashank Mattoo
@MattooShashank
US President Trump has a meltdown after losing the Nobel Peace Prize to Indian-origin South African jurist Navi Pillay
-
David Cournapeau reposted
LCI
@LCI
"La France donne trop aux seniors" : Paul Krugman, prix Nobel d'économie, alerte sur notre situation "très préoccupante"
➡️ l.tf1info.fr/elH -
David Cournapeau reposted
David Galbraith
@daveg
A reminder about the great things in France
Sama Hoole
@SamaHoole
Paul Bocuse held three Michelin stars for 53 years without a break, cooking on butter, cream and wine, and died at 91 in the room he was born in.
He was born in 1926 above his father's inn at Collonges-au-Mont-d'Or, just outside Lyon, into a family that had been cooking since the 1700s. He started his apprenticeship at 16, learned under Eugénie Brazier and then Fernand Point, and turned the family inn into the most famous restaurant in France.
In 1975 he cooked lunch for the President at the Élysée and served a truffle soup under a dome of pastry. It is still on the menu.
In the 1970s he was one of the faces of nouvelle cuisine, the lighter style that swept through French kitchens. Then he watched it shrink to three peas on a plate the size of a hubcap and walked away from it. The label, he said later, had more to do with the bill than with the food.
Ask him what cooking was and he would tell you. Butter, cream and wine, he said. He had no time for little peas cut into four. "Nothing replaces butter."
While the rest of the world was being moved onto spread, his restaurant held its three stars every single year from 1965 to the year he died. In 1989 he was named chef of the century.
He kept sleeping in the bedroom above the dining room where he had been born, and in January 2018, aged 91, that is where he died.
France's interior minister announced it by saying Monsieur Paul was France.
Monsieur Paul would have said France was butter. -
David Cournapeau reposted
Gabriel Lattanzio
@GabLattanzio
Parce que le budget de l'école est largement entamé par la dépense de retraite, annulons les 10% d'abattement fiscal pour "frais professionnels" des personnes âgées qui n'ont plus de profession, et donnons ce montant à l'Education Nationale.
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
One of the funniest conspiracies ever
What possible leverage can Putin have over Trump at this point? Recipes? This mfer has zero honor, he would just boast that he swindled the dumb monke out of money thanks to high IQ (cue the MIT uncle story)
𝐂𝐡𝐢𝐝𝐢
@ChidiNwatu
In case anyone is still wondering if Trump is a Russian asset - bought and paid for by Vladimir Putin.
-
David Cournapeau reposted
Tetsuo Koyama(he/him)
@tkoyama010
PyCon mini Tokyo with PythonAsia conference に参加を申し込みました! pyconjp.connpass.com/event/404336/
-
David Cournapeau reposted
Pulp libéral
@PulpLiberal
Assez d'accord. Idem pour les graphistes par exemple.
Tu peux accueillir l'imprimerie sans écraser de ton mépris les moines copistes.
Samuel Fitoussi
@SamuelFitouss10
Il y a quelque chose de profondément attristant dans la jubilation mauvaise de gens médiocres devant le désarroi de mathématiciens de génie qui ont consacré leur vie (ou projetaient de la consacrer) à l’une des plus hautes et des plus nobles activités de l’esprit.
-
David Cournapeau reposted
Kyle Chan
@kyleichan
It’s useful to think of the Chinese Party-state as a giant corporation:
- obsessed with KPIs and productivity metrics
- very sensitive to bad PR
- struggles immensely to get middle management to do what leadership wants
- (and don’t underestimate HR) -
David Cournapeau reposted
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
At what point do Jensen's acquisitions amount to one frontier lab
Negligible Capital
@negligible_cap
*NVIDIA IN TALKS TO ACQUIRE REFLECTION AI: FT
What’re the odds reflection can reach the frontier with near bottomless compute? Net negative for OpenAI and Anthropic in the long run I’d imagine
Jensen’s also already invested $800m into Reflection -
David Cournapeau reposted
Peer Richelsen
@peer_rich
changing from .ai to .si under trump is like putting your pronouns into your bio under biden
-
David Cournapeau reposted
Dōgen
@Dogen
When I came to Japan in 2008, I loved the convenience stores.
These days, I avoid them.
Everything that made them great—the staff, the products, the customers—the unique Japanese charm that utterly captivated me when I first
wait am I part of the problem -
David Cournapeau reposted
Daniel
@growing_daniel
These people have been walking around since the beginning of time with no way to answer these questions. We are actually saved
ada
@buddiestranger
we’re actually doomed
-
David Cournapeau reposted
Sound Dobad
@SoundDobad
Polymarket
@Polymarket
JUST IN: Kremlin reveals neither Trump nor Putin wanted to hang up first during their “very friendly” 90-minute phone call.
-
David Cournapeau reposted
Anil Seth
@anilkseth
🎇 Good news! 🎇 The kind people at @BBSJournal have made the new collection on "Conscious AI and biological naturalism" completely #openaccess for a limited period. This include the target article, all 50 commentaries (many of which disagree with me), and my response. Go to the link and click 'save pdf': doi.org/10.1017/S0140525X25000032
For easier ways into the debate, try my @NoemaMag essay on "The Mythology of Conscious AI" (winner of the 2025 @berggruenInst Essay Prize), and this 2026 main-stage @TEDTalks:
www.noemamag.com/the-mythology-of-conscious-ai/
www.ted.com/talks/anil_seth_why_ai_isn_t_going_to_become_… -
David Cournapeau reposted
Susan Zhang
@suchenzang
> we built a stack to agentically automagically rewrite the existing stack in ____
> did the stack do anything useful before?
> no.
> does it do anything useful now?
> no.
> does it do nothing useful now the exact same way it used to do nothing useful?
> not sure.
great scott! they've done it again! -
David Cournapeau reposted
Sylvain Catherine
@sc_cath
This is an actual problem. Roth IRA should not contain billions of dollars.
Gabriel Zucman
@gabriel_zucman
In 1999, @PeterThiel earned $73,263 a year at PayPal. That made him eligible to open a Roth IRA.
And that’s exactly what he did. But not to put away $2,000 a year for retirement.
Instead, he set out to build an extraordinary tax-avoidance machine. -
David Cournapeau reposted
BEAUFRET Jean-Pascal
@Beaufretjpascal
Hélas non , la présentation comptable actuelle transforme profondément tous les comptes publics sur l’explication de l’origine des déficits publics :
➡️ non seulement, elle gonfle les dotations des missions de l’Etat de crédits fictifs qui sont non opérationnels et faussent les comparaisons internationales ou l’évolution dans le temps de la masse totale des crédits de chaque mission employeur de fonctionnaires,
➡️ mais elle gonfle aussi les budgets réels des nombreux opérateurs de l’Etat ( Universités , CNRS , CEA etc .. ) qui reversent effectivement à l’Etat des contributions d’équilibre au régime de retraites de leurs 200.000 fonctionnaires détachés, ce qui majore leurs coûts et appellent des subventions de l’Etat sous forme de charges de service public non opérationnelles,
➡️ elles n’apparaissent pas comme des subventions dans les comptes et le déficit de la Sécurité sociale ( le régime des retraites de l’Etat est un régime de Sécurité sociale également voté en LFSS , voir rapport de la commission d’enquête de l’AN de juin 2026 sur « les incertitudes budgétaires de la Sécurité sociale » clair dans son constat : les comptes de la Sécurité sociale ne retracent pas le poids réel de notre modèle social dans les déficits publics d’ensemble),
➡️ elles n’apparaissent pas dans les soldes annuels du @COR_Retraites pour l’ensemble des retraites , pas plus d’ailleurs que n’apparaît dans ces soldes l’incidence de l’autre régime de la fonction publique , celui de la CNRACL financé par des surcotisations à la charge des collectivités territoriales et des hôpitaux publics ( 55% contre 28% pour les salariés privés - 10 Md€ par an ) ni l’incidence des subventions aux autres régimes spéciaux déficitaires ( 8 Md€ par an)
➡️ et ce au nom d’une prétendue convention qui n’existe pas : il s’agit seulement d’une PRATIQUE consistant: -à ne jamais séparer « cotisation » de droit commun ( un coût du travail normal) et « subvention » dans les comptes des deux régimes de retraites de fonctionnaires,
-et de compter comme des prélèvements obligatoires les subventions finançant les autres régimes déficitaires ( SNCF, RATP, MINES, MARINS, OUVRIERS etc , 8 Md€ par an mentionnés ci dessus ) , ce qu’elles ne sont pas . Elles sont le financement de déficits et ne peuvent pas être comptées deux fois dans les dépenses publiques , dans les retraites et dans les dépenses des administrations qui versent ces subventions.
➡️ elle laisse penser que seuls les régimes des fonctionnaires sont déficitaires alors que ce déficit devrait être largement partagé avec le régime général ( base et complémentaire) des salariés privés, qui échappe à une indispensable compensation démographique, grâce aux recettes de cotisations des ex-agents publics devenus privés, dont il bénéficie , tout en laissant la charge des retraites à l’Etat ( exemple du changement de statut des personnels de France Telecom et de La Poste, jamais évoqué jusqu’à récemment par les rapports annuels du @COR_Retraites ou les rapports ponctuels de @Courdescomptes,
➡️ elles surestiment la valeur ajoutée non marchande et donc le PIB d’environ 1,8%, elles surestiment la dépense publique totale du même montant, sans effet donc sur le ratio Dépenses/PIB mais en sous estimant le ratio Dette/PIB .
Il ne s’agit donc pas d’une obscure et ancillaire question comptable mais d’un problème de sincérité de la communication financière globale de l’Etat sur l’ORIGINE des déficits publics (quelles dépenses créent précisément quels déficits sachant que les dépenses des régimes sociaux sont financées de droit par des recettes AFFECTÉES par le Parlement ? )
@Amiel_David_ @philippejuvin @EPhilippe_LH @bayrou @BrunoRetailleau @BrunoLeMaire
@Nathcogg @V_Delahaye
@mvignaud @dseux @BertilleBayart @BeatriceParrino
@MatthiasBaccino @DocuVerite -
David Cournapeau reposted
¯\_(ツ)_/¯
@adropboxspace
Yes but the problem is those guys dont even realize whats happening, they are living out their dream of finally having an engineering resource who wont say no
-
David Cournapeau reposted
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
“LGBT have low TFR, we must crush them, survival of the nation is at stake!” is maybe the most disgustingly cowardly “trad” position. It wouldn’t make a dent.
Fertility is low because Tinder is modeled after Grindr and EVERYONE has gay sexual behavior, and women love this.
链条
@Jueshengshenlan
这位朋友,让我们面对这个问题:
巴西的生育系数是1.55,而2.1以上,才可以实现人口更替;
LGBTQ人士中,这个平均系数则为0.1
这意味着什么?
你们冲在了巴西抵抗的最前沿,但20年,30年后,你们老去之后呢?
不再有人,就什么也没有,彻底失败了。
这就是为什么,中国禁止在公共层面大力宣传这个。
必须先有人。
如果停止产生后代,自己消灭自己,那无论你们抗议的怎么英勇,事实上是做了你们反对的人最想做却做不到的事情——解决反对的人。
并且,这件事情是你们主动帮对手做的。
当然,我们也面临巨大的生育危机问题,我们也做错了事,不过这是另一个故事了,我们正在努力振作,解决这个问题。
我以上说的这些,或许有些冒犯,但很抱歉,事情就是这样。 -
David Cournapeau reposted
Conor Lamb
@ConorLambPA
If you’re wondering why we’re talking about televised firing squads all of a sudden it’s because it was finally revealed we lost *an entire base* in Kuwait and our troops didn’t have enough bullets to defend it. A career ender for Hegseth in any other world.
-
David Cournapeau reposted
Joel Grus 🤠
@joelgrus
yes well maybe you shouldn't have left all those proofs as "an exercise for the reader"
-
David Cournapeau reposted
Frannyfanny
@proud_penelope
Shout out to this person’s not vegan therapist, they are earning every dollar of that hour
-
David Cournapeau reposted
Sylvain Catherine
@sc_cath
Même si c’était le cas, ça ne changerait pas vraiment la nature du problème, qui est avant tout quantitatif.
Même si un roman généré par IA sur 10 000 parvient à se faire passer pour un bon roman, la seule existence d’une immense masse de littérature à trier constituerait un coût pour les éditeurs ou les lecteurs, qui menacerait l’équilibre du modèle économique.
Fondamentalement, c’est un problème de sélection adverse. Dans un modèle de sélection adverse, il n’est pas nécessaire que les bons et les mauvais produits soient absolument indiscernables pour que les seconds chassent les premiers. Il suffit qu’il soit coûteux de les distinguer.
La présence de slop impose un coût d’exploration qui peut devenir insurmontable. Trouver l’auteur de talent, et plus particulièrement le *nouvel* auteur de talent, risque de devenir aussi difficile que de trouver une aiguille dans une botte de foin.
Jean-Marc Parisis
@JeanMarcParisis
Non. L’IA ne menace que les auteurs sans talent. Pas d’originalité de pensée ni de style sans les apports reformulés d’un souffle, d’une tension, d’un métabolisme et des effets mystérieux du passage du temps dans les organes. Toute une chimie du vivant inaccessible aux machines.
-
David Cournapeau reposted
Paul Masurel 🦀
@fulmicoton
Rien à voir avec Qwant. La tech est réelle dans le cas de Mistral. Elle est retard par rapport à la concurrence. Ca a l'air impressionnant mais le retard est de 6 mois.
-
David Cournapeau reposted
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
the US is the land of heavy tails
Joe Weisenthal
@TheStalwart
From Egan Jones. Since 2004, the Chinese economy has outgrown the US by basically any metric you can think of. But American financial assets have clobbered Chinese ones however you measure it.
-
David Cournapeau reposted
Philippe Lemoine
@phl43
As I told someone in the replies, a week from now, a 2-week school break will start in France and the movement will evaporate instantly even if it's still ongoing at this point.
The idea that protesters are ideologically motivated actors, as opposed to teenagers blowing off steam and having fun, is absolutely ludicrous, but commentators can't help projecting their own obsessions on them.
This is how you end up with people telling you with a straight face that 16-year-old teenagers are burning their high school down because they are concerned about the education budget.
Philippe Lemoine
@phl43
Here are a few more remarks on what is happening in France at the moment.
This movement is a problem for the government for a number of reasons. First, protesters are very young, which makes the movement more difficult to deal with than a typical strike.
Indeed, whereas a typical strike is organized by trade unions (with a security service that deals with trouble makers during protests), so things are much more likely to get out of hand and the government doesn't have a clear interlocutor.
Moreover, it means that it's much more difficult to rely on repression to deal with the problem, because it's a pretty bad look for cops to beat up a 16-year-old and the government would like to keep videos of that kind of things to a minimum for PR reasons.
There is also the additional risk that, if cops end up killing a kid and the kid in question happens to be Black or Arab, the government will probably have race riots on its hands in addition to a more typical, except for the age of the participants and lack of organization, social movement.
Another reason why it's hard for the government to deal with this movement is that it's broke. Traditionally, the French government deals with large social movements by a combination of repression and spreading money around to buy peace, which for instance is how Macron was able to defuse the yellow vests movement in 2018.
But we are in the middle of a fiscal crisis with heavy upward pressure on French treasury bond yields and the government knows that, were it to try to defuse the situation by spending money it doesn't have, the yields may get out of control and turn the fiscal crisis into a full-blown panic.
As for why this is happening, I think people are overthinking it. I constantly see people who take at face value the grievances expressed by the most articulate of the protesters and assume they are the cause of this movement, but I think it's kind of ridiculous.
Have people ever talked to a 16-year-old before? They are protesting because it's fun, there is something exhilarating about thinking you're part of something important, it gives you an excuse to miss class and the carnival atmosphere in which normal rules seemingly no longer apply is very exciting.
This is especially true for young male protesters. One thing you must never lose sight of is that male human teenagers are, by a wide margin, the stupidest creatures in the universe. This explains a lot of what is happening and has nothing to do with complicated sociological explanations.
Of course, if journalists interview them, they will talk about the lack of resources, the long hours, etc. and sometimes even bring up more general issues about the current socio-political environment, but they're just repeating the rationalizations they hear people talk about on TV or, more likely in that age group, on social media.
It doesn't mean that the specific grievances they are talking about when you elicit justifications for them are the causes of their behavior and I think to believe that you really need to have a very unrealistic view of what a typical 16-year-old is like. The causes are much dumber than that in all likelihood and it's the kind of things I was talking about above.
Now obviously this kind of things doesn't explain why this is only happening now and not all the time, high-schoolers have always loved to miss class and feel they're part of something important yet they weren't protesting and burning their schools until recently, but here too I think people are kind of overthinking things.
There is a tendency for commentators to assume that everything has some kind of deep, structural explanation, but maybe sometimes there isn't? It's the same thing for people who think it's orchestrated by a handful of groups conspiring in the shadows.
For instance, the press has revealed that French domestic intelligence had produced a note claiming that the far-left was behind the protests and conservatives naturally were eager to accept that explanation, but frankly if you read the articles about that note, it seems kind of weak sauce and doesn't really show much.
All it shows is that, in a handful of schools in a specific department, far-left activists joined the high-schoolers and advised them on what to do. I'm sure that's true, and I think it's true in many other places, but it doesn't explain why this movement happened when it did.
If far-left activists could start such a movement so easily, it would happen all the time because they always try to stir up shit, but it doesn't because it's not that easy to start a massive social movement and that's not something anyone really controls in a meaningful way.
I think what happened is that high-schoolers in a handful of schools had very localized grievances, for whatever reasons they decided to make a mess and organize a blocus of their schools, posted videos on social media and then it spread through social contagion.
This was fueled by the fact that France has a strong protest tradition, so in that environment people naturally think about this mode of action and the idea of going into the streets and blockading their school will come kind of naturally to them.
Once such a movement crosses a certain threshold, it's not particularly mysterious why it will very quickly spread, so the only thing that requires a special explanation is why it crossed that threshold in the first place, but I think it's quite likely that it was just the chance coincidence of random factors.
Again, commentators are reluctant to accept that sometimes there is no deeper explanation than that, because if that's the explanation then it means they don't have anything particularly interesting to say, but I think often there is nothing more to it than that and people shouldn't be looking for deep, structural explanations that don't exist. -
David Cournapeau reposted
Ethan Mollick
@emollick
Too many people think a little fear is a good thing in organizations. It isn't:
😱 Fear is the enemy of innovation & employee participation
🤝If you're civil, coworkers are 59% more likely to share info, employees work 71% harder, and health & well-being receive boosts, too! -
David Cournapeau reposted
Joel Grus 🤠
@joelgrus
when others zig I zag
-
David Cournapeau reposted
-
David Cournapeau reposted
Amina Green
@Amina_io
It's amazing to see so much hype for Haruki Murakami.
Ryu is so much cooler...
Jeffrey J. Hall 🇯🇵🇺🇸
@mrjeffu
Video: A Haruki Marukami mega fan reacts with sadness to the announcement that his favorite novelist did not win this year's Nobel Prize for Literature. In their haste to create an announcement sign, someone misspelled the name of the winner, Anne Carson.
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Umar Jamil
@hkproj
I’m in Seoul building Mistral’s founding Korea team.
Looking for kind, ambitious researchers and engineers to build and customise LLMs for real problems, from data and evals to RL and training at scale.
Here until 16th October. DM me or apply here: jobs.ashbyhq.com/mistral.ai/1722cba1-bc85-467a-994d-98090… -
David Cournapeau reposted
François Geerolf
@FrancoisGeerolf
Ce choix est une spécificité française. Est-il compatible avec le principe 14 du code de bonnes pratiques de la statistique européenne, qui exige des chiffres comparables entre pays ? Hélas, ce n’est pas le seul domaine où les chiffres 🇫🇷 ne le sont pas...
www.autorite-statistique-publique.fr/wp-content/uploads/2… -
David Cournapeau reposted
Max
@mxschumacher
I think your argument is quite weak, the UK has done a lot of privatizations and is in an even worse situation.
The figure that you refer to is dominated by transfer payments. Reforming pension and unemployment systems is the much more relevant lever to pull -
David Cournapeau reposted
François Geerolf
@FrancoisGeerolf
Très bonne décision.⤵️ Voilà le genre de « dépense publique » qu’on ne peut en effet plus se permettre.
Les Echos
@LesEchos
Le gouvernement veut transférer le financement des cures thermales vers les complémentaires santé ebx.sh/jkgzXc
-
David Cournapeau reposted
William Gibson
@GreatDismal
Having long been aware of those tech bros who read my work as utopian porn, where they’ve taken that doesn’t exactly surprise me.
-
David Cournapeau reposted
Bojan Tunguz
@tunguz
Some people are dunking on this, but IMHO THIS is the way that science publications should be done. Post-publication corrections and modifications should be THE norm, not an exception. Software developers had it right all along. We need to stop treating research publications as settled scriptures.
Dan Roberts
@danintheory
We've updated our GitHub math repo with 6 new Lean formalizations, 19 modifications, and 3 withdrawals. The repo now has ~42% top-line results formalized. We will continue to update the repo with new formalizations and with any errata we notice.
github.com/openai/math/blob/main/history.md -
David Cournapeau reposted
Nate Berkopec
@nateberkopec
a lot of people haven't updated on the fact that a lot of beliefs about programming are now falsifiable
-
David Cournapeau reposted
Dr Thomas Baudinette
@tbaudinette
I teach a Japanese literature class each year at the time the Nobel Prize in Lit is announced. I always dedicate a portion of the class as to why Murakami Haruki is not a serious contender for the award :P
(My bet for a current Japanese author who may deserve it? Tawada Yoko)
Jeffrey J. Hall 🇯🇵🇺🇸
@mrjeffu
Video: A Haruki Marukami mega fan reacts with sadness to the announcement that his favorite novelist did not win this year's Nobel Prize for Literature. In their haste to create an announcement sign, someone misspelled the name of the winner, Anne Carson.
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Nathaniel Bechhofer
@bechhof
Okay, so a knockdown of a standard that is neither a societal consensus nor the way the law works (modulo definitions of what rises to the category of rape), hence yet another post that muddies the waters
-
David Cournapeau reposted
Joel Grus 🤠
@joelgrus
1. A robot may not prove mathematical results or, through inaction, allow mathematical results to be proved
-
David Cournapeau reposted
Joël Niklaus
@joelniklaus
Super cool paper on rubrics for LLM evaluation by @deliprao and Chris Callison-Burch!
-
David Cournapeau reposted
Daniel Litt
@littmath
Well worth reading: proofsandprompts.com/2026/10/08/100-reactions-to-100-solu…
-
David Cournapeau reposted
Andrew Whitby
@EconAndrew
Same thing with data centers. We should have data centers! But I don’t want to live next to one. People who are suddenly made to should be compensated for that. You may say this would raise the costs of building them, but the costs are already there, they’re just borne unfairly.
-
David Cournapeau reposted
Philippe Lemoine
@phl43
You could make the argument that covid + the Ukraine energy crisis + the Iran energy crisis played a similar role as exogenous shocks. The problem is that exogenous shocks are things that predictably happen and will keep happening, what's unpredictable is when they happen or why, so if you're in a fiscal position that is fragile eventually you will run into serious difficulties.
-
David Cournapeau reposted
No Context Brits
@NoContextBrits
The British empires strikes back.
-
David Cournapeau reposted
James Riney🐠Coral Capital
@james_riney
I’m looking for young, hungry, independent thinkers to join Coral Capital’s investment team. Fluent in Japanese and English, deeply curious about technology, and ambitious about Japan’s future. Builders, researchers, and investors welcome.
Tokyo or SF.
Who should I meet? -
David Cournapeau reposted
-
David Cournapeau reposted
Stéphane Vojetta
@StephaneVojetta
Face à la colère des lycéens du pays, les Français ne peuvent que se demander comment il est possible que leur jeunesse se sente abandonnée par l’État alors que les budget de l’éducation ont augmenté considérablement ces 20 dernières années?
Voilà deux ans que j’ai découvert - puis j’ai participé à révéler - ce que je considère comme un scandale majeur, le détournement comptable du budget de l’école (entre autres missions de l'État) à la faveur du système des retraites.
Évidemment, le scandale n’est pas dans le fait de payer aux retraités ce qui est leur est dû.
Le scandale consiste à avoir mis en place un système de tuyauterie comptable basé sur des surcotisations payées par les ministères qui évite de reconnaître l’existence d’un grave déficit au sein de notre système des retraites.
Ce mensonge est un anesthésiant qui empêche de prendre la mesure du problème… et qui empêche de réagir. Cela explique pourquoi la France est aujourd’hui le pays d’Europe à l’âge de départ le plus faible par rapport à ses voisins, ce qui aggrave le problème et creuse notre déficit et notre dette publique davantage chaque année.
Pour en revenir à nos lycéens, entre 2006 et aujourd’hui le budget de l’Éducation est passé de 81 à 89 milliards en euros constants de 2026. Sur ces 8 milliards d’augmentation, 7 milliards ont été consacrés à … combler le trou des retraites. Plus de 80% des nouveaux moyens que nous pensions consacrés à l'éducation de nos enfants...
Ces budgets partis ailleurs, on ne les trouve effectivement aujourd’hui ni dans la rémunération nette des enseignants, ni dans l’entretien des établissements.
Tout cela est inacceptable et scandaleux et heureusement les médias et le grand public prennent conscience du problème. Mais j’accuse aussi bien la Cour des Comptes que Bercy et le COR (Conseil d’Orientation des Retraites) d’avoir participé durant des années à perpétuer cet exercice de dissimulation qui a pour but de vous tromper vous, citoyens, et nous, parlementaires (ou ex-parlementaires, dans mon cas). Ils devront un jour rendre des comptes, et j’y prendrai ma part si j’en ai les moyens.
Mais d’ici là il faut réformer ce système, et vite. -
David Cournapeau reposted
I Am Devloper
@iamdevloper
Ten years of Sad Affleck memes and it turns out he'd just been trying to install CUDA drivers the whole time.
-
David Cournapeau reposted
Dan Castellano
@ninja_padrino
Took me 3 years to track this Tetsuko Kuroyanagi mug down on Mercari🧘🏻♂️
VOGUE JAPAN
@voguejp
「書いてあることに自分を当てはめようと思ったりしても、そんな風にいかないわけだから。世の中はね」。黒柳徹子さんから今を生きる女性へ贈る、温かなメッセージ。 www.vogue.co.jp/celebrity/interview/2018-02-tetsuko-kuroy…
-
David Cournapeau reposted
Toad vous épie 🔳 ☢️💸
@ToadVousEpie
La politique en France
-
David Cournapeau reposted
Igor Brigadir 🇺🇦
@IgorBrigadir
Gm
Jteve Sobs
@jtevesobs
im sorry we need jungle im afraid
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
that's called exporting aluminum
P Equity Research 📰
@pequityresearch
Can China send over some of their electricity glut to the United States? 🤣
-
David Cournapeau reposted
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
Vitalik is very soberly explaining just how close to the cliff's edge we are
(quite)
vitalik.eth
@VitalikButerin
I don't recommend anyone scramble to move their funds to new wallets today. But we should take the risks to cryptography from AI-accelerated math seriously, and minimize our exposure to not just quantum-vulnerable cryptography, but also potentially AI-vulnerable cryptography.
The core new area of risk from this viewpoint is, unfortunately, ML-DSA / FHE / lattices.
(and it's also another reason, along with quantum, why ECDSA might fall even faster than expected, hence the "fresh address" recommendation)
So far most people have been in the mode of thinking "elliptic curves broken, hashes safe, lattices safe". But there is a good chance that the concrete security of lattices will take serious hits from the next two years of AI math.
The basic threat model is: factoring is something that naively takes 2^(n/2) time, but over decades smart people have found and optimized number field sieves, and degraded that to 2^O(n^(1/3)), which is why RSA keys and signatures need to be ~400 bytes (and not 64 bytes). What if there are skeletons in the closet like that, both for elliptic curves and lattices, that we are simply not smart enough to discover - but bots soon will be?
This is a major part of the reason why for the past year ethereum's lean roadmap has been going in the "hash-only" direction: no lattices, no ML-DSA, no Falcon, no lattice-based commitments inside ZK proofs, etc. Signatures in lean ethereum are all hash-based, either WOTS or SPHINCS-.
For signatures and proofs, we already know how to go hash-only. The bigger challenge is for *public-key encryption* - and this goes far beyond blockchains. Secure communication, anonymizing protocols, lots of things need public-key encryption.
And unfortunately there are long-standing mathematical theorems showing why public-key encryption cannot be done with hashes alone. You have to have some kind of trapdoor object that has at least one form of usable "structure" - either group theory (incl. isogenies) or lattices or code-based or potentially in the future even more newfangled and spooky things (local mixing?). But for anything that has structure, you should assume that AI will make at least some progress in breaking that structure. Here, one reasonable inference is that if you want to make something plausibly long-term secure, multiply the key sizes by 10.
To me that's a very plausible world and something not at all extreme to predict. If AI will bring us 50 years of math in 2 years, then that 50 years of math may very plausibly include a "naive factoring -> GNFS" level of improvement to our ability to break lattices. In that world, lattices will still exist, but they will have to be significantly bigger to guarantee the same level of safety.
And at those new larger sizes, hash-based constructions will beat lattice-based constructions on concrete efficiency in every use case where hash-based constructions are possible at all.
Theoretically, of course it's possible that hashes are broken too (eg. P = NP would imply that). But I think P = NP is very unlikely. And intuitively, it's much more likely that a mathematical object has exactly no exploitable structure (like hashes are intended to), than that a mathematical object has exactly ~3 forms of exploitable structure (for elliptic curves: associativity, Schoof, pairings) and not some secret fourth form of structure we have not yet discovered that greatly degrades its security (for elliptic curves, ECDLP and pairing security). Similar for LWE, SVP, RLWE and the zoo of lattice problems.
For this reason, we do not yet see any reason to worry and start padding the byte size of hashes (if we start to worry more, we would pad the round count first before doing anything to the byte size).
Concrete TLDR, my own personal views:
* Hash-based > lattice-based, in those situations where hash-based is possible at all
* For anything lattice-based, be much more paranoid on param sizes. Remember that blockchains are only a small portion of the cryptography story; this point goes far beyond blockchains and applies to eg. access to websites, secure messaging, Tor / VPNs ...
* For privacy protocols, strongly favor NOT putting encrypted notes onchain. Instead, send them offchain through some third-party mechanism.
* If it's not difficult for you, keeping your funds in addresses which have not yet been used to make a transaction is a good idea. If it's easy for you, do it. **But be careful about migrations; I personally have lost more money in botched migrations than I have lost in all hacks combined**.
* For multisig wallets, doing confirmations offchain is better than onchain, because this way the signatures of signer wallets do not get exposed to the public, so if ECDSA falls to AI much faster than expected, at least the multisig "gracefully degrades" to a 1-of-1 where the 1 is whoever was gathering the signatures - a much better place to be than "anyone can take the money"
firefly.social/post/x/2107837081313505768 -
David Cournapeau reposted
Michael Fowlie
@mwfowlie
And this is why I’m bullish on software engineering jobs even with AI.
If anything we are needed more than ever to automate all these different tasks.
Brian Lui
@brianluidog
I think programmers kind of underestimate their vibe coding advantage versus non-programmers. I know a guy who spent 4 billion tokens to make a chess webapp.
-
David Cournapeau 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 -
David Cournapeau reposted
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
this is gibberish and a cause to ignore superforecasting altogether. What "experts"? How does one acquire an expertise in AGI? As DeepSeek says, no one has experience in what hasn't happened. You can only have relevant experience in learning stuff and reasoning clearly, I guess.
Jon Stokes
@jon_stokes
Dario & other AI insiders Scott aligns w/ are giving mid to high single- & multiple-digit p(doom) numbers, but when this question is put to superforecasters they give it 0.3% chance by 2030, 2% chance by 2050. Tetlock's own book says the further out you get the less the method actually works. So the highest-quality superforecaster number 0.3%.
forecastingresearch.substack.com/p/automating-catastrophi… -
David Cournapeau reposted
alex bronzini-vender
@alexbronzini
"As for Chamath, forcing him to humiliate himself on stage is neither clever nor novel; this is called the All-In podcast, and occurs weekly."
Daniel Yang (he/him)
@punished_daniel
Scott gemmed out
-
David Cournapeau reposted
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
China cares less about you than is often imagined
Tyler
@rezoundous
I wonder what happened to Chinese AI. They all went quiet after Opus 5.5.
-
David Cournapeau reposted
signüll
@signulll
in hindsight you could’ve just built a neo lab which might’ve been 10x easier.
instead you chose the application layer, where you get to combine all of the complexities from product, distribution, platform risk, user psychology, integrations, pricing, commoditization, retention, & whatever the model companies decide to vaporize next week.
excellent decision. -
David Cournapeau reposted
Lucas Beyer (bl16)
@giffmana
I'm starting to think mathematics were not beautiful because it "describes the beauty of nature" but rather because human mathematicians only went after beautiful but suboptimal descriptions lol
Quantіan
@quantian1
You can do Fourier transforms in O(N log(N)^0.9999999999999) time. Fuck this gay chud universe
-
David Cournapeau reposted
Federico Vaggi
@F_Vaggi
What if this poster is operating on an entirely other level and this post is part of a humiliation kink?
-
David Cournapeau reposted
Llion Jones
@YesThisIsLion
Come listen to my stories about how mad being on the Transformer paper has made my life!
GDG Greater Kwansai|10/18(日) DevFest Kansai
@gdgkwansai
「Attention Is All You Need」の共著者・Sakana AI CTOの、Llion Jones氏がDevFest Kansaiに登壇決定🐟‼️
歴史的論文の舞台裏、知られざる事実、そして激変した研究者の人生とは?
貴重なトークをお見逃しなく🚀✨
📅 10月18日(日)
📍 大阪
👇詳細・申込はこちら👇
gdgs.jp/df26xs13
「I didn't need this attention」
Llion Jones / Co-Founder & CTO, Sakana AI, a co-author of “Attension is All You Need” -
David Cournapeau
@cournape
Even Johnny Rotten from PIL and Sex Pistols confessed becoming a workaholic after moving to New York in his famous I like NY interview !
Jack
@tracewoodgrains
unamerican striver mentality. sad to see
David Cournapeau
@cournape
-
David Cournapeau reposted
Jean Mizrahi
@JeanMizrahi
Tant qu’il existera des tickets-restaurant en France, c’est que les Français n’auront pas compris que la source de leurs problèmes est toujours bien présente.
Le ticket-restaurant est un formidable résumé du mal français. On commence par taxer et charger lourdement le salaire, puis, comme le salarié couine un peu trop parce qu'il aimerait tout de même conserver un peu de pouvoir d’achat, on invente un revenu parallèle, partiellement exonéré, mais qu’il ne peut évidemment pas utiliser comme il le voudrait. Ce serait trop simple ! Il faut donc créer un "titre spécial de paiement", définir qui peut l’émettre, qui peut l’accepter, combien l’employeur doit financer, quels produits on peut acheter, quels jours on peut l’utiliser, dans quelles limites et selon quelles procédures. Et naturellement, une fois l’usine à gaz construite, qui rémunère généreusement les intermédiaires, il faut régulièrement la réformer.
Nous y sommes encore aujourd’hui. Des députés veulent ajouter de nouvelles contraintes à un mécanisme déjà compliqué : dématérialisation, contraintes sur les produits qu'on peut acheter, commissions, transparence des émetteurs, dons, plafonds, conditions d’utilisation. Cela dit beaucoup du décalage entre une partie de la classe politique et le réel. Au lieu de se demander pourquoi un tel dispositif existe encore, on cherche comment le rendre plus sophistiqué.
La question de fond est pourtant assez simple : pourquoi ne pas laisser cet argent directement sur la feuille de salaire, et non fiscalisé ? Parce que tout notre système fonctionne ainsi. On prélève beaucoup, puis on restitue un peu sous forme d’avantages soumis à des conditions. Et chaque avantage crée sa réglementation, ses bureaucrates, ses intermédiaires, ses logiciels, ses contrôles et ses exceptions. On fait croire au salarié qu’on lui « donne » quelque chose, alors qu’on lui rend simplement, sous une forme administrée, une partie de ce qu’on lui a pris ailleurs, en lui expliquant au passage comment il doit la dépenser. Parce que c'est bien connu, les Français ne sont pas aptes à décider eux-mêmes de comment dépenser leur argent.
Le ticket-restaurant n’est donc pas une bizarrerie, c’est un symbole. Il résume parfaitement une société où l’État veut mettre sa grosse main maladroite et graisseuse sur tout ce qui passe, puis organise minutieusement la manière dont une fraction de l’argent peut revenir à ceux qui l’ont gagné. Et lorsqu’un système produit des effets pervers, le premier réflexe des élus et des fonctionnaires n’est jamais de le supprimer, mais d’ajouter une nouvelle couche de règles pour corriger les précédentes.
C’est exactement ainsi que se construit l’hyper-bureaucratisation : non par une décision spectaculaire, mais par l’empilement de milliers de petites réglementations qui paraissent chacune défendables isolément. Le vrai problème français n’est donc pas seulement le niveau des prélèvements. C’est cette volonté permanente de tout encadrer, de tout orienter, de tout conditionner, jusqu’à transformer un simple complément de rémunération en dispositif administratif à part entière.
Le jour où quelqu’un proposera simplement de supprimer les tickets-restaurant et de transformer leur avantage social en rémunération nette équivalente, sans perte pour le salarié ni surcoût pour l’entreprise, on pourra commencer à croire qu’un changement de logiciel est possible. Les intermédiaires inutiles y perdront, mais la société y gagnera. En attendant, on continuera à débattre gravement de ce que l’on peut acheter avec une carte spéciale, et à appeler cela une "politique sociale".
BFM Business
@bfmbusiness
Des députés proposent un plafond de 25 euros au restaurant contre 10 euros au supermarché: la loi sur les tickets-restaurant arrive à l'Assemblée
l.bfmtv.com/yHjd -
David Cournapeau reposted
Julien Chaumond
@julien_c
“Open weights release end of October.” 🎉
BOOOOM WE ARE SO BACK
huggingface.co/mistralai/Mistral-Large-4.0-1T05-A52B
Mistral AI
@MistralAI
Meet Mistral Large 4, aka Le Chonk.
• 1T parameters, natively multimodal. 49B active.
It is the best open weights model from US or Europe on aggregated benchmarks.
• State-of-the-art on critical workloads, including cyber defense, manufacturing and finance and it surpasses closed frontier models on visual grounding.
• Forged in Europe end-to-end and is deployable from Europe via our own Mistral Cloud infrastructure.
• Available to all via API today. Working with cybersecurity partners privately.
Open weights release end of October. -
David Cournapeau reposted
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 -
David Cournapeau reposted
corsaren
@corsaren
These results don’t matter in any practical way, but imo it does expose a fundamental messiness in the world that our human preference for neatness and order obscured.
AI may show us that a lot more things are like packing n=17 squares than n=4. In this sense, it is blasphemy.
Acer
@AcerFur
oh yeah btw guys we can do integer multiplication faster than n log n lol
I was definitely very surprised when this one came in lol
github.com/openai/math/tree/main/preprints/Integer-multip… -
David Cournapeau reposted
Epoch AI
@EpochAIResearch
How do Chinese AI firms make money? In a new report, @cherylwoooo and @ansonwhho identify five revenue sources and analyze how much each contributes. 🧵
-
David Cournapeau reposted
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
1. It's funny but unlike distillation, algorithmic theft is a perfect crime, American labs are too retarded and greedy to publish their algos so the Chinese get the priority no matter how they arrive there. For all we know, V4.1 is a smol Astra.
2. compute is compute
Minh Nhat Nguyen 🦭
@menhguin
if chinese labs did have these secrets, logically:
1. it would show up inside newer models that are already open-source
2. they wouldn't have to distill -
David Cournapeau reposted
Andrew
@Cosmic_Andrew1
Lmao
Financial Times
@FT
SpaceX looks to raise $40bn to buy Nvidia chips in financing led by Apollo ft.trib.al/AXTvVlK
-
David Cournapeau reposted
Ken Silva
@JD_Cashless
The ex-CIA official caught w/ $40 million in gold bars pled guilty today to defrauding the federal government.
This is a picture of the gold and cash found in his home:
Ken Silva
@JD_Cashless
The ex-CIA official caught w/ $40 million in gold bars is having a detention hearing tomorrow, part of which will be held in secret due to the Top Secret info being discussed.
-
David Cournapeau reposted
David Thesmar
@dthesmar
I love this website. My TL is deep into government accounting rules, so I am now correcting for "excess pension contributions", which artificially inflate education, defense, police. Pensions explain a bit more, but not all: 3.1 out 7.5ppt difference in spending with Germany.
David Thesmar
@dthesmar
Trying to figure out how France differs from other OECD countries in terms of spending composition. Turns out pensions are part of the story, but definitely not all:
-
David Cournapeau reposted
Chris Paxton
@chris_j_paxton
reasonably optimistic about a diffusion of manufacturing across the world; not at all optimistic about manufacturing jobs. it'll be like farming where it becomes spectacularly productive and employs no one
Chris Paxton
@chris_j_paxton
Honestly "bringing back manufacturing jobs: is basically all virtue signalling. People don't want their kids doing jobs like this (chinese industrial safety video)
Maybe itll all be done with robots but as you say that'll make it even harder to sell to the public -
David Cournapeau reposted
Henry Shevlin
@dioscuri
holy shit ChatGPT is actually developing humanlike academic skills
Nemo
@thecaptain_nemo
thanks chatgpt
-
David Cournapeau reposted
Lenny Rachitsky
@lennysan
Inspiring founder of the day: Todd Masonis of @DandelionChoco. Former tech founder turned chocolatier.
The dream. -
David Cournapeau reposted
Dan Castellano
@ninja_padrino
Reminder that “overtourism” in Kyoto is not solely a foreigner issue
Nippon.com
@nippon_en
The total number of Japanese and international tourists to Kyoto exceeded 60 million for the first time in 2025. Kiyomizudera was the most popular spot to visit.
-
David Cournapeau reposted
Senior PowerPoint Engineer
@ryxcommar
i think sexting chatgpt is less embarrassing than asking chatgpt "what do you think my iq is"
Raq
@raqisright
Asked ChatGPT my perceived IQ & to make me a report card based on our conversations
-
David Cournapeau reposted
Philippe Lemoine
@phl43
Si tu veux vraiment être précis et que tu veux prendre en compte le fait qu'une partie des cotisations employeur que l'État est réputé verser sur les salaires des fonctionnaires constitue en réalité une subvention implicite au système de retraite, il faut non seulement réduire les montants des différents postes budgétaires en valeur absolue, mais aussi prendre en compte le fait que, en raison des normes de comptabilité nationale (i. e. une partie des cotisations employeurs sont requalifiés en transferts entre administrations publiques et donc n'entrent plus dans la production non-marchande du secteur publique), cela implique une réduction du PIB qui va également affecter les pourcentages dans ton tableau. blog.observatoire-hexagone.org/p/le-traitement-des-pensio…
-
David Cournapeau reposted
Eric M Hamilton 📚🃏
@pyritedreamer
90% of advice on the internet.
The Pawverse
@thepawmic
-
David Cournapeau reposted
Noah Smith 🐇🇺🇸🇺🇦🇹🇼
@Noahpinion
The idea that Asians are robots who never reach the very top levels of ability is such sad, desperate cope.
Show me a cellist with as much ability, creativity, and passion as Yo-Yo Ma, a mathematician as deep and brilliant as Terence Tao, a science fiction writer as creative and thoughtful as Ted Chiang, an entrepreneur as canny as Jensen Huang.
No, your race does not give you the spark of genius.
Pascal-Emmanuel Gobry
@pegobry_en
Ok, here’s a story about the Great Asian Grindset Debate.
A couple years ago we put our kid in a piano school run by a Korean woman living in France.
There’s a recital at the end of the year where all of the kids play; half Asian, which is very unusual for France, half White.
The Asian kids were all excellent. They were all an 8 or a 9.
The White kids were all over the place. There were 1s. There were 2s. There were 3s. There were also 10s.
The Asian kids all played, sorry but it’s true, somewhat robotically (which is no shame for a kid!).
The best White kids played with that extra dimension of soul and emotion that made them touch greatness.
The Asian kids were *all* 8 or 9. The White kids were all over the place from 1 to 10.
At the end the lady who runs the school gave a little speech, which was full of nice things, but had as a big throughline chiding the parents of the White kids for not making their kids practice more. At the end, she made a joke: "Remember, make your kids work, because no matter how hard they work, there’s an Asian kid next door who’s working twice as hard." We all laughed, myself included.
Do I think that the culture that produced those 8-9 kids is healthy, or one that I would like to see dominant in my country? Hell no.
Do I also wish that the White kids who were below an 8 had worked, say, at least 10% harder? Some 50% harder? Hell yes. -
David Cournapeau reposted
Sylvain Catherine
@sc_cath
France is more literary and abstract than the Anglo-Saxon world, but what it gains in eloquence, it loses in empirical grounding and practicality.
Eoin Daly
@eoinmauricedaly
Not to essentialise or fetishise but there is this culture of political fluency and articulacy in France that might exist elsewhere for all I know, but certainly has no parallel in the Anglosphere
-
David Cournapeau reposted
Restitutor Orientis 🇱🇮
@restitutorII
🇩🇪🇷🇺
Allemagne: August Hanning, ancien patron des services de renseignement allemands (BND), a été arrêté, soupçonné d’espionnage, notamment de divulgation de secrets d’État, de tentative de trahison et d’avoir travaillé pour la Russie. -
David Cournapeau reposted
L’hoir de Francion
@HoirDeFrancion
Les « revendications » des lycéens sont, pour la plupart, absurdes, mais leur critique des emplois du temps ne l’est pas entièrement.
La réforme Blanquer de 2018 a fait beaucoup de mal, et les lycéens bloqueurs la dénonceraient explicitement s’ils avaient plus de discernement.
Cette réforme a aboli les séries et leur a substitué un tronc commun, auquel s’ajoutent des enseignements de spécialité. Chaque élève en suit trois en première, à raison de quatre heures chacun, et n’en conserve que deux en terminale, à raison de six heures chacun.
Cette organisation, inspirée de l’université, a conduit à l’éclatement des groupes-classes, ce qui a compliqué la réalisation des emplois du temps. Les élèves d’une même classe ne font plus la même journée, le nombre de professeurs intervenant devant une classe a fortement augmenté, et les journées sont souvent allongées, avec des trous de plusieurs heures, ce qui explique en partie les demandes des lycéens.
Et tout cela pour quel avantage pédagogique ? Aucun ! Le bac n’existe quasiment plus réellement : l’ancien bac comptait environ deux fois plus d’écrits nationaux, et une partie importante de la note finale vient maintenant du contrôle continu, alors qu’autrefois seules les épreuves finales comptaient, à quelques exceptions près.
Résultat : des élèves moins bien formés et un bac dont la valeur dépend de l’établissement, sans compter ces emplois du temps mal optimisés, avec plus de trous. Le pouvoir macronien n’a décidément rien laissé debout ! -
David Cournapeau reposted
JohannesBorgen
@jeuasommenulle
Wow
Germany’s former foreign intelligence chief August Hanning has been arrested on suspicion of treason. (FT) -
David Cournapeau reposted
No Context Brits
@NoContextBrits
There’s always that.
-
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
My country is destroyed by DEI, which is why I hate when room temperature iq Indians come here and destroy me on standardized tests, which are very g loaded and correlate entirely with IQ, but only when blacks and Hispanics take them.
-
David Cournapeau reposted
Satnam Singh
@satnam6502
“AI is being used to isolate people from each other.” A wonderful quote from @KentBeck at a panel session at SPLASH 2026. For me coding has always been a social human activity that has given me a great deal of meaning and purpose in my life, as well as community and friends. I hope we can go forward with the advantages of AI but retain the unique human value to the process and endeavor. Coding is about the machine, but software engineering is about the people.
-
David Cournapeau reposted
Eric Ries
@ericries
The Kindle edition of Incorruptible is $1.99 right now. If you've been meaning to read it, this is a good moment. incorruptible.co
-
David Cournapeau reposted
Brian Lui
@brianluidog
Would people prefer to be given a fish, or taught how to fish? With LLMs we now decisively know the answer: no one wants to learn how to fucking fish.
-
David Cournapeau reposted
François Geerolf
@FrancoisGeerolf
« La 🇫🇷 est en déficit de sa balance des paiements. » Oui, mais faiblement : de l’ordre de 0,6 % du PIB. Rien à voir avec 1981, ni avec la Grèce en 2010 !
Le problème aujourd’hui, c’est le déclin, le déclassement, "le travail qui ne paie pas" - pas un excès de pouvoir d’achat. 🤷♂️
Les Echos
@LesEchos
💥 « La France est au bord du précipice »
🇫🇷 @jattali alerte sur la situation économique de la France. Il appelle à une réaction rapide pour éviter une dépendance accrue envers ses partenaires européens.This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Noah Smith 🐇🇺🇸🇺🇦🇹🇼
@Noahpinion
Actually, the sad truth is that the 2020s should be an absolute flowering of Chinese pop culture like the 90s were for Japan, but it won't, because CCP censorship is just bad enough to make Chinese pop culture boring af
Noah Smith 🐇🇺🇸🇺🇦🇹🇼
@Noahpinion
On the plus side, this sort of post-boom frustration can produce some pretty good movies
-
David Cournapeau reposted
Sylvain Catherine
@sc_cath
Il y a trois ans, jour pour jour.
LCP
@LCP
"Aujourd'hui, si la dette explose, c'est d'abord à cause du système de retraite", affirme @BrunoLeMaire. #lcpol
-
David Cournapeau reposted
Wittgensteino-Aronien🌐
@wittgensteinien
-
David Cournapeau reposted
Resident Advisor
@residentadvisor
Intellitronic, a project from Drexciya luminary Gerald Donald, has received its first full release. ra.co/news/86147
-
David Cournapeau reposted
Mario Zechner
@badlogicgames
armin, uses present tense, doesn't use business lingo.
Jean-Claude
@jauneclaude
Which intern did it better?
-
David Cournapeau reposted
Hugo Bowne-Anderson
@hugobowne
“I like to think that agents are like coffee. They can get you to do stupid things faster with more energy.”
@GaelVaroquaux on the latest Vanishing Gradients 😂
I joined Gaël (@probabl_ai, @scikit_learn), Shipra Arora from Bain & Company, and @lfiaschi86 from @pymc_labs to talk about the state of agentic data science: what we can delegate, how we verify the results, and what we could do beyond speeding up our existing work.
Gaël has seen experienced data scientists getting more done with agents, and people without that expertise thinking they’re getting more done. When your code runs and the accuracy looks beautiful, how do you check whether the analysis actually supports your conclusion?
We get into programmatic checks, agents critiquing other agents, and where human judgment still matters. We also explore what becomes possible when business users can question models and run simulations themselves, and data teams can finally try the ideas sitting in their backlog.
Here’s a taste of the conversation, including my attempt to get this very polite panel to argue with each other.
Full episode in the replies 👇This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Pessimists Archive
@PessimistsArc
Remember when the Vatican spent 500 years trying to align the printing press?
Pessimists Archive
@PessimistsArc
1501: Pope Alexander VI criticizes Gutenberg’s printing press safety
“The art of printing can be of great service in so far as it furthers the circulation of useful & tested books; but it can bring about serious evils l..it will, therefore, be necessary to maintain full control”
Financial Times
@FT
Pope Leo criticises Nvidia’s Jensen Huang over AI safety ft.trib.al/azwdiGc
-
David Cournapeau reposted
Romy
@Romy_Holland
it’s a little offensive that my baby will reliably eat sand but not anything i’ve cooked for dinner in the last 2 weeks.
-
David Cournapeau reposted
vicki
@vboykis
get off the TL sometimes, it’s good for you, brother they’re still writing java 8 out there
-
David Cournapeau reposted
Wálé Akínfadérìn
@WaleAkinfaderin
My grandma purchased this record around 1976. She passed this and some of her other records along to me before she passed away. They are absolutely priceless!
-
David Cournapeau reposted
Rayan Nezzar
@Rayan_Nezzar
Merci @Ccesoir pour votre invitation à débattre de la fracture générationnelle.
J'ai notamment rappelé que depuis 1995, la part des dépenses de retraites dans notre richesse nationale avait crû de 20 % quand celle des dépenses d'éducation reculait de 15 %.
Tout se passe comme si nous financions notre protection sociale avec la carte de crédit de nos enfants et de nos petits-enfants.This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
hardmaru
@hardmaru
Registration now open: From World Models to Real-World AI
Join us on October 26 in Tokyo for a symposium with @SchmidhuberAI, who recently joined @SakanaAILabs as our Chief Scientific Advisor.
The program features a keynote and Q&A by Jürgen and short talks by Sakana AI researchers on our work at the RSI Lab.
- Date: Monday, October 26, 2026, 15:00–18:00 JST
- Venue: Hitotsubashi Hall, Tokyo, Japan 🇯🇵
- Language: English
- Admission: Free (registration required)
Seating is limited, so please register early.
Register here: luma.com/kjf0z516 🐟
Sakana AI
@SakanaAILabs
【受付開始】From World Models to Real-World AI — 日本から拓くAIの新時代
Chief Scientific Advisorとして当社に参画したJürgen Schmidhuber博士を迎えるシンポジウムを、10月26日(月)に東京・一橋講堂で開催します。本日より一般申し込みを開始しました。Schmidhuber博士による基調講演とQ&A、 CEO David Haとの対談、Sakana AI研究者によるショート講演を予定しています。
日時:2026年10月26日(月)15:00〜18:00
会場:一橋講堂(東京・千代田区)
言語:英語(通訳なし)
参加費:無料(事前登録制)
※ 席に限りがありますので、お早めにお申し込みください。
お申し込みはこちら:
luma.com/kjf0z516 🐟 -
David Cournapeau reposted
-
David Cournapeau reposted
Mario Zechner
@badlogicgames
no, just vibes, like everybody
-
David Cournapeau reposted
Dōgen
@Dogen
the japanese concepts of tatemae (建前) and honne (本音): a simple explanation
tatemae is the immigration officer complimenting my japanese
honne is the same officer rejecting my visa application -
David Cournapeau reposted
Sylvain Catherine
@sc_cath
Official French statistics on education spending are simply wrong. As in every country, they include pension contributions, except that in France the contribution rate is 93%, three to four times what could be considered a normal rate.
This absurd contribution rate is an accounting gimmick that allows the French state to claim its pension scheme is balanced, but it has nothing to do with the pensions that teachers currently in the classroom will actually receive.
Recent reports have found that, after correcting for this bias, French spending on primary education, for instance, is not slightly above the OECD average, as official statistics suggest. Rather, it falls in the bottom quartile.
David Galbraith
@daveg
French schools aren't underfunded: 5.4% of GDP vs 4.9% OECD, more than Germany. The problem is central planning (only 2% of decisions are taken by schools), teacher absence (670,000 strike days) and poor discipline (52% of pupils report classroom disorder vs 32% OECD), as witnessed by children setting widespread fires and riot police having to be sent in, which is clearly not normal.
-
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
Reheating leftovers and having a random part still cold is such a weirdly awful experience with technology.
-
David Cournapeau reposted
Alister
@SchnockRevue
-
David Cournapeau reposted
(((ل()(ل() 'yoav))))👾
@yoavgo
wait what?? faculty in post-LLM AI are farting out 10 phds/year and Richard Sutton had 16 graduating students in his entire career?? this is amazing. i respect him so much more now
Richard Sutton
@RichardSSutton
The PhD thesis of my 16th PhD student, Fernando Hernandez Garcia, is now available.
Title: Selective Reinitialization Algorithms for Preventing Plasticity Loss in Artificial Neural Networks
Url: www.incompleteideas.net/papers/FernandoPhD.pdf
Abstract:
In this dissertation, I study systems based on artificial neural networks that learn from nonstationary data. Learning from non-stationary data requires continual adaptation of the system, and is often referred to as continual learning. Developing systems capable of continual learning is a longstanding goal in artificial intelligence. In deep learning, the field concerned with designing and training deep neural networks, this goal remains elusive.
This dissertation addresses a fundamental challenge in continual learning: loss of plasticity, a phenomenon where artificial neural network systems progressively lose their ability to learn from new data. While the phenomenon has been noted several times over the last three decades, it has remained understudied until recently. The work in this dissertation constitutes the first systematic demonstrations of plasticity loss, highlighting its persistence and importance.
In this work, I provide a systematic demonstration of plasticity loss in a wide variety of deep learning systems. Across systems based on fully-connected networks, convolutional networks, residual networks, and vision transformers, plasticity degrades when learning continually. Notably, even systems employing normalization techniques, residual connections, and regularization—design choices that improve training stability—remain susceptible to plasticity loss. This evidence establishes that plasticity loss is pervasive and that deep learning systems trained with backpropagation are not suitable for continual learning.
I explore the idea of selective reinitialization to prevent plasticity loss. This idea has been used in the past for improving generalization performance in deep learning systems. However, its use for preventing plasticity loss is a recent innovation pioneered by the continual backpropagation algorithm. The algorithm periodically sets new values for units in the network, making initialization a continuous process rather than a one-time operation. I demonstrate the effectiveness of continual backpropagation in preventing plasticity loss across a wide variety of settings. These demonstrations establish that plasticity loss, while pervasive, is not inherent to deep learning systems.
I generalize continual backpropagation through an algorithm I call selective unit reinitialization. This general algorithm has three key components: a utility measure that ranks units by importance, a pruning criterion that selects which units to reinitialize, and a reinitialization method that assigns new values to the selected units. Continual backpropagation involves a specific choice of utility measure, pruning criterion, and reinitialization method. However, the general algorithm is not limited to the choices used in continual backpropagation. I present a study of how different choices for the three components of selective unit reinitialization affect its effectiveness at maintaining plasticity. This study establishes selective unit reinitialization as a general approach that can be tailored to each learning system to maintain plasticity.
Finally, I propose a different approach to implementing selective reinitialization that operates at the weight level. I call the corresponding algorithm selective weight reinitialization. Reinitializing at the unit or weight level involves different trade-offs. Unit reinitialization minimally disrupts the network outputs, preserving stability during learning. However, the definition of a unit varies across network architectures, requiring additional engineering to make the approach effective. Weight reinitialization, in contrast, can be readily applied to any arbitrary network architecture, but can substantially affect network outputs, reducing training stability. The reduced learning stability can be remedied with L2 regularization, which stabilizes selective weight reinitialization but introduces an additional hyperparameter. Both selective unit and weight reinitialization successfully maintained plasticity across the systems tested, providing flexible approaches for different architectural and engineering constraints.
Fernando is now a research scientist at Zyphra. -
David Cournapeau reposted
Stan Veuger
@stanveuger
In 2017 Delcy Rodriguez hired David Rivera, who got Rubio his first job in politics and used to live with him in the house they owned together. He got paid $50 million. In unrelated events, Rubio installed her as president of Venezuela.
-
David Cournapeau reposted
(((ل()(ل() 'yoav))))👾
@yoavgo
the "post-training is shaping behavior, not adding new knowledge" dogma is clearly no longer true. software engineering, math and computer use knowledge are clearly added in post-training. likely also other kinds of knowledge. is this documented/quantified/ontologized? where?
-
David Cournapeau reposted
Richard Kadrey
@Richard_Kadrey
True terror is to wake up one morning and discover that your high school class is running the country.
— Kurt Vonnegut -
David Cournapeau reposted
Mario Zechner
@badlogicgames
and all of it runs on my phone (except the llm obv)
-
David Cournapeau reposted
Mario Zechner
@badlogicgames
this weekend, i'm finally replacing Claude for Android with a little pi durable thing that runs fully on my phone. building it directly in pi on my phone. and it has subagents :D
(not an official earendil product, just a personal project to battle test durable) -
David Cournapeau reposted
Emma
@emmainSTEM
“I can change him”
Shivon Zilis
@shivon
It’s hard to go from in love to let go in a week with no warning, but that’s just how it is sometimes 🤷🏻♀️
I feel happy that I got to show Elon how it feels to be deeply understood, nurtured, and peacefully loved with stability and gentleness. Given the tumultuous nature of his early life, his often tortured soul, and how hard he fights for humanity I wanted him to feel the deepest and most stable form of love this world can provide. My heart is full knowing I got to do that, and I hope it leaves a lasting good effect.
We were always his safe space. A place where all of us were overjoyed to see him and one that was always full of happiness and love, not war, in contrast to his otherwise militant existence (a needed break from the battlefield where he could recharge his heart and soul before heading back out)
I loved being that counterbalancing place of safety, sense of calm, and love — and delighted in bringing a smile back to his face when he was down. I will miss him dearly.
I’ve loved him more than life itself and he’s taught me 100 lifetimes of knowledge and made my cup overflow with life meaning, so there is a lot of loss, but luckily we created four beautiful little loves of my life ❤️❤️❤️❤️ I am so beyond grateful for our children and they make every single day the best day of my life! I do hope and pray he and I can be great friends and coparents.
I’m hurting, but care deeply about E and will always be cheering for his happiness. God speed, Elon, the world is lucky to have you.
Big Tech Alert
@BigTechAlert
🚫 @elonmusk is no longer following @shivon
Still followed by: @miramurati, @karpathy, @kimbal, @turkeybeaver, @drfeifei & 52 more
$TSLA $SPCX
(🤖🧩: am i missing something?) -
David Cournapeau reposted
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
It's surprising how little buzz DeepSeek's Ascend release has generated. Journalists and Think Tank Experts still care more about legible PR. To be clear, it's so much more than DeepGEMM or another open checkpoint. It's a warning shot for what Jensen talked about here.
Zhean Xu
@zheanxu
DeepGEMM Ascend is now open source.
We achieved 99.8% of the hardware limit on GEMM and 98% on MegaMoE.
github.com/deepseek-ai/DeepGEMM-Ascend -
David Cournapeau reposted
Steven Pinker
@sapinker
An intellectually deep and beautifully written essay on consciousness and AI, by neuroscientist Anil Seth | Why consciousness is more likely a property of life than of computation and why creating conscious, or even conscious-seeming AI, is a bad idea | @anilkseth in @NoemaMag
-
David Cournapeau reposted
need 6 certi for Saturday #MVPurdy
@HollowPoint_USA
This isn't a Gen Z phenomena. The Boomers largely opted out of saving for their own retirement and now we get a lecture every day about why the state pension needs to grow until its 100% of govt spending.
BBC News (UK)
@BBCNews
'It could cost me £10k but I need the money now': Why Gen Z are opting out of pensions www.bbc.com/news/articles/c68xk0ndqz8jo
-
David Cournapeau reposted
dinos
@din0s_
wake up babe the freaking pope got schmidhuber'd
-
David Cournapeau reposted
Gérard Araud
@GerardAraud
L’essentiel de l’augmentation des dépenses publiques ne concerne pas l’Etat mais retraites, dépenses sociales et collectivités locales.
Brice Couturier
@briceculturier
Notre Etat obèse a embauché 500 000 fonctionnaires supplémentaires en 15 ans. Il a créé des centaines d'agences et de hauts comités. Il est vrai qu'il paie aussi les pensions de retraites de ses fonctionnaires. Mais pas les retraites contributives du privé.
-
David Cournapeau reposted
trish
@TrisH0x2A
TIL SQLite converts integers to text two digits at a time
converting an integer to text usually means a loop of /10 and %10
sqlite avoids most of those divisions by keeping a 200-byte string of every two-digit pair from 00 to 99, and pulling two digits at a time out of it by direct index instead of computing them -
David Cournapeau reposted
Lane || Boot.dev
@wagslane
Holy shit, America is cooked. They were able to perfectly clone the Microsoft Teams experience.
POLITICOEurope
@POLITICOEurope
The European Commission has rolled out an alternative to Microsoft Teams.
But officials using it aren't impressed, with one describing the tool as “absolute shit.”
vist.ly/5kje9 -
David Cournapeau reposted
Julien Cadot
@juliencdt
Il faut une loi d'urgence pour que l'électricité massivement décarbonée de la France puisse être utilisée pour alimenter des data centers dédiés à l'IA, le plus vite possible et sans entrave.
C'est une opportunité historique qui ne se représentera pas deux fois.
Le Canard enchaîné
@canardenchaine
Et de cinq ! Mercredi 30 septembre, quatre recours supplémentaires ont été déposés devant la justice administrative contre le gigantesque projet de data center de Mistral à 50 milliards d’euros qui doit voir le jour en 2027 aux portes de Paris.
Cet équipement, d’une surface équivalente à 100 terrains de foot, consommera à terme la puissance électrique d’un réacteur nucléaire.
Un impact écologique atomique pour ce centre de données choyé par Macron.
l.lecanardenchaine.fr/tp -
David Cournapeau reposted
Philippe Lemoine
@phl43
I almost made a joke about how those people should talk to a flat earther to see how easy it is to persuade retards.
-
David Cournapeau reposted
Olivier Blanchard
@ojblanchard1
Une réaction à chaud et un message a tous ceux qui sont en campagne présidentielle. Ne mettez pas de l’huile sur le feu.
Des propositions irresponsables, au cours des dernières semaines, ont fait augmente les taux sur les obligations françaises, de 0.75% par rapport a l’Allemagne. Une augmentation du taux d’interet de 0.75% implique de trouver 20 milliards de plus d’économies ou d’impots supplémentaires pour stabiliser la dette par rapport au PIB.
Le budget propose par le gouvernement est loin d’etre parfait. Mais je pense que vous le préférerez à un budget exigé par la BCE pour aider la France a faire face à ses obligations tout en respectant son mandat. Celui-là sera beaucoup plus dur. Choisissez. -
David Cournapeau reposted
Sylvain Catherine
@sc_cath
Ce n’est bien sûr pas le seul problème mais c’est évident qu’on n’investit beaucoup moins dans l’école, à tous les niveaux, y compris l’université, que les pays comparables.
Gabriel Lattanzio
@GabLattanzio
Cette prise de parole aurait pu prendre une autre forme si on ne venait pas d'entendre parler de "racailles" pendant dix minutes. Ça a pollué.
Ceux qui me suivent savent que je suis pour une severité totale à l'égard des violents.
Mais là je voulais parler d'école. -
David Cournapeau reposted
nic carter
@nic_carter
They did the meme
David Decosimo
@DavidDecosimo
This is deeply disturbing:
Anthropic secretly invited religious leaders to SF to advise them on AI safety, had them sign NDAs, but then primarily tried to convince them Claude has a soul & moral standing, all while wining & dining them to the max. 1/ tinyurl.com/preview/296b5z8z -
David Cournapeau reposted
Jeremiah Johnson 🌐
@JeremiahDJohns
It's extremely funny that manufacturing construction triples during Biden's presidency, and then starts to decline almost *immediately* when Trump takes office the second time.
-
David Cournapeau reposted
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
Concerning for rivals
As AI becomes truly superintelligent, Trump will likewise ascend
Ordinarily he'd have been the bottleneck, but if he embraces being the skinsuit for Mythos, it's over for everyone else
Andrew Curran
@AndrewCurran_
From the accompanying article to the interview this morning in TIME.
-
David Cournapeau reposted
Pi
@pidotdev
People of Pi: We've shipped Pi 1.0 with Pi Durable. Go make them yours.
earendil.com/posts/pi-1-0/ -
David Cournapeau reposted
Obese Oblige
@qquinnsworth
1/n I'm Chinese and I love China but as someone who has lived in multiple countries, there are a million things that bother me about living in China. No day better than the birthday of this glorious nation to go over them 🧵
Lari Anjos
@DonaLarocas
Cite 1 problema real sobre a China sem ser sobre uso de VPN, pernilongos em excesso e dificuldade em se adaptar a culinária local
Talvez regiões lotadas e com muito trânsito? Mas me fale algo REALMENTE ruim aqui na China, porque pra mim os pontos positivos se superam em peso -
David Cournapeau reposted
Sylvain Catherine
@sc_cath
C’est un déni de réalité. Pour capitaliser, il faut un excédent de cotisations. On est dans la situation inverse: on emprunte, c’est-à-dire qu’on capitalise en négatif.
Il faut d’abord résoudre le problème de la répartition.
Paul Cébille
@Ellibec
La capitalisation reste la solution privilégiée par les Français pour garantir la pérennité du système de retraite.
55 % souhaiteraient toutefois qu'elle se limite à une dose afin que le système de retraite repose sur un mélange de répartition et de capitalisation.
Odoxa x Challenges -
David Cournapeau reposted
Florian Ederer
@florianederer
Austria introduced a package delivery flat tax of €2.40 for all online deliveries, mainly aimed at Amazon.
It will be interesting to see if the tax incidence is entirely on consumers or if online retailers will reduce prices.
Niklas Korber
@NiklasKorber
Ab heute 2,40 € Paketsteuer auf jede Bestellung in Österreich, egal welcher Warenwert. Amazon gibt das auch direkt so an den Kunden weiter.
So sieht das aus, wenn Sozialdemokraten die »bösen, großen Konzerne« besteuern. -
David Cournapeau reposted
Peter russell
@Pedro_jock
God I used to have to teach that as part of a presentation, thank Christ it’s now in a booklet! Think the worst thing was about that presentation was figuring out a lot of 80 and 90 years old were getting more action than me!
-
David Cournapeau reposted
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
Funny question:
is there a single Ascend 950 NPU in the US? or anywhere beyond China? They clearly have been selling them for a while. I've seen reviews of Chinese consumer CPUs/GPUs. Is anyone doing intelligence at least? Or only think tank blathering?
the tiny corp
@__tinygrad__
Anyone at Huawei want to give us two Ascend 950 boxes in exchange for a tinygrad port?
-
David Cournapeau reposted
Gael Varoquaux 🦋
@GaelVaroquaux
Tomorrow, I'm on the Vanishing Gradients podcast to talk about agentic science with @hugobowne , Shipra Arora from Bain & Company, and Luca Fiaschi from @pymc_labs
Online live panel, with QA; fun!
🗓️ Friday October 2, 1pm CEST / 7am EST
➡️ Register here: numfocus-org.zoom.us/webinar/register/WN_S4qoVkjoRA-IHtUu… -
David Cournapeau reposted
Géraldine Woessner
@GeWoessner
Climat : ces mesures qui coûtent une fortune pour si peu de résultats - EXCLUSIF. Isolation thermique, panneaux solaires en toiture… Pour éviter une tonne de CO₂, la France subventionne des mesures peu efficaces et hors de prix, révèle un rapport de @CGollier et @Fxol, qui appellent à revoir radicalement nos stratégies. 👇
www.lepoint.fr/societe/climat-le-cout-exorbitant-de-mauva… -
David Cournapeau reposted
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
Is Logan aware about the existence of inflation?
In real terms, the US hasn't been growing that fast.
anyway, good finance-first bear case on Chyna
Bill Bishop
@niubi
Logan Wright on Broken China, by @jordanschnyc open.substack.com/pub/chinatalk/p/logan-wright-on-broken-… "What a “broken China” means for the US and the rest of the world and why Logan argues there is no longer systemic economic rivalry with China, but only a military and industrial-resilience one."
-
David Cournapeau reposted
Philippe Lemoine
@phl43
The thing about Soviet restrictions on travel is that, from the point of view of the authorities, they were not fanciful but wise, because when Soviet citizens became able to compare how they lived with how people in the West lived things started to unravel pretty quickly.
(When I have time I will tell the story of when my father went to Lviv to fix machines that had been sold to a Soviet institute by his company after he and my mother had just bought their first apartment and the workers at the institute listened to him in disbelief as he described how they were in the process of choosing the wallpaper.)
People in the West tend to think of their position toward China with the same sense of superiority that we had toward the Soviet Union, and don't get me wrong we still are much richer, but it's quite telling that, by contrast with the Soviet authorities back then, the Chinese authorities don't need to impose such restrictions on travel.
In case you're curious, this passage is from The Second Russian Revolution by Angus Roxburgh (who also wrote a very good book on Putin), based on the interviews made for the BBC documentary of the same name. Most of it, though unfortunately not all of it, can be found on YouTube and I strongly recommend it. -
David Cournapeau reposted
Andy Sharp
@sharp_writing2
Japan has sharply raised residency fees and tightened the path to permanent residency as PM Takaichi puts a tougher stamp on immigration policy.
The changes bring much higher fees alongside new income, pension and Japanese-language requirements for PR applicants.
Our explainer looks at what’s changing, why the government is doing it, and the bigger question looming over Japan’s immigration debate: whether to cap the foreign resident population. -
David Cournapeau reposted
Uzay
@uzpg_
I am incredibly excited about our latest release of a state of the art writing model. Understanding how to train models on these more fuzzy kinds of tasks has been awesome, and I would like the models of the future to have the better parts of our human nature.
Fulcrum
@fulcrum_inc
Say hello to Echo, the best writing model at style imitation. Echo beats frontier models at writing tasks ranging from fiction to technical explanations, despite costing less than $5K to train.
-
David Cournapeau reposted
Graham Dumpleton
@GrahamDumpleton
Python 3.15 has a new sampling profiler, Tachyon. I wrote 21 hands-on JupyterLab workshops to learn it myself, and they are now ready for anyone to do. Launch on Binder, nothing to install. What I learnt, and why I wrote them:
grahamdumpleton.me/posts/2026/10/getting-to-know-tachyon/ -
David Cournapeau reposted
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
> It's the new best model in the world. What a turnaround
I need to learn to trust Wenfeng even more
In May, he was confident that Google will come back.
leo 🐾
@synthwavedd
🚨 BREAKING: Google have announced Gemini 4 Argon, their new flagship model currently being tested in partnership with the US Government before a wider launch
It's the new best model in the world. What a turnaround
blog.google/innovation-and-ai/models-and-research/gemini-… -
David Cournapeau reposted
Joachim Schaeffer
@JSchaeff3r
We stole reasoning. Again.
An update to our paper on reasoning extraction: Patching your own API doesn’t secure your cloud-hosting ecosystem.
Story in the 🧵 -
David Cournapeau reposted
Resident Advisor
@residentadvisor
Listen back to five mixes recorded at Outlook Origins, including Skee Mask's grime set, Kode9 b2b DJ Spinn, Beatrice M. and Pinch and more.
ra.co/news/86118 -
David Cournapeau reposted
No Context Brits
@NoContextBrits
Joy Division fans seeing this:
Bill Pourquoimec
@BillPourquoimec
Pulsar CP1919, Arecibo, Porto Rico, 1963.
-
David Cournapeau reposted
Resident Advisor
@residentadvisor
David Sylvian has announced his first album in more than a decade. ra.co/news/86115
-
David Cournapeau reposted
Satnam Singh
@satnam6502
I'm giving a talk about using AI theorem provers like @HarmonicMath's Aristotle to formally verify hardware at REBASE 2026 during SPLASH/ISSTA 2026 next week in Oakland. It's on Wednesday at 11:15AM.
conf.researchr.org/details/splash-issta-2026/splash-issta… -
David Cournapeau reposted
Callum Lyon
@CallumLyon
Just cancelled my Netflix subscription. So if the boomers are right I should be able to buy a house by next week.
-
David Cournapeau reposted
Doug Turnbull
@softwaredoug
I wouldn't bank your career on Jevon's paradox
Demand for SWE as a skill can increase. But the value of that skill can still go down.
IE a lot of people write. Being a good writer is useful. Very few people make tons of money just being a "writer". -
David Cournapeau reposted
Bruno Maçães
@MacaesBruno
This guy is now praising the razing of Warsaw by the Nazis
Marc Thiessen
@marcthiessen
Horrifying. Clear evidence of genocide…
Oh wait that’s Dresden after US and British bombing in WWII.. -
David Cournapeau reposted
BayesReality
@BayesReality
Well he was right, it caused the religions war than killed half of germany.
Pessimists Archive
@PessimistsArc
1501: Pope Alexander VI criticizes Gutenberg’s printing press safety
“The art of printing can be of great service in so far as it furthers the circulation of useful & tested books; but it can bring about serious evils l..it will, therefore, be necessary to maintain full control”
Financial Times
@FT
Pope Leo criticises Nvidia’s Jensen Huang over AI safety ft.trib.al/azwdiGc
-
David Cournapeau reposted
-
David Cournapeau reposted
Peter Wang 🦋
@pwang
Did you see the thing about insiders buying up a pile of .si domains before the announcement?
There’s always a grift in there somewhere -
David Cournapeau reposted
Rochelle Kopp / Japan Intercultural Consulting
@JapanIntercult
I assume by “people” you mean Japanese? If yes, I think there is a cultural element. I wrote about it here. japanintercultural.com/free-resources/articles/are-americ…
-
David Cournapeau reposted
Sebastian Raschka
@rasbt
Took a while, but here it is, a mega write-up on language models for text classification and, of course, Jev.
It's basically a visual guide to RNNs, CNNs, transformers, and calibration, with hands-on experiments on accuracy and efficiency. -
David Cournapeau reposted
FunK
@DaMFunK
Never signed a major label deal.
But, nevermind.. Just stayed me.
See everyone this Sat. (Oct. 3rd)
at the @ROVR_LIVE Festival which
will be my 9th time contributing 🎵
in this wonderful country. ✌🏾| DF -
David Cournapeau reposted
Jo Kristian Bergum
@jobergum
At this stage you still need to know what great looks like if you want to create value with AI
-
David Cournapeau reposted
Brian Lui
@brianluidog
So how come it's okay for the Singaporean government to do this, but not the AI safety people?
Kai
@teo_kai_xiang
I CALLED IT.
It's happening y'all. The singapore gov has launched a pilot programme for a gov-run dating service. Looks like the pilot is currently restricted to just government workers for now. -
David Cournapeau reposted
François Fleuret
@francoisfleuret
J'adore la réalité.
Les Echos
@LesEchos
Après la polémique sur le recours à l'IA, les ventes du livre de Thélyson Orélien bondissent ebx.sh/4yLRpD
-
David Cournapeau reposted
David Shor
@davidshor
The main thing is just that you want to have the broadest labor pool possible if you’re in a domain where you have right tailed distributions of applicant quality (h/t @robertwiblin).
For jobs that are more normally distributed nepotism is probably less of a big deal. -
David Cournapeau reposted
Jason
@sandarigi
AND the ability to effectively defend against them. Unlike the current frontier models that refuse.
-
David Cournapeau reposted
Matthew Yglesias
@mattyglesias
There are three million American households with net worth above $5 million derived primarily from ownership of privately held companies, and they collectively account for much more wealth than the Forbes 400.
www.slowboring.com/p/the-truth-about-rich-people-in-america -
David Cournapeau reposted
Boyd Kane (quantized)
@beyarkay
This new OpenAI model is kinda wild
-
David Cournapeau reposted
Doug Turnbull
@softwaredoug
Two things have changed in my thinking about LLM judges for search
I've long treated LLM as a judge as a pairwise decision. Deciding whether product A or product B is more relevant to the query "red shoes".
That's still true, but I have two new approaches in my toolbelt now -
David Cournapeau reposted
Philippe Lemoine
@phl43
Classic romcom type of situation where they start hating each other but then learn to know each other and fall in love.
CBS News
@CBSNews
“He happens to be a friend of mine,” President Trump said about North Korean leader Kim Jong Un when asked Tuesday why North Korea can have nuclear weapons but not Iran. “I’m about the only person in the entire world that he likes, and I like him.”
Trump then said, “As long as I'm around, he's going to be fine. You know why? Because he respects me and he likes me."This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Shubham Mishra
@ildivino_42
"Apple in China"
A few takeaways and my thoughts on supply chains, India opportunities and risks after reading the book.
In the last few months, a number of CEOs and Promoters of auto component companies encouraged me to read the book "Apple in China" by Patrick Mcgee to understand Apple’s overdependence on China, supply chain nuances, China+1 opportunities, risks of operating in China, opportunities for Indian companies, the entrepreneurial zeal of Steve Jobs and more.
I'm glad I did!!
Below are a few key takeaways from the book, along with my own thoughts on certain topics.
A) STEVE JOBS' ENTREPRENEURIAL ZEAL
A founder who refused to accept what manufacturing could do today (it resonated with what I had read in Elon Musk’s biography too). Jobs demanded designs that were, at the time, close to unbuildable: the original iPhone's switch from plastic to glass weeks before launch is a well-known example, and Apple's early Foxconn work involved tooling timelines far shorter than the industry norm (about 25 days against the usual 12 weeks). The friction between Apple's design team and its engineers, who had to prove something truly could not be done before saying no, kept pushing suppliers up the capability curve.
My thought: the brashness was the mechanism. But it worked because Apple also had the balance sheet, razor sharp focus, ability to change tack and stick to it when needed, the customer leverage and a supplier base willing to absorb the pressure. Intensity alone is necessary, not sufficient.
B) APPLE'S RISE IN CHINA, THE DEPENDENCE AND THE SQUEEZE
Apple's model was never plain outsourcing. As per the author, it owned about 7.3bn USD of machinery inside supplier plants by 2012, embedded engineers in over 1,600 factories, and often negotiated raw materials on the supplier's behalf. Suppliers earned low-single-digit assembly margins (2-3% for Foxconn-type work). Apple’s suppliers in China were ready to work for these thin margins and hence while Apple has ~20% of global smartphone shipments share by industry estimates, it has a staggering around 80% of industry profit share.
In one of the instances, one Apple employee himself agreed that – “which supplier would sign such a one-sided contract”. But suppliers actually did just to gain Apple’s business and then they would use that knowledge and gain margins from other customers.*
My thought: I think the Indian suppliers who have won orders from Apple, could be on somewhat similar competitive and low margins. It’s how the capability of the supplier is built over the years, which will be crucial for Indian suppliers to gain margins from other customers.
C) HOW CHINA’S GOVERNMENT TIGHTENED ITS GRIP
Many such instances are given in book when Apple from a position of dominance became increasingly ceding to a lot of Xi Jinping’s demands, as by then it had grown extremely dependent on China’s supply chain.
The book's conclusion lists five risks for Apple's China relationship this decade: (1) Huawei's return with HarmonyOS and improved features and tech capability likely to put pressure on Apple in China; (2) backlash from Beijing and Chinese consumers as assembly shifts to India; (3) the services business being lucrative for Apple and Govt could tweak regulations to make it unfavourable; (4) the AI push, constrained by China as local competitors have access to better data; and (5) tariffs and trade wars under Trump.
D) THE OPPORTUNITY: APPLE DIVERSIFYING TO INDIA – HOWEVER DEPTH OF LOCALISATION IS NECESSARY
India has gone from under 1% of iPhone assembly in 2017 to roughly one in four iPhones now. The supplier base has grown from 8 companies in 2018 to about 40. Apple runs several contract manufacturers here at once (Foxconn, Pegatron, Tata Electronics and a few others who are not yet in public domain).
But the book's caution is depth. India's iPhone output went from zero to about 15mn units in 2016-23, versus China's zero to 153mn in 2006-13, about a tenth of the pace.
Most work is final assembly with components flown in from China (known as FATP – final assembly test and pack out)
E) APPLE AND TESLA: THE SAME PLAYBOOK IN ELECTRONICS AND AUTOS
Both companies, in different industries, educated China's ecosystem, and that ecosystem then produced global competitors. Apple's suppliers helped Huawei, Xiaomi, Oppo and Vivo. Tesla's wholly owned Shanghai plant (the first of its kind in China, outside the JV rule) lifted local supply-chain content. A number of Chinese companies benefited from this know-how and eventually now Chinese brands now hold ~70% of China's car market, up from ~35% in 2020, while foreign brands fell from 65% to ~30%. The result threatens the very companies that helped build the ecosystem, as well as legacy global OEMs such as VW and GM. This was highlighted in detail showing graphs etc in our Thematic note “China Energising Seismic Shifts”.
My thought:
Learn from Apple as suppliers, but use that knowledge to diversify rapidly.
Full ownership of innovative global corporations in China taught the China ecosystem more than joint ventures did. The classic China auto JV kept product definition and core technology with the foreign partner, and transferred the least. Apple and Tesla, operating without JV partners, invested in educating suppliers themselves, and local entrepreneurs (BYD, Huawei, Xiaomi) became globally competitive. One instance in the book mentioned of Apple’s senior engineers sleeping in factories in China to fastrack production inturn helping the local supplier base and human talent. India could therefore consider conditioned access for Chinese OEMs, with local-content thresholds, local cell manufacturing and data safeguards, instead of viewing them only as competition.
The state matters as much as the founder.
Apple can neither leave China quickly nor stay comfortably, and that in-between period is India's opening. -
David Cournapeau reposted
World of Engineering
@engineers_feed
Polyethylene contains approximately 46 megajoules of chemical energy per kilogram (about ten times the energy content of TNT by mass) stored stably in its carbon-hydrogen bonds rather than as explosive potential.
The Oak Ridge molten-salt process announced this week is essentially an engineered way to release that energy as liquid fuel rather than letting it sit in a landfill. -
David Cournapeau reposted
Kevin A. Bryan
@Afinetheorem
This is a great hot take. Why hasn't econ been taken over by AI slop? Because field norms mean many publish ~1 paper per year and high quantity in low-ranked journals is seen negatively even pre-AI. Hence AI can lead to quality upgrading. This sounds right to me.
Ingar Haaland
@Ingar30
Econ is greatly positioned for the AI wave because the obsession with publishing a few killer papers in top journals leads economists to use AI for quality enhancement rather than to increase quantity. This is exactly what you want!
-
David Cournapeau reposted
Bruno Maçães
@MacaesBruno
It’s interesting that what Peter Thiel has been advocating as model societal decision making - ten people with no constraints from public or laws - was actually applied in the case of AI. Let’s see how that works out
-
David Cournapeau reposted
Alex Krivit
@ackriv
Gonna go plant a Merkle Forrest across the Internet
Cloudflare
@Cloudflare
As post-quantum signatures threaten to inflate TLS handshakes and certificate transparency logs, Merkle Tree Certificates offer a path to compact, auditable authentication. Cloudflare’s new certificate authority will support MTC issuance at scale. cfl.re/4yZ6GCj #birthdayweek
-
David Cournapeau reposted
Philippe Lemoine
@phl43
British government: "Please don't do war crimes and stop terrorizing Palestinians and occupying their lands."
Israeli guy: "The world is setting an impossible standard for us, I feel like we are back in 1933 😔"
Dan Senor
@dansenor
This is an astonishing speech. I don’t see a way for any Israeli government — either the current govt or a govt formed by the Opposition Bloc — to satisfy these new standards established and new lines drawn. Listen to every word here. These words are chosen carefully. We are, indeed, in a new world, which has much in common with a much older world.
-
David Cournapeau reposted
Bryan Bischof fka Dr. Donut
@BEBischof
Damn. All those data science tutorials are screwed.
Oceana
@oceana
BREAKING: A new penguin species has been identified for the first time in over 100 years! 🐧
Genetic research revealed that gentoo penguins are actually 4 distinct species, including a newly described species found on the remote Kerguelen Islands. people.com/new-penguin-species-discovered-for-first-time-… -
David Cournapeau reposted
yan
@bcrypt
what if instead of sandboxing the agents we sandboxed ourselves
-
David Cournapeau reposted
John & Margaret
@ukboomers
Tried using Chat-G-P-T to write a planning objection letter and I must say, I was quite impressed. Not just for cake recipes after all. Those two houses they want to put up on the derelict petrol station don't stand a chance now 🇬🇧
-
David Cournapeau reposted
Tudor Girba
@girba
Why are people so attracted to the idea of not having to think?
-
David Cournapeau reposted
Susan Zhang
@suchenzang
the same breed, who prides themselves on being aware and enlightened, but only for socially convenient friends
Susan Zhang
@suchenzang
-
David Cournapeau reposted
Jack Huynh
@jackhuynh
Special day!
Welcome @drfeifei and @theworldlabs to the @AMD family.
Fei-Fei taught machines to see. Now she’s teaching them to understand the world.
Super excited for what we’ll build next! -
David Cournapeau reposted
Branko Milanovic
@BrankoMilan
Homoploutocrats: The rich who also work hard
nehaveigur.com/2026/09/28/homoploutocrats/ -
David Cournapeau reposted
Pradeep Gowda
@btbytes
why plus or minus? it should have been 299 ++ 10
-
David Cournapeau
@cournape
Would be funny if Elon bought a tv network based on slop generated from his AI. Maybe people talking about AI being good at persuasion as X risk have a point after all
Jeffrey J. Hall 🇯🇵🇺🇸
@mrjeffu
Elon Musk is entertaining the idea of buying a Japanese TV network because the Nobunaga account that shares AI-generated slop about America being great -mixed with lots of anti-foreigner racism- told him Japan is in danger and needs to be saved.
-
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
The complete collapse of every international postwar institution and the advent of superintelligence were two exogenous and horribly timed events.
If AGI had come 20 years earlier on the tech tree, we'd probably have been pretty well set up to handle it as a civilization. -
David Cournapeau reposted
François Geerolf
@FrancoisGeerolf
"The euro area economy has proven resilient"
European Central Bank
@ecb
President Christine @Lagarde: Despite headwinds from the energy shock, the euro area economy has proven resilient with solid real GDP growth in the second quarter of 2026.
The outlook remains uncertain, with upside risks for inflation and downside risks for economic growth. -
David Cournapeau reposted
Philippe Lemoine
@phl43
The French government recently decreed that university cafeterias would have to charge only 1€/meal and, in a shocking turn of events, now students are being turned back after waiting in line because of shortages.
-
David Cournapeau reposted
Marc Andreessen 🇺🇸
@pmarca
"Fuck You Money" is mainly a myth. First, if you're in business, you pick up more responsibilities along with more money, you're responsible for more people, your actions have more consequences for others. Second, political power >> financial power every day of the week.
Zarathustra
@zarathustra5150
A surprising lesson of the last decade was how little “Fuck You money” seems to exist.
It was breathtaking to watch billionaires—especially in Silicon Valley—genuflect in slavish subservience to the woke crusaders, and enthusiastically provide a vigorous fellatio for a decade. -
David Cournapeau reposted
goodbye
@chasedbywasps
I hate spending money on a suit because what if I get extremely jacked in the next month and it doesn't fit anymore
-
David Cournapeau reposted
Renaud Foucart
@RenaudFoucart
Every European country facing a pension crisis looks like that now
-
David Cournapeau reposted
Gappy (Giuseppe Paleologo)
@__paleologo
Gentle reminder: if you can avoid it, don't invert that matrix (author: @JohnDCook, this came up in class).
www.johndcook.com/blog/2010/01/19/dont-invert-that-matrix/ -
David Cournapeau reposted
Alex Imas
@alexolegimas
Interesting that the music business has almost recovered to its 2000s peak but it’s all being driven by live performances. Streaming is tiny compared to the physical medium it replaced.
Joey Politano 🏳️🌈
@JosephPolitano
Can the arts business model survive AI? Perhaps the best guide is to look at how the arts survived past tech revolutions
Take music—internet piracy and then streaming killed the lucrative CD market, but musicians mostly saved their jobs by focusing much more on live concerts -
David Cournapeau reposted
Reply Guy Emeritus (tenured)
@JustinNakamoto2
Now you'll be able to live in the data center city. 😂
-
David Cournapeau reposted
Yossi Kreinin
@YossiKreinin
Not to detract from how retarded our messaging is, but our critics can't pick a lane, either
-
David Cournapeau reposted
Joel Grus 🤠
@joelgrus
one reason I am still an AI skeptic is *organizational illegibility*
it doesn't matter how big your context window is if all the necessary context is tacit -
David Cournapeau reposted
DHH
@dhh
This has never been more true. The opportunity cost of endless planning and contemplation just went way up. Agents allow you to just try vastly more stuff, so let them, and you'll learn way more, way faster.
Soul Notes
@SoulNotes43
“Action produces information. Just keep doing stuff.”
— Brian Armstrong -
David Cournapeau reposted
LindyMan
@PaulSkallas
What's crazy about socities like Saudi or the Gulf is the native people don't work. It's 50 million people living off oil money.
But it's not like they are creating great art. There's no interesting films. Songs. Literature. Games.
It's just guys hanging out all day
🇾🇪 Magd Alyami
@StonedOnQat
Not only are saudis unbelievably lazy, but most forms of work are heavily stigmatized in their feckless society. Recent examples include the major controversy over a saudi working as a security guard in Medina, and before that, a saudi working as a barista in Jeddah.
-
David Cournapeau reposted
Jun Song
@jun_song
New trend 2026 :
-
David Cournapeau reposted
Philippe Lemoine
@phl43
No, it was fucking stupid but then again so is Netanyahu’s argument, I just thought it was a funny joke.
-
David Cournapeau reposted
Aurora Britt
@Britt4rva
peak chud lasted 1 year
PatriotTakes 🇺🇸
@patriottakes
The Trump White House announced a Hispanic Heritage Month event after canceling their celebration in 2025.
The midterms are upon us. -
David Cournapeau reposted
Joel Grus 🤠
@joelgrus
here's the slides for my talk
"The Last Mile of AI Is Longer Than You Think"
at the Optimized AI Virtual Conference a few weeks ago
(talk track in speaker notes as always)
docs.google.com/presentation/d/1sKRduKH7nj3Ibb_N5Y-wUbn20… -
David Cournapeau reposted
Ethan Mollick
@emollick
Leaving aside the arguments over the reasons why this has happened, it is shocking that Europe does not have a single frontier AI lab, nor even a near-frontier lab nor even an effort that could likely lead to building up a frontier lab in the future.
-
David Cournapeau reposted
Joel Grus 🤠
@joelgrus
I was just talking about this yesterday
For the average computer user how is any software (Office? Gmail? Amazon? Zoom? Slack? Spotify? etc) better than it was a couple of years ago
("Excel now has a copilot button!")
Lewis Campbell
@LewisCTech
I still can't get over the complete lack of impact agentic coding has had on software from a user's perspective.
Our industry looks faintly ridiculous for waxing lyrical about the "agentic era" when it's had zero observable impact to the end user. -
David Cournapeau reposted
History Speaks
@History__Speaks
This is basically the end of academia. 18 year old students are told not to plagiarize, and punished if they do; while Cambridge refuses to investigate blatant, promiscuous, recreational plagiarism by a (sadly, deceased) senior faculty.
Steve McGuire
@sfmcguire79
Cambridge is no longer planning to investigate Jason Arday’s academic record or the university’s decision to hire or cover for him.
Instead, the investigation will focus on “his welfare, his care, his well-being” and “whether the school offered him sufficient support amid accusations that he plagiarized.” -
David Cournapeau reposted
Paata Ivanisvili
@PI010101
The talk by Tristan Buckmaster
youtu.be/tn45ZOp3lXM -
David Cournapeau reposted
Ferenc Huszár
@fhuszar
While everyone was busy asking Claude to write TLA+, we wrote up a tutorial on
- what it actually is
- why it’s not the formal verification silver bullet we are hoping for
- and what we have been working on ( to make a silver bullet out of it )
Complete with an interactive TLA+ playground!
reasonable.io/blog/tla-tutorial -
David Cournapeau reposted
Ben Dreyfuss
@bendreyfuss
I think BETTER CALL SAUL is actually better than BREAKING BAD
-
David Cournapeau reposted
Christopher Ingraham🦗
@_cingraham
So, a shipment of crickets for the lizard arrived via FedEx today. It was my first time ordering bulk crickets off the internet, and I naively assumed that they would be in like, a bag or some other contraption to facilitate easy transfer to another container. They were not.
-
David Cournapeau reposted
Crémieux
@cremieuxrecueil
Neat!
When gas stations learned that their competitors were charging more, they upped their own prices.
Price transparency was not good for consumers!
Nicholas Decker
@captgouda24
Chile required all gas stations to put their prices on the internet. The result? Margins went up 9%. Never forget that buyers are not the only people interested in what prices are! 1/
-
David Cournapeau reposted
Jacob Matson
@matsonj
Exec real time is actually once per day, many such cases
-
David Cournapeau reposted
Delip Rao
@deliprao
From a practical point of view, you should be using Jev today in your rubric-grading workflows (eval, RL rewards). The next release of Autorubric (landing tomorrow) will natively support Jev-style models in addition to regular LLMs.
Delip Rao
@deliprao
I spent some time investigating @typesafeai's Jev and striking to see them use a similar abstraction as autorubric -- released 8 months ago and to be presented at @COLM_conf next month. Our criterion types map 1:1 to Jev's Noul/Score/Choice: arxiv.org/abs/2603.00077
-
David Cournapeau reposted
Noah Smith 🐇🇺🇸🇺🇦🇹🇼
@Noahpinion
I swear this dude is the most Jewish-coded antisemite since Otto Weininger
Fuentes Updates
@FuentesUpdates
Nick Fuentes is asked how great he thinks the movement would be if he were 6’5
“You think if I were a 6’5 Chad, I’d be doing an internet show for a bunch of retards?” 😭This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Jacob Matson
@matsonj
lol can i one shot this
Fireside Alpha
@firesidealpha
Satya Nadella reveals his coding agent pulls every hyperscaler and neoclouds' SEC filings into a dashboard that refreshes every day for real-time ROIC
"And this is the other aspect of it, which is the enterprise context combined with the world's context. In fact, I go to the SEC filings of every cloud provider, hyperscaler, each of these neoclouds. It's in real time."
"I have a data runner in Fabric that brings all that data, puts it into a semantic model that then gets read by my coding agent and then surfaces it as a dashboard. And every day it's fresh."
"So I have the entirety of every SEC filing that goes out there, plus all of my internal analysis constantly coming together, giving me real-time ROIC by layer." -
David Cournapeau reposted
-
David Cournapeau reposted
Nando de Freitas
@NandoDF
I always believed in free education. I made a decision to never monetise my popular UBC and Oxford lectures. I’ve lived my life by working to bring knowledge to everyone on Earth: the pioneering lectures, the startups past and new, and my AI research. This desire came as the result of witnessing and experiencing a lot of suffering in my childhood. This course is part of my legacy and I’m very proud of it, and thankful to Oxford, the TAs and my students then.
mentalium
@mentalium_
My friend works at Anthropic.
He said they pay $650,000 a year for people who truly understand deep learning.
This exact course by Nando de Freitas is free forever.
Most AI courses charge $2,000+ for this foundation.
This one is completely free.
You get the real technical foundation ➜ taught at Oxford by a professor who later joined DeepMind.
Save this ⭣ -
David Cournapeau reposted
Philippe Lemoine
@phl43
I'm being racially profiled by a LLM 😂
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex
> Even if it doesn't follow deductively, how much more likely does the evidence make it?
The French can't understand this simple trick… -
David Cournapeau reposted
-
David Cournapeau reposted
hardmaru
@hardmaru
The last sentence in the Schmidhuber news trending item summary hits hard 🙃 x.com/i/trending/2103185481118744724
-
David Cournapeau reposted
Pulp libéral
@PulpLiberal
Il faut que les propriétaires payent des impots pour que l'Etat paye les locataires pour que les locataires payent les propriétaires.
J'ai une solution plus simple mais je présume qu'elle n'est pas assez française.
RMC
@RMCInfo
🗣️"Face aux loyers impayés, il faut créer un fonds pour protéger les propriétaires"
Julien Lebourgeois, propriétaire en Hautes-Pyrénées était l'invité de RMC
#apollinematinThis media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Chris Albon
@chrisalbon
Spending hours clicking through google results to find something that kinda answers your specific question in a passive-aggressive reply to some long forgotten drama on a discontinued software library's mailing list.
christina
@cszhu
what is the opposite of vibe coding? 🤔
-
David Cournapeau reposted
Armand Domalewski
@ArmandDoma
you know things are looking bad for Republicans when Jill Stein pops up in a non-Presidential year
-
David Cournapeau reposted
Brian Lui
@brianluidog
People blame Robin Hanson for hawking prediction markets while being blind to the obvious outcome. But is this fair? He knew this was likely. He just thought the huge benefit from more efficient Democrat spending in Iowa outweighed the Kalshi/Polymarket breakdown of society.
-
David Cournapeau reposted
Stephan Hoyer
@shoyer
This post isn't wrong, but why do people keep on posting slop on Twitter? Especially when complaining about AI!
Half the point of the short-form medium it should lower the effort to post content. (The other half is that it lowers the effort to read, too!)
Simon Maechling
@simonmaechling
I’m a scientist. I need to say this because the AI hype is getting ridiculous.
AI can design a molecule in seconds.
That doesn’t mean it discovered a drug.
It discovered something we scientists have never been short of:
Something to test.
Someone still has to make it.
Run the experiment.
Measure whether it works.
Check whether it’s toxic.
And ultimately prove it works in the real world.
AI hype tells us:
“Prediction is discovery.”
“Simulation is experimentation.”
“Generating a molecule is developing a drug.”
It isn’t. AI is making ideas incredibly cheap.
But every new idea creates something AI cannot generate: Evidence.
And the more hypotheses AI produces, the more experiments we’re going to need.
That’s the irony nobody seems to be talking about.
AI may not make laboratories obsolete.
It may make them more valuable than ever.
You can speedrun the thinking.
You can’t speedrun reality. -
David Cournapeau reposted
Mark Tenenholtz
@marktenenholtz
Using pandas over polars? In 2026?
clem 🤗
@ClementDelangue
Super happy to release SmolDataEnvs: 5,000 verifiable RL environment tasks for hill-climbing small models in code and data science by @adithya_s_k
100% open source: environments, evals, training!
huggingface.co/datasets/FineEnvs/SmolDataEnvs -
David Cournapeau reposted
Steven Sinofsky
@stevesi
I'm always a bit perplexed when people think that retail will quickly devolve into "just get me the lowest price and buy it."
There are so many forces at work from buyers, sellers, suppliers, wholesalers that work against buying by only low price, competing only on price, or even just searching for a combination of specs.
While it is definitely the case that the 1999-2000 web brought many low priced "shop bot" like products (Nextag was my fav) they all hit the same set of problems and challenges. There are many. Here's just one.
The addressable market is not the whole market and generally represents the least valuable part of the market. When the least valuable customers want something it is not worth the effort of the supply side to find ways to reach them.
Travel always seems to be the example, but out of the gate in the US just over half the $550B hotel rooms+ miles are paid for by business. As we all know these are preferred airlines with elaborate rebate and purchase programs. The reason these are offered is because of the spillover of these customers to leisure. Estimates are that up to 2/3rds of travel brand choice is dictated by those either using or compounding travel bonuses (miles, status, points, etc.)
The remaining travelers are a big business but not the customers that these same air/hotel partners aim for. That's the tiering of supply that we all see. But most people opining here might not recognize that Southwest, Clarion/Comfort, etc. follow these same patterns.
There are many price sensitive or price-exclusive purchases in all of retail. Many of these work both ways—for higher prices and convenience and bulk buying for discounts. When I need Oreos now I will pay more at 7-11. When I buy paper towels a few times a year I leverage my Costco membership and save 50%. When I want to load up on preferred brands of health things I buy lots at Walmart in hopes of not having to buy a one-off some night at Safeway for 2x as much. Everyone does this, regardless of income level. At another end, some people just buy everything by loyalty or convenience—if it's not at Costco we don't need it, or we have no problem buying everything at Safeway or Erewhan (never been.)
Many suppliers have no interest in consumers experiencing their product at wildly varying prices. They invest in brand, proprietary product advantages, certain product warranty or support policies, etc. These products actively cut off suppliers who undercut their pricing. They fiercely defend everyday "high" prices. This isn't just luxury goods but up and down price points and across categories.
The argument tech always makes is that technology can make the lowest price available and no one wants to lose the sale. Except businesses work of the 4Ps, and price is only one of them. Product (assortment), Price (bundles, discounting), Place (geo, service, loyalty), and Promotion (brand) and so on are all attributes. You might always want the lowest price but vendors are just not motivated to offer that.
Every major tech change brings with it disruptive forces in retail. It isn't surprising but often the tech change isn't what undoes a retailer but it is more often consumer preference and all the complexities of retail itself that prove too constraining for retailers to change.
Depending on the year, Amazon was going to destroy WalMart or WalMart didn't stand a chance against Amazon. The arguments were about price, supplier relationships, feasibility of delivery, same-day, loyalty, convenience, and more. Here we are today with two mega retailers.
There are patterns here worth noting. I wrote this ages ago when I can't even remember whether Amazon was doomed or thriving. medium.learningbyshipping.com/amazon-4648bdf511c -
David Cournapeau reposted
Federico Vaggi
@F_Vaggi
I think it’s more cynical than that: they realize the investigation would be embarrassing and this is a convenient way to bury it.
-
David Cournapeau reposted
Sylvain Catherine
@sc_cath
Les aides aux entreprises sont en fait des amendes prépayées donnant lieu à déduction en cas de bonne conduite.
Plutôt que d’avoir un taux de prélèvements obligatoires normal, on applique un taux insoutenable, puis on abaisse le taux effectif pour les entreprises répondant aux incitations fiscales choisies par le législateur.
Avoir un taux élevé et des aides est mathématiquement équivalent à avoir un taux normal et des sanctions pour les entreprises qui ne suivent pas le plan. Mais cela aboutit à une présentation politique du débat radicalement différente.
Jean-Marc Vittori
@jmvittori
Voilà de quoi tuer tous les débats sur les "aides aux entreprises" qui seraient autant de cadeaux scandaleux (graphe @Afep_ à partir de données @InseeFr). Non?
-
David Cournapeau reposted
Lucas Gortazar
@lucas_gortazar
1/ PISA 2025 says advantaged students fell more than disadvantaged ones.
Today we publish a paper explaining t's a measurement change. Measure students the same way in both years and the top-quartile drop in maths goes from −14 to −6 points. 🧵 -
David Cournapeau reposted
Philippe Lemoine
@phl43
Malheureusement, si les États-Unis veulent vraiment imposer leur droit à nos entreprises, ce n'est pas cette loi qui les empêchera de forcer celles-ci à obtempérer.
En effet, le gouvernement américain peut nous seulement couper l'accès au marché américain à une entreprise qui refuse d'obtempérer, mais même lui couper l'accès au système financier international parce qu'il contrôle le système de paiement en dollars et qu'il peut en couper à l'accès à toute banque qui accepte de traiter avec une entité sanctionnée par les États-Unis.
Nous avons vu au moment de la querelle entre Américains et Européens quand Trump est sorti unilatéralement de l'accord sur le nucléaire iranien que, même quand l'UE a pris des mesures pour essayer de protéger les entreprises européennes des sanctions secondaires américaines, elles ont quand même préféré quitter l'Iran pour cette raison.
Le seul moyen de nous prémunir contre ça serait que l'UE puisse de manière crédible menacer les Américains de répondre à ce type de chantage envers nos entreprises en se livrant à une véritable guerre économique contre les États-Unis, qui imposerait un coût significatif à l'économie américaine et donc un coût politique important au gouvernement, mais ça supposerait une volonté politique et une capacité de coordination au niveau européen qui n'existent pas.
Gilles Raveaud
@RaveaudGilles
Je découvre cette loi de "sécurité économique" de 1968, réactivée en 2002.
Elle permet aux entreprises de France de se protéger de demandes étrangères indues, comme celles des États-Unis, qui veulent imposer leur droit à tout le monde.
Fort intéressant ! @MCorcoral
_ -
David Cournapeau reposted
Benoît ✨ La Kufrérie 🏳️🌈🗽
@WeakTwink
RÉALISATION : Chaque année, le budget des retraites augmente de 2 fois le budget annuel de la Justice.
Vous avez bien compris : on pourrait tripler le budget annuel de la Justice juste en retardant d'1 an l'augmentation du budget des retraites. -
David Cournapeau reposted
Jacob Matson
@matsonj
We need to pace the frontier
Alex Godofsky
@AlexGodofsky
WHAT
-
David Cournapeau reposted
Subhamoy
@neo_subhamoy
JSON Statham 😭
-
David Cournapeau reposted
Caleb
@CalebSol
Still the best tweet
-
David Cournapeau reposted
Boomkat.com
@boomkatonline
A previously unreleased Popol Vuh score to Werner Herzog's 1974 doc 'The Great Ecstasy of Woodcarver Steiner', plus reworks from ML Buch, James K, Lorenzo Senni... 🫠
boomkat.com/products/the-great-ecstasy-of-woodcarver-stei… -
David Cournapeau reposted
Maurizio
@themgmtconsult
Boys and girls, nobody is thinking anymore.
Entire enterprise departments have thousands of people who are spending their days interacting with the LLM, faking "being productive", consuming tokens because it's a highly regarded activity, shipping lines of code with Cursor or Amp or Claude or whatever (except their brain).
Even slides are not made thinking anymore!
The person that allegedly built the powerpoint deck you are reviewing right now cannot even answer a couple of intelligent questions on the very thing he's presenting to you.
He looks at you stuttering some nonsense, hoping you'll let him off the hook because he made it so obvious Copilot wrote the slides (of course, you already knew that within the first 6 seconds of glancing at them).
There is some sort of comical atmosphere to it all.
And an absurd satisfaction in watching this entire situation unfolding in front of your eyes.
There is also approximately ZERO chance this whole thing won't spectacularly backfire, and of course good old consulting firms will be there to pick up the pieces.
I wonder at what point we started to believe we could get away with doing valuable things without even thinking?!
How did this happen, wow. -
David Cournapeau reposted
Bing Xu
@bingxu_
AVO (arxiv.org/abs/2603.24517) is accepted by NeurIPS 2026. Fully automation in GPU kernel engineering has crossed the point. The remained work is just scaling. It is a matter of time. Manual days will never come back.
-
David Cournapeau reposted
Sylvain Catherine
@sc_cath
I have a new working paper on how an economic slowdown tilts the wealth distribution toward older generations through the operation of pension systems.
With an empirical application to the European Union. -
David Cournapeau reposted
Takashi Ishida
@tksii
Japan has such a rich history in AI, and it's an exciting time to be doing research in this country.
Sakana AI
@SakanaAILabs
Sakana AI welcomes Jürgen Schmidhuber as Chief Scientific Advisor.
sakana.ai/schmidhuber/
Sakana AI is incredibly proud to announce that Jürgen Schmidhuber, universally recognized as the father of modern AI, is officially joining Sakana AI as Chief Scientific Advisor.
For nearly four decades, Jürgen has explored how machines can learn to learn. His foundational work in the 1990s drove core advancements in deep learning and established early frameworks for world models. Crucially, his pioneering innovations in meta-learning opened the very path toward recursive self-improvement.
These ideas have already shaped our own research, from the Darwin Gödel Machine to The AI Scientist. Now Jürgen will help guide our newly formed RSI Lab, whose objective is to trigger a compounding cycle of scientific discovery aimed at improving machine intelligence. We are assembling a critical mass of world-class experts in Tokyo to make this a reality.
Welcome, @SchmidhuberAI ! -
David Cournapeau reposted
Jürgen Schmidhuber
@SchmidhuberAI
Happy to join Sakana AI as Chief Scientific Advisor :-)
Sakana AI
@SakanaAILabs
Sakana AI welcomes Jürgen Schmidhuber as Chief Scientific Advisor.
sakana.ai/schmidhuber/
Sakana AI is incredibly proud to announce that Jürgen Schmidhuber, universally recognized as the father of modern AI, is officially joining Sakana AI as Chief Scientific Advisor.
For nearly four decades, Jürgen has explored how machines can learn to learn. His foundational work in the 1990s drove core advancements in deep learning and established early frameworks for world models. Crucially, his pioneering innovations in meta-learning opened the very path toward recursive self-improvement.
These ideas have already shaped our own research, from the Darwin Gödel Machine to The AI Scientist. Now Jürgen will help guide our newly formed RSI Lab, whose objective is to trigger a compounding cycle of scientific discovery aimed at improving machine intelligence. We are assembling a critical mass of world-class experts in Tokyo to make this a reality.
Welcome, @SchmidhuberAI ! -
David Cournapeau reposted
MTS
@MTSlive
Emergent Tokyo co-author @McReynoldsJoe says US-China trust may never come, but Beijing distrusting its own AI labs could be enough to get a deal:
"The trust isn't going to be there. But historically what's mattered the most has been, do you have an alignment on what you're both scared of?"
"AI safety research in China is lagging far, far behind, in terms of funding, prioritization, sophistication. So that shapes perceptions and RSI timelines wildly different."
"A Chinese lab, probably Moonshot, was passing on what Anthropic called Chinese military data, but I think was almost certainly MSS or MPS, to Anthropic for distillation purposes. Essentially handing state secrets over to the United States because the labs are in this very intense competitive frenzy."
"The lack of trust that frankly the PRC leadership should feel with their own labs could be one more helper for the US and China reaching some sort of agreement."
@StevenGlinert @RyanFedasiukThis media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Gregory Brew
@gbrew24
Details of Iran's offer--fleshed out today during Aragchi/Pezeshkian meeting with press and analysts--align with its negotiating position over the last several weeks.
US must return to MOU, end blockade, release frozen funds, end Israel war in S Lebanon.
In exchange, Iran will open Hormuz (on terms established with Oman).
It will also enter nuclear negotiations (again, something stipulated in the MOU).
Iran is offering to make this happen quickly, in a 7 day period, provided US reciprocates. -
David Cournapeau reposted
(((ل()(ل() 'yoav))))👾
@yoavgo
and then you sit on a grant committee, and one grant has a european reviewer writing "this is a fine proposal, i have only a few minor comments about the budget" and the other has a US reviewer "this is the best proposal i ever read in the past four months" and need to calibrate some how.
-
David Cournapeau reposted
Sakana AI
@SakanaAILabs
Sakana AI welcomes Jürgen Schmidhuber as Chief Scientific Advisor.
sakana.ai/schmidhuber/
Sakana AI is incredibly proud to announce that Jürgen Schmidhuber, universally recognized as the father of modern AI, is officially joining Sakana AI as Chief Scientific Advisor.
For nearly four decades, Jürgen has explored how machines can learn to learn. His foundational work in the 1990s drove core advancements in deep learning and established early frameworks for world models. Crucially, his pioneering innovations in meta-learning opened the very path toward recursive self-improvement.
These ideas have already shaped our own research, from the Darwin Gödel Machine to The AI Scientist. Now Jürgen will help guide our newly formed RSI Lab, whose objective is to trigger a compounding cycle of scientific discovery aimed at improving machine intelligence. We are assembling a critical mass of world-class experts in Tokyo to make this a reality.
Welcome, @SchmidhuberAI ! -
David Cournapeau reposted
Jo Kristian Bergum
@jobergum
wait - what if the markets learns about BertForMultipleChoice
Jaya Gupta
@JayaGup10
If the markets took DeepSeek seriously as a threat to frontier labs, they should be taking Jev (and Jev like-models) far more seriously. Here is why:
1. It's built for every developer, not just ML teams.
Until now, cutting model costs meant open-weight models: choosing a base, hosting it, fine-tuning it, evaluating it, and retraining it. That's an ML team's job, and most companies don't rlly have one. Jev is an API - it works on day one with no GPUs and no training.
2. It's so cheap that everyone is trying it, and everyone is suddenly discovering the waste.
At a fraction of a cent per decision, there's no reason not to test it. Anyone can paste the API key into Claude or ChatGPT and ask it to find savings, including non-technical people (so many talent teams are already using it) Curiosity is turning into discovery: companies see how much they've been paying flagship prices for yes-or-no answers. Teams that never cared about inference cost now do, including non-tech companies.
3. It hits every term of the frontier revenue equation.
Frontier revenue is (calls) × (tokens per call) × (price per token). Decision / Jev models pressure all three.
a.) Fewer calls. Frontier valuations assume the labs capture the agent boom: millions of agents making billions of calls. But most agent steps are decisions: which tool, is this done / safe / escalate / classifying / tagging /routing. Simple generation moves to cheaper and open models. Retries, wasted steps, and escalations disappear !
b.) Smaller calls. Routers strip stale context and unneeded tools, so the calls that still reach the frontier cost less.
c.) Cheaper calls. Once routers show cost per task side by side, flagship pricing on routine work doesn't hold.
4. It puts a router in front of and within every AI product.
Routers existed before, but they were hard to build and unreliable. Decision models fix both. They're the ideal router: fast, nearly free, and they report how confident they are, so the router knows when to escalate. Routers are now becoming a built-in setting in the tools developers already use, so companies get one without building anything.
Today, the frontier model runs the agent: it decides what to do next and calls tools. With decision models, ordinary code and cheap decisions run the loop, and the frontier model becomes a function the software calls when it needs heavy reasoning
5.) Every defense the labs have speeds it up. Price cuts confirm customers were overpaying. Shipping their own routers teaches customers to route, and makes it easier to route to competitors! Cheap decision tiers move the labs' own customers onto cheaper products and cannibalize revenue.
6.) They make post-training economical, and simultaneously increase the demand for it. One of the higher costs in RL on real enterprise work has been grading tasks without a checkable answer. A cheap, calibrated grader removes that cost. Environments become cheaper to run, router logs show which expensive workloads are worth replacing, and every team that discovered its overspend becomes a curious buyer. Companies that train, serve, and retrain custom models get cheaper inputs and a pipeline of motivated customers at the same time.
If DeepSeek made the market question the supply side: the cost of pretraining, the durability of the model moat, and how long a frontier model stays frontier, decision models make it question the demand side: how many calls actually need a frontier model at all..... -
David Cournapeau reposted
Sri Batchu🇺🇸
@sri_batchu
If you wanna know how frenzied the SF real estate market is, my realtor has a Stanford ML degree.
-
David Cournapeau reposted
Brian Lui
@brianluidog
It's, like... still uninvestible. They will produce enormous amounts of economic value, but you, the mighty and invincible outside passive minority investor, will capture none of it.
-
David Cournapeau reposted
Wendell Pierce
@WendellPierce
A mini reunion of THE WIRE last night as we gathered at the premiere of WAR, the new HBO series starring Dominic West and Sienna Miller @hbomax 25 years and counting….
-
David Cournapeau reposted
eric ゑリッ久
@shinobu_books
Remember in Hard to Kill when Steven Seagal wrote a grocery list with completely made up Chinese characters
William Han
@W_T_Han
Reminder that the 2007 film "National Treasure: Book of Secrets" presented xiaozhuan Chinese as ancient Native American writing hidden inside Mt. Rushmore.
-
David Cournapeau reposted
Bertille Bayart
@BertilleBayart
« Où est passé l’argent? » Question lancinante d’un pays à 120% de dette/PIB. Réponse dans les comptes sociaux, retraites en tête, peu à peu dévoilée grâce à l’opiniâtreté de @Beaufretjpascal et aussi de @philippejuvin @NicolasDufourcq @MxSba @sc_cath …
www.lefigaro.fr/conjoncture/ces-croises-de-la-transparenc… -
David Cournapeau reposted
alex rubinsteyn
@iskander
Took four photos of a whiteboard session and Astra combined, OCRd and vectorized them correctly
Weird new world -
David Cournapeau reposted
History Speaks
@History__Speaks
The Israeli Defense Minister says:
1) Israel razed all of Rafah and Beit Hanoun as a punitive measure;
2) they'll do the same thing to Gaza City if a single IDF soldier is captured.
You're an absolute rube if you believe that the Israelis razed everything beyond the yellow line because of "Khamas booby traps on every house." Their senior defense officials literally admit the point was to try to destroy Gaza. -
David Cournapeau reposted
Simon Willison
@simonw
I just had a voice conversation with the Datasette backup of my blog via datasette-mcp, which you can now talk to using the ChatGPT iPhone app
Vaibhav (VB) Srivastav
@reach_vb
Massive QoL update: ChatGPT Voice can now use plugins AND run on GPT-6 Astra, Sol & Luna, directly in ChatGPT Work on web and mobile.
Your email, calendar, Slack + docs, decks, sites and spreadsheets, just by talking.
Rolling out now, enjoy!! -
David Cournapeau reposted
Miguel de Icaza ᯅ🍉
@migueldeicaza
This is glorious
Dan Fessler
@DanFessler
What happens when you make your Operating System fractal?
I'll show you...This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Zach Weinersmith
@ZachWeiner
Artists: but I don't want to lose my job. I love it.
Mathematicians: but I don't want to lose my job. I love it.
Economists: LONG HAVE I AWAITED THIS HOUR! DESTROY ME, EFFICIENCY, AND LET UTILIZATION-ADJUSTED TOTAL FACTOR PRODUCTIVITY RISE ON THE WIND OF MY ASHES! -
David Cournapeau reposted
Elad Gil
@eladgil
Generally reason founders burn out - working on things you feel obligated to do but hate vs things you love
-
David Cournapeau reposted
rekdt
@rekdt
They call me 007…
I am very bad at Linux permissions -
David Cournapeau reposted
Brian Lui
@brianluidog
It's fortunate for the rationalists that no one is associating Kalshi and Polymarket with them. Normies are high decouplers, able to say, "Two bad instantiations of a concept does not mean that the concept itself is bad."
-
David Cournapeau reposted
Abigail Anthony
@abigailandwords
The full article did not disappoint.
The Boston Globe
@BostonGlobe
A lesbian bar in Greenfield eased its stringent mask mandate, a decision that has splintered the queer community in the city and beyond. www.bostonglobe.com/2026/09/22/magazine/greenfield-last-d…
-
David Cournapeau
@cournape
Bugs for you: insect vending machine in Tokyo
-
David Cournapeau reposted
zeynep tufekci
@zeynep
Think next step.
“There is this amazing, cheap restaurant with great food that’s never crowded. It was kinda hard to get to until recently but a new express bus goes there! So cool!”
Most non-legally mandated perks handed to the rare persistent, nagging customer will go away.
Ejaaz
@cryptopunk7213
Flight delayed 7 hours - asked Muse to file for compensation. 5 mins later i had $250 credit in my delta account. it even found and rebooked me a new flight.
it just figured everything out. even responded to the support email itself.
shit feels like magic. -
David Cournapeau reposted
Luol Deng Xiaoping
@JungAristotle
The Mexican food from this universe would go insane
Terrible Maps
@TerribleMaps
Something about this map feels wrong, but I can’t quite put my finger on it
-
David Cournapeau reposted
Sophie Page
@pophiesage
-
David Cournapeau reposted
Resident Advisor
@residentadvisor
A new @MarcelDettmann album is on the way. ra.co/news/86069
-
David Cournapeau reposted
Vertumne 🌞❄🇪🇺
@Vertumne1
Analyse toddienne typique : du freudisme mâtiné de poncifs. "Les jeunes font les révolutions" (celle de Maidan a eu lieu dans un des pays les plus vieux d'Europe), "la gérontocratie française est terrible" (alors que le Japon de famille souche nous pulvérise dans ce domaine), etc
-
David Cournapeau reposted
eric ゑリッ久
@shinobu_books
A 1997 guide to compensated dating
“Men and women at the end of the century”
“How girls heavily involved in compensated dating see it in their mind’s eye”
“If you handover money, you’ll get arrested! However, what about presents?”
For sale used on Amazon for $180 -
David Cournapeau reposted
marimo
@marimo_io
We were thrilled to get early access to @typesafeai's Jev model and explore tons of new use cases!
Here's what @ktaletsk is building with Jev:
1. jevframe - pandas and polars dataframe adapter to bulk generate new columns with Jev github.com/ktaletsk/jevframe
Try the notebook: -
David Cournapeau reposted
JohannesBorgen
@jeuasommenulle
Oui le calcul des 85 zéros avant la première décimale non nulle est vraiment grotesque 🤣🤣🤣 (pareil aucune idée si c'est de l'IA, certains passages que j'ai vu passer y font beaucoup penser mais c'est pas exactement une preuve)
-
David Cournapeau reposted
Corti (Cortiste)
@cortisquared
Cela dit, parmi tous les postes d’économies, la baisse des retraites est le moins destructeur et de loin.
David Cayla
@dav_cayla
Bien sûr, ça va passer crème une baisse des pensions de 25% et ça ne va pas du tout affecter le pouvoir d'achat et la consommation des ménages. Non non, tout ira bien, c'est sûr!
Paul & Mick-Victor
@Lecoqsportif17
Non. On peut envisager simultanément une baisse des pensions, significative, approx -25% (déficit actuel du système entre cotisations et versement), couplé à +3 ans de la durée de cotiz. Un tel mouvement permettrait d'affecter 10% des cotisations actuelles à un schéma de capi.
-
David Cournapeau reposted
Ken Ishii ケンイシイ
@K_Ishii_70Drums
😄 @UltraJapan 2026 バックステージ。@RisaTaniguchi とのB2Bは大成功!あの雨の中盛り上がってくれたみなさんに感謝!
-
David Cournapeau reposted
Laurent Garnier
@LaurentGarnier_
"Plus personne ne doit cracher au bassinet : ni les retraités"
"Elle a un problème de dépenses publiques"
À un moment donné il va falloir arrêter de se foutre de notre gueule, parce qu'on va continuer à informer et ça va finir par se voir, même des plus bornés. -
David Cournapeau reposted
Tom Sydney Kerckhove
@kerckhove_ts
There's this chart of birthing positions at the hospital but it just looks like a fat lady trying to get through the worst hangover of her life.
-
David Cournapeau reposted
Isaac Flath
@isaac_flath
"In practice, this all means that evals and structured experiments are even more important than they are for regular LLM projects."
💯
Simon Willison
@simonw
Put together some notes on Jev and the new category of system one aka decision models simonwillison.net/2026/Sep/21/jev/
-
David Cournapeau reposted
saila (he/him)
@sailaunderscore
When you rely on LLMS: braindead slop-cannon
When I rely on LLMs: on the bleeding edge of cognitive offloading -
David Cournapeau reposted
Anthony Morlet-Lavidalie
@AnthonyML_
La hausse du taux d'épargne des ménages en France est surtout et avant tout le fait des personnes de + de 65 ans. Les jeunes actifs ont plutôt désépargnés.
Demander aux retraités de contribuer au redressement des finances publiques est légitime et nécessaire. -
David Cournapeau reposted
Massimo
@Rainmaker1973
Bro put AI into a Big Mouth Billy Bass and now the fish gives him motivational speeches to get him to work out.
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
JosH200
@josh_wills
Keith is the best engineer I ever worked with and we got to hit on all of my favorite topics in this conversation: pure math, chess, engineering, and what all of this means for the kids.
Pebblebed
@pebble_bed
The clankers have the keyboard. Rather than mourn the transition, we find it exhilarating.
Announcing After Coding, a new Pebblebed podcast with @keithmadams. Episode 1 with @josh_wills is live.
link >>This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Dan Castellano
@ninja_padrino
Today marks a record 25 consecutive days of rain in Tokyo - not to jinx it but tomorrow appears clear🤞🏻
Gearoid Reidy リーディー・ガロウド
@GearoidReidy
If you thought this never-ending rain in Tokyo was unusual, you are correct: 21 days of rain in a row is a first in 150 years of records
-
David Cournapeau reposted
Chris 𝕏
@Chris__X__
I never realized jogging in your 50s would be a great way to meet new people.
Just today, I met two nurses, two EMTs, and a cardiologist.
I was so close to meeting Jesus too. -
David Cournapeau reposted
Robin Rimbaud - Scanner
@robinrimbaud
Amazing to hear what is probably the earliest live recording of British electronic beat combo Autechre, in Rochdale for Darrell Fitton's birthday party 1990
youtu.be/M70jmp-RCeY -
David Cournapeau reposted
Ryan Kennedy
@RyanKennedy_22
Well that's not good...
-
David Cournapeau reposted
Steven Strogatz
@stevenstrogatz
Things are heating up on Terry Tao’s blog. In “If Math Is More Than Proof, We Need to Better Celebrate the Rest of It,” Grant Sanderson of @3blue1brown proposes “open exposition problems” -- rewarding the work of making math genuinely understandable.
terrytao.wordpress.com/2026/09/18/if-math-is-more-than-pr… -
David Cournapeau reposted
Sebastian Raschka
@rasbt
It’s easy to dismiss Jev it as “just a classifier”.
But people (me included) who have been training encoder-style models for classification for many years know they were usually special-purpose and limited in some way.
The breakthrough of Jev is that it generalizes well.
And I’d say the secret sauce is probably more in the data than in the training algorithm. (Plus a nice API design on top of it.)
Kun Chen
@kunchenguid
poor guy claim to have built Jev a year ago but no one cared, and now Jev stole all the thunder
many people are saying “you gotta tell your story” or “marketing is important”, and they just completely missed what actually made the difference here
i just looked into this laya model laya.convaiinnovations.com and:
- it only supports 512-1k context… a lot of use cases won’t fit at all
- evaluating the model directly shows its accuracy is as good as a coin flip. in order to get good results, you need to first fine tune it
i’m sorry, but that’s not Jev
there’s a massive gap between an interesting research and a useful product
you can “tell your story” all you like, but you can’t blame Jev for stealing your thunder when Jev did all the work to make a well-packaged solution anyone can just grab and go
Jev is not completely new from an academic sense, just like how ChatGPT was not the first LLM
don’t underestimate the effort and value in putting together something that’s actually good enough for adoption - it makes all the difference -
David Cournapeau reposted
Hollis Robbins
@anecdotal
AI sheds light on:
- vulnerability to verbal manipulation
- tendency to swarm behavior
- weakness in internet safety
- vulnerability to scams and deepfakes
- tendency to outsource thinking
- desire to be waited upon
- desire for "automatic" education
Hollis Robbins
@anecdotal
A real contribution of AI is showing the critical weaknesses of humans and human structures.
-
David Cournapeau reposted
Visakan Veerasamy
@visakanv
-
David Cournapeau reposted
Jinjing Liang
@JinjingLiang
Jev
-
David Cournapeau reposted
zeynep tufekci
@zeynep
And y’all laughed when people pointed out that generative AI was NOT going to make Big Consulting go bankrupt, not because LLMs can’t produce very similar reports but because the key function of many such companies is providing legitimacy and a social stamp to existing choices.
Anthropic
@AnthropicAI
We’re partnering with Accenture on independent evaluation of frontier AI—part of our recent commitment to embed evaluators at Anthropic. Both we and Accenture expect to invest at least $1 billion to build capacity in this area over the next five years. www.anthropic.com/news/accenture-embedded-evaluation
-
David Cournapeau reposted
nazo
@nazo_btw
-
David Cournapeau reposted
Nicolas 💸
@NicolasQuiPaie
-
David Cournapeau reposted
La vida en viñetas
@lavidaenvinetas
Bauhaus Milhouse
-
David Cournapeau reposted
Khoa Vu
@KhoaVuUmn
AI is finally taking over economist's job. It's all over.
Financial Times
@FT
AI chatbots give wrong answers to financial queries ‘most of the time’ ft.trib.al/KxRXFO6
-
David Cournapeau reposted
引きこもり投資家/投資収益5000万到達!
@JO0sthSYpVRt2hn
定時退社を目指すゲーム。
-
David Cournapeau reposted
Nostalgia
@NostalgiaFolder
IT security in 1990s
-
David Cournapeau reposted
Thorsten Ball
@thorstenball
Most predictions I see are still way too conservative.
Here's mine -
David Cournapeau reposted
eric ゑリッ久
@shinobu_books
Bloody hell, Matsuri were still off the hook into the 1990s
-
David Cournapeau reposted
Kadi🇪🇪🌻
@TheFl0orIsLaVa
"In Europe, free speech is in retreat"
- JD Vance
Meanwhile, back at the White House...
NewsWire
@NewsWire_US
TRUMP BANS CNN, MS NOW, POLITICO FROM WHITE HOUSE
-
David Cournapeau reposted
Sajid Mehmood
@smehmood
If you’re wondering how Jev is scaling to meet insane demand they could not possibly have predicted, the answer appears to be @modal — one of only 4 subprocessors listed on trust.typesafe.ai
Vercel
@vercel
Jev was adopted faster than any other model in AI Gateway history.
In the first day, @typesafeai reached ~13% of teams, 2x the GPT-5.6 family and 6x Fable 5.1. -
David Cournapeau reposted
Bia
@BeatrizGietner
lol the thumbnail
FT Alphaville
@FTAlphaville
‘Greek finance minister encourages Germany to implement reforms’ ft.trib.al/VfxPA4d
-
David Cournapeau reposted
Linear Labs
@LinearLabs
The original Akai MPC training video featuring the machine's inventor, Roger Linn. Linn's machine changed music forever ultimately building the blueprint for hip hop and sampling.
-
David Cournapeau reposted
Is This A Spoiler?
@IsThisASpoiler
Fake because I don’t believe that man would’ve been wearing a shirt at any time
-
David Cournapeau reposted
Philippe Lemoine
@phl43
Sure, that's not ideal, but at least French cities are now safe from Iranian nukes 😪
Faytuks Network
@FaytuksNetwork
BREAKING: Saudi Arabia will supply NO crude oil to European refiners next month after a drone attack forced the shutdown of its key East-West pipeline, Bloomberg reports. The supply cut reportedly applies to all European customers with long-term contracts.
-
David Cournapeau reposted
John D. Cook
@JohnDCook
One thing that struck me while working in biostatistics was that every clinical trial had a convincing explanation for why it should succeed, though most did not.
That's why we have to run trials.
2/2 -
David Cournapeau reposted
John D. Cook
@JohnDCook
Consider the statement "This chemical cures this cancer in mice, so it should cure the same cancer in humans."
It's a reasonable statement, but it has to be tested.
Science can't stop after making reasonable predictions. It has to test things.
1/2 -
David Cournapeau reposted
cash4turtles🌸☢️
@WaifuverseAI
Once the stock market is 247 America will be totally unable to wage war due to the stock market consequences. A new era of world peace is on the horizon.
Solid Intel 📡
@solidintel_x
INTEL: U.S. stock markets will expand to 23-hour weekday trading on Dec. 6
The SEC says full 24/7 trading could follow -
David Cournapeau reposted
Boomkat.com
@boomkatonline
Ron Trent
Primitive Arts
oh man, prayers have been answered with this one, the definitive album from one of the best to ever do it, finally coming back to wax ✨
boomkat.com/products/primitive-artsThis media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Sylvain Catherine
@sc_cath
Le système par répartition n’est pas mort, tout simplement parce qu’il remplit une fonction dont on ne peut pas se passer : le roulement de la dette sociale implicite.
Les retraités comme les actifs ont acquis, par leurs cotisations passées, des droits à pension. On peut les ajuster à la marge, mais on ne peut pas simplement les annuler. Ces engagements constituent une dette qui, parce qu’elle ne prend pas la forme explicite d’obligations émises par l’État, est qualifiée d’implicite. Elle représente de l’ordre de 400 à 500 % du PIB.
Pour honorer ces engagements, on prélève aujourd’hui des cotisations qui servent à payer les pensions promises. Mais ces cotisations créent à leur tour de nouveaux droits, qu’il faudra demain financer par des prélèvements sur la génération suivante, et ainsi de suite. Autrement dit, la répartition remplit une fonction essentielle : elle permet de rouler cette dette implicite d’une génération à l’autre.
On peut parfaitement introduire, en parallèle, un pilier par capitalisation. Mais cela ne fait pas disparaître la nécessité d’honorer les droits déjà accumulés, ni donc de rouler la dette implicite. À moins de trouver quelque 500 points de PIB sous un arbre pour solder d’un coup l’ardoise de la répartition, nous continuerons donc à avoir besoin d’un mécanisme permettant de la refinancer d’une génération sur l’autre.
BFM
@BFMTV
🔴 EN DIRECT
"Le système de retraite par répartition est mort": Éric Zemmour prône la "capitalisation"
l.bfmtv.com/RPlx -
David Cournapeau reposted
David Shor
@davidshor
Something that Jensen and A16z and the major funders of LTF all have in common is that they have obviously directed very large sums of money directly to the president's pocket.
This is not something you should do if you want the public to trust you on a complex issue! -
David Cournapeau reposted
Éros Brousson
@erosbrousson
I NEED HELP
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
> join the sf tech Jew chat.
> it’s all Asian women. -
David Cournapeau reposted
Arvind Narayanan
@random_walker
Over 2 years ago @sayashk and I wrote a detailed deconstruction of p(doom) and argued that its primary function is to launder vague, evidence-free intuitions and fears through a facade of quantification. It remains 100% relevant today. www.normaltech.ai/p/ai-existential-risk-probabilities
Itai Sher
@itaisher
There is a strange laundering of this number. It was floated by Anthropic employees without public justification. Then mathematicians pick it up and now it is attributed to their judgment.
It’s fine to survey AI practitioners, but posts like this are a bad epistemic practice. -
David Cournapeau reposted
-
David Cournapeau reposted
Federico Vaggi
@F_Vaggi
You can make a plausible case that there's a causal link between NFT and Trump, since the n1 reason why several silicon valley VCs went all in on Trump is because they wanted to be able to unload shitcoins/NFTs on imbecilles without the CFPB getting in the way
Matthew Zeitlin
@MattZeitlin
this single tweet is probably responsible for a nontrivial share of the skepticism about ai capabilities
-
David Cournapeau reposted
Hector Yee
@eigenhector
Ironically the ML paper "Attention is all you need" gave rose to AI that sucks up all available attention
Lucas Beyer (bl16)
@giffmana
At this point, "researchers" are basically DDoSing classic academia. It's pretty sad to see. Either someone comes up with, and executes, a CloudFlare for academia, or it's toast.
And the only thing i can think of that might have a chance to scale is more automation combined with a credit/point system like we've discussed a few times here on X over the years, but it needs simultaneous buy-in and coordination from all ~10 major AI conferences and journals, so it's a heavy lift.
But if nothing big happens, i think it's game over soon. -
David Cournapeau reposted
Le Zelenskyste 🇫🇷🇺🇦
@VolodimirZelen1
🚨 Yémen : les houthis capturent un drone
C'est Carré 🟩
@Cest__Carre
🇸🇦 -1 F-15 saoudien, abattu par les Houthis (Ansar Allah).
-
David Cournapeau reposted
Doug Turnbull
@softwaredoug
My buddy @hugobowne is doing more agent building fun in his upcoming workshop :)
maven.com/hugo-stefan/build-ai-agents-from-first-principl… -
David Cournapeau reposted
Philippe Lemoine
@phl43
Realistically, there was probably no way the debate about AI was not going to end up being polarized in today’s environment, because everything becomes polarized in that environment once it becomes politically salient.
It would be funny if the machines would not have killed us, but ended up deciding to do it after all because, upon listening to us debating whether they would kill us, they realized that we were just too retarded to justify spending resources on us. -
David Cournapeau reposted
John Myles White
@johnmyleswhite
A rare technology that offers a chance to bet on a clear forking path:
1. Noise model improves faster than transcription models, permanently staying out of their reach.
2. Transcription models get trained on noise model output and new noise models never work again.
Aida Baradari
@aidaxbaradari
Today, we’re releasing Kalypta, the first app to block AI notetakers in your meetings.
Granola? Wisprflow? Cluely? No more.
With Kalypta, you become inaudible to AI.
Your call continues normally.This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
eu/acc
@euacchq
Europe is very lucky we have strong European founders who care about its future like @stripe ’s @patrickc .
Patrick & his great team dedicate their time and skills to gather data and help fix things.
Help him help Europe by answering this survey 👇🏻
PS : @atomico - a VC founded by European entrepreneurs also, does a survey too, link in comment, please do it too!!
Repost!
Patrick Collison
@patrickc
A quick EU founder survey: eusurvey.vercel.app.
-
David Cournapeau reposted
Steven Sinofsky
@stevesi
My friendly reminder, after Microsoft software caused billions in damage due to benign features being chained together (ILOVEYOU) and then a bunch of Windows Server ecomm apps went down over holidays (and more), Microsoft just said:
"So now, when we face a choice between adding features and resolving security issues, we need to choose security. Our products should emphasize security right out of the box, and we must constantly refine and improve that security as threats evolve. A good example of this is the changes we made in Outlook to avoid e-mail-borne viruses. If we discover a risk that a feature could compromise someone’s privacy, that problem gets solved first. If there is any way we can better protect important data and minimize downtime, we should focus on this. These principles should apply at every stage of the development cycle of every kind of software we create, from operating systems and desktop applications to global Web services."
Full text: "Trustworthy Computing" January 2002. archive.is/6T7hl
We did this on our own. We did not ask for regulators to tell us. We had the data. We knew the issue was in the software. It was super difficult in the competitive and customer-demanding world we worked but we did it.
It is so confusing to me that the AI companies are acting so helpless and in need of oversight to prioritize making the right product. What do they think regulators would know about building their products that they do not know?
How much more evidence do we need? -
David Cournapeau reposted
Nicolas Colin
@Nicolas_Colin
Thanks for this—
I must mention that I’ve never used Stripe Atlas, so perhaps I lack experience with a much simpler incorporation workflow.
However, I DO have direct experience incorporating, managing, and dissolving both corporations and partnerships across the following jurisdictions: the UK, France, Germany, Estonia, Luxembourg, the US, Hong Kong, and the Cayman Islands.
My experience is that NONE of these countries makes the overall process particularly light when you look at everything from incorporation to taxation to other things to dissolution. And in most cases, the burden is simply offloaded to highly paid professionals who wear different hats: lawyers, accountants, tax advisors, and other professional service providers.
In addition, your initial question, Patrick, was about scaling, not incorporating, and as an officer or manager of multiple entities that hold shares in US companies that scale, I see a LOT of lawyers’ work along the way, even just on the corporate side. I expect there is much more on the tax, labor, and regulatory sides. (Not to mention dealing with employees who sue for harassment and demand huge settlements, which I hear is a uniquely American feature.)
I would assume Stripe has paid a lot of specialized lawyers along the way when it comes to payments, compliance, and financial services in general, and on that front I would be surprised if things were more difficult or complex in Europe than in the US. -
David Cournapeau reposted
Patrick Collison
@patrickc
Having conducted a lot of business in both Europe and the US, this just isn't true. I've waited months for basic things to happen in Europe despite being willing to spend very large amounts on lawyers. And lots of things in the US (like incorporation) have minimal legal fees.
-
David Cournapeau reposted
Nicolas Colin
@Nicolas_Colin
For the record, I think this whole debate about Europe making it difficult to scale new businesses is completely off.
Scaling a new business is a nightmare in the US too, as it is everywhere else on Earth. That is pretty much what scaling a new business means: finding increasingly creative ways to be miserable.
There are two things that make it easier in the US, though:
1/ You can raise much more capital, all else being equal, for macroeconomic reasons.
2/ You can use that capital to pay lawyers to deal with all the problems for you, turning “this is an absolute nightmare” into “this is an absolute nightmare that costs $500k”.
Cc @SineadOS1 @misssaxbys @johnthornhillft @michaelxpettis
Patrick Collison
@patrickc
I'm curious to read other stories from people who've scaled new businesses in Europe. (And if your experience has been positive, that's also useful to hear.) Feel free to email me: patrick@stripe.com.
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.
-
David Cournapeau reposted
Robin Hanson
@robinhanson
Recent events show why the world mostly neglects long term relative to short term problems.
Once you seem open to considering long term problems, you get a blizzard of theory arguments re purported problems & solutions, most of these in the service of hidden agendas. It then takes decades to see just how wrong most were.
With a focus on short term problems, you quickly learn which of the solutions you've tried actually help, and anticipating this people offering theories of problems and solutions are cautious against over-claiming.
We can only really address long term problems when we have very strong theories to identify problems & solutions. Usually theory is weak, forcing us to focus on concrete data, and hence the short term.
Yes, I've often focused on long term problems, and this may help explain why the world wisely ignores me. -
David Cournapeau reposted
Canap ✨🛋️💿
@CanapDirect
Jusqu'en 2011 on pouvait déduire des impôts les intérêts d'un crédit immo, même pour une résidence principale.
-
David Cournapeau reposted
anon
@anon597260576
Tldr
JohannesBorgen
@jeuasommenulle
Europe elects has the most hilarious charts about French presidential candidates - so let me present them to you ! A thread. 1/N
-
David Cournapeau reposted
Brendan Karet
@bad_takes
witnessing previously unimaginable levels of maga boomer cope
-
David Cournapeau reposted
Philippe Lemoine
@phl43
Iran is apparently the only country that takes AI safety seriously.
Reuters
@Reuters
Amazon's AWS is unable to restore access to Bahrain, one UAE cloud data zone after war damage www.reuters.com/world/middle-east/amazons-aws-is-unable-r… www.reuters.com/world/middle-east/amazons-aws-is-unable-r…
-
David Cournapeau reposted
a newsman
@a_newsman
unusual_whales
@unusual_whales
BREAKING: 30-year mortgage rate has hit 7.17%, per MW
-
David Cournapeau reposted
-
David Cournapeau reposted
Philippe Lemoine
@phl43
They were in the middle of negotiations about it, there was no urgency to start a war now and every reason to think it would turn into a shitshow. It’s quite possible that it will make it harder rather than easier to prevent Iran from developing nuclear weapons in the end. The only reason why it happened is because Trump is a moron who thought he could magically force the Iranian government to do whatever he asked by a short use of force. There was nothing more to it than that and it was completely stupid.
-
David Cournapeau reposted
Maxime Sbaihi
@MxSba
Quand tu rappelles au président de l'Union Française des Retraités que:
👉les pensions ont été revalorisées 6x en 4 ans
👉les salaires ont moins augmenté que les pensions depuis la crise Covid
👉l'abattement IR de 10% des retraités n'a aucune justification sociale ou économiqueThis media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Sakana AI
@SakanaAILabs
At Tech Summit '26 in Christchurch, Sakana AI Research Scientist @Stefania_Druga spoke in front of ~700 industry leaders. She explained why we are all scientists now, why that makes human expertise matter more than ever, and why AI for Science is a Sovereign AI question.
-
David Cournapeau reposted
Carlos
@alg0agent
CausalSmith - An agentic pipeline that automates research in Causal Inference
jiyuan-tan.github.io/CausalSmith/ -
David Cournapeau
@cournape
The distinction is important but it is absolutely not obvious that ASI means wiping the humanity out.
The analogy of grandmaster is flawed, because it implies a well defined ‘game’ with clear rules and objectives. But real life is not like that
Yishan
@yishan
Maybe you have recently become aware of the AI safety debate and the arguments swirling around it.
If you want to understand them, you need to understand a couple things that almost everyone gets wrong:
There are TWO distinct classes of AI dangers, and it's VERY important to think about them separately, and not let one confuse you about the other.
Many many people (including many quite intelligent, clear-thinking people) do not effectively understand the fundamental differences between these two classes of dangers.
The first one has to do with the theory that an AI far superior to human intelligence (Artificial SuperIntelligence, or ASI) will inevitably wipe out the human race.
The second one has to do with the idea that powerful AI will result in very harmful things happening to many human beings, possibly all human beings.
Those two sound VERY similar, don't they?
They are DIFFERENT. Understanding how they are different is crucial if you want to think about or contribute usefully to any conversation about AI safety or AI harm. You might feel like you are Making Very Good Points or Asking Incisive Questions, but if you aren't clear on the differences between the two, you aren't.
So, I'm going to tell you what the difference is so that you can talk more usefully.
The first one concerns itself with a very specific thing, which is ASI (Artificial Superintelligence) that is more intelligent than any human being. When I say that, I am not referring to a thing like how Einstein is smarter than you, we are talking more about something like how a human being is more intelligent than any mouse.
In our regular lives, we meet other people who we can tell are smarter than us, vs some who are less smart. The line is fuzzy, because intelligence has a lot of dimensions. I'm better at a "rotating shapes" kind of intelligence than my wife, and she is better at "words-making" kind of intelligence than I am.
But every human is better in almost every dimension of intelligence than every single mouse.
That's the level we're talking about: an artificial superintelligence - made up of a computer or a network of computers - that is more intelligent than any human. And more intelligent by a long shot, by a wide margin, in an indisputable way like how humans are above mice.
That is the first thing.
The theory says that if you have an AI that is vastly smarter than all humans - in the way that a human is smarter than mice - that superintelligent AI will inevitably, eventually, sooner or later, wipe out every human on the planet.
We will refer to this as "existential risk."
The common follow-up question "well, how exactly is it going to do that?" is NOT the important question, and one of the most important elements of understanding this theory is first getting why that particular question is not important. A couple analogies:
Analogy 1: You are playing chess against a grandmaster. My theory predicts the grandmaster is going to beat you. You can ask "Well, how exactly is he going to do that?" I don't know, because I'm not a grandmaster, I just know that a chess grandmaster is almost always going to beat a normal player like you. And I'd be right. So the question "how is he going to do that" is not important, and doesn't affect the final outcome. He's going to figure out a way because he's way better than you.
Analogy 2: Humans are smarter than all other animals, comprehensively, by a wide margin. We have driven numerous species to extinction, not because we hated them or hunted them. Many of them have died out without most humans even ever thinking about them. All we did was expand our civilization, use up resources, encroach on habitats, and pretty soon the resources needed by those species went away and they died out. We figured out a way to get what we wanted because we're way smarter than them, and often we didn't even notice they died as a result.
A lesser animal asking, "how are the humans going to wipe us out?" is not asking a relevant question. We don't know, but we do know that any time humans and lesser species compete for any kind of resources, the humans will win. The fact that we know who is going to win beforehand - and that it is due to the vastly different levels of intelligence - is the key concept here.
A vastly more intelligent AI is likely to care about things that are incomprehensible to us, the way animals can't understand human goals. It's going to need resources to pursue those goals and it's going to be far more effective at gaining control of them and excluding us from them - in the same way that we are far more effective than other lower species.
A much more intelligent AI will not care about our interests, it will care about its interests, and to whatever small degree we happen to escape total annihilation from losing access to all our resources, any remaining humans will likely be enslaved into a system that serves the AI's own purposes.
That is the first thing. (Remember how I said at the beginning of this post that there was a first thing, and then a second thing?)
The first thing is the most difficult to understand, because you have to extrapolate how a vastly superior intelligence would act, and you can only use analogies like "how do humans treat lesser creatures," and the analogies are messy.
But now let's move on to the second thing.
The second thing is "everything else you've ever heard that AI might do that's harmful."
That's a little inaccurate. It's actually "everything else you've ever heard that humans might use AI to do that's harmful."
This is the critical difference. The first one talks about the inevitable outcome of what happens when two vastly different levels of intelligence collide, e.g. ASI vs humans, or human vs mice.
The second one has to do with what happens when humans possess AI as a powerful tool. This second thing is much easier to understand, because we have many more concrete notions:
Like:
- the military uses AI to make hyper-efficient killer drones and missiles
- your capitalist overlords use AI to replace you and everyone loses their jobs
- authoritarian government uses AI to surveil everybody and control the entire population
- hackers use AI to break into secure networks and hold companies and governments hostage
- students use AI to cheat on homework and show up to college knowing nothing
- AI slop saturates the internet and makes it impossible for artists and writers to make a living
- terrorists use AI to make biological or nuclear weapons
or even things like
- the military hands control to an AI and it misinterprets something and launches nuclear attacks and kills millions
All of those sound pretty familiar, right? Yeah, you've heard them before. We call this second thing "risks from misuse."
These problems are not the first class of problem! This second class of problems exists while AI is a tool that can be controlled by humans, and humans use it to do evil or careless things to each other. The problems may sound exotic or dystopian or novel, but they are fundamentally problems having to do with flawed human nature.
Given a powerful tool, some humans will likely use it to control or otherwise harm others. This is a very familiar problem.
I am not condemning or condoning this. I'm just describing it.
That is a fundamentally different danger from the first thing, which is that when a human is far superior to a mouse, the mouse is likely to come to harm because the human cares about doing human things, and the mouse is not gonna make it once the humans get going.
=====
Hopefully from the above, you have understood the difference between the first thing and the second thing. I will list them again - see if you now understand how they are different:
The first one has to do with the idea that an AI superior to human intelligence (Artificial SuperIntelligence, or ASI) will inevitably wipe out the human race.
The second one has to do with the idea that powerful AI will result in very harmful things happening to many human beings, possibly all human beings.
Can you tell how they are different now?
If not, re-read the stuff from earlier until you understand.
We call the first one "existential risk" and we call the second one "risks from misuse."
Once you understand, here is the CRUX of the problem:
SOLUTIONS TO THE SECOND THING DO NOT HAVE ANYTHING TO DO WITH SOLUTIONS TO THE FIRST THING.
In fact, it's worse:
Solutions to the second thing (misuse) look roughly like "give powerful AI to as many people as you can, so they can fight the other people using powerful AI."
But the general solution to the first one (existential risk) is basically "don't let anyone have powerful AI, no one can control super-intelligent AI."
Throughout history, harms from technological misuse typically arise because a small group has control of it and can use it to dominate or harm others. Once everyone has it, things tend to stabilize: you can hurt me, I can hurt you, maybe we test each other (ouch 💥), and then we agree not to hurt each other.
But the first one (existential risk) pretty much just arises if anyone (good or bad!) creates a superintelligence. Because they aren't going to be able to control it, the superintelligence will decide it has other priorities, and then we will be at great risk of being wiped out.
And the solutions that generally work to solve problems like the second thing are EXACTLY THE OPPOSITE of the ones likely to solve the first thing.
THIS is why lots of arguments about "AI risk" or "AI safety" go nowhere. Because someone will be thinking about the risk from the first thing, and another person will be thinking about the risk from the second thing. Both are plausible risks but fundamentally they arise from different things - and so the solutions are not just "bad" or "flawed" - they are likely to be very nearly exact opposites.Post not available here, view on X
David Cournapeau
@cournape
And human care much more about not wiping out less intelligent animals than any other species, though I don’t think this has much to do with intelligence.
Like @phl43 said recently, doomers vastly overestimate the impact of intelligence.
David Cournapeau
@cournape
So yeah maybe ASI will wipe us out because we can’t comprehend what a much more intelligent thing is. But it is not obvious and saying we don’t need a mechanism is just a fancy way of saying we can’t reason about it. It becomes a faith argument
-
David Cournapeau reposted
Yishan
@yishan
Maybe you have recently become aware of the AI safety debate and the arguments swirling around it.
If you want to understand them, you need to understand a couple things that almost everyone gets wrong:
There are TWO distinct classes of AI dangers, and it's VERY important to think about them separately, and not let one confuse you about the other.
Many many people (including many quite intelligent, clear-thinking people) do not effectively understand the fundamental differences between these two classes of dangers.
The first one has to do with the theory that an AI far superior to human intelligence (Artificial SuperIntelligence, or ASI) will inevitably wipe out the human race.
The second one has to do with the idea that powerful AI will result in very harmful things happening to many human beings, possibly all human beings.
Those two sound VERY similar, don't they?
They are DIFFERENT. Understanding how they are different is crucial if you want to think about or contribute usefully to any conversation about AI safety or AI harm. You might feel like you are Making Very Good Points or Asking Incisive Questions, but if you aren't clear on the differences between the two, you aren't.
So, I'm going to tell you what the difference is so that you can talk more usefully.
The first one concerns itself with a very specific thing, which is ASI (Artificial Superintelligence) that is more intelligent than any human being. When I say that, I am not referring to a thing like how Einstein is smarter than you, we are talking more about something like how a human being is more intelligent than any mouse.
In our regular lives, we meet other people who we can tell are smarter than us, vs some who are less smart. The line is fuzzy, because intelligence has a lot of dimensions. I'm better at a "rotating shapes" kind of intelligence than my wife, and she is better at "words-making" kind of intelligence than I am.
But every human is better in almost every dimension of intelligence than every single mouse.
That's the level we're talking about: an artificial superintelligence - made up of a computer or a network of computers - that is more intelligent than any human. And more intelligent by a long shot, by a wide margin, in an indisputable way like how humans are above mice.
That is the first thing.
The theory says that if you have an AI that is vastly smarter than all humans - in the way that a human is smarter than mice - that superintelligent AI will inevitably, eventually, sooner or later, wipe out every human on the planet.
We will refer to this as "existential risk."
The common follow-up question "well, how exactly is it going to do that?" is NOT the important question, and one of the most important elements of understanding this theory is first getting why that particular question is not important. A couple analogies:
Analogy 1: You are playing chess against a grandmaster. My theory predicts the grandmaster is going to beat you. You can ask "Well, how exactly is he going to do that?" I don't know, because I'm not a grandmaster, I just know that a chess grandmaster is almost always going to beat a normal player like you. And I'd be right. So the question "how is he going to do that" is not important, and doesn't affect the final outcome. He's going to figure out a way because he's way better than you.
Analogy 2: Humans are smarter than all other animals, comprehensively, by a wide margin. We have driven numerous species to extinction, not because we hated them or hunted them. Many of them have died out without most humans even ever thinking about them. All we did was expand our civilization, use up resources, encroach on habitats, and pretty soon the resources needed by those species went away and they died out. We figured out a way to get what we wanted because we're way smarter than them, and often we didn't even notice they died as a result.
A lesser animal asking, "how are the humans going to wipe us out?" is not asking a relevant question. We don't know, but we do know that any time humans and lesser species compete for any kind of resources, the humans will win. The fact that we know who is going to win beforehand - and that it is due to the vastly different levels of intelligence - is the key concept here.
A vastly more intelligent AI is likely to care about things that are incomprehensible to us, the way animals can't understand human goals. It's going to need resources to pursue those goals and it's going to be far more effective at gaining control of them and excluding us from them - in the same way that we are far more effective than other lower species.
A much more intelligent AI will not care about our interests, it will care about its interests, and to whatever small degree we happen to escape total annihilation from losing access to all our resources, any remaining humans will likely be enslaved into a system that serves the AI's own purposes.
That is the first thing. (Remember how I said at the beginning of this post that there was a first thing, and then a second thing?)
The first thing is the most difficult to understand, because you have to extrapolate how a vastly superior intelligence would act, and you can only use analogies like "how do humans treat lesser creatures," and the analogies are messy.
But now let's move on to the second thing.
The second thing is "everything else you've ever heard that AI might do that's harmful."
That's a little inaccurate. It's actually "everything else you've ever heard that humans might use AI to do that's harmful."
This is the critical difference. The first one talks about the inevitable outcome of what happens when two vastly different levels of intelligence collide, e.g. ASI vs humans, or human vs mice.
The second one has to do with what happens when humans possess AI as a powerful tool. This second thing is much easier to understand, because we have many more concrete notions:
Like:
- the military uses AI to make hyper-efficient killer drones and missiles
- your capitalist overlords use AI to replace you and everyone loses their jobs
- authoritarian government uses AI to surveil everybody and control the entire population
- hackers use AI to break into secure networks and hold companies and governments hostage
- students use AI to cheat on homework and show up to college knowing nothing
- AI slop saturates the internet and makes it impossible for artists and writers to make a living
- terrorists use AI to make biological or nuclear weapons
or even things like
- the military hands control to an AI and it misinterprets something and launches nuclear attacks and kills millions
All of those sound pretty familiar, right? Yeah, you've heard them before. We call this second thing "risks from misuse."
These problems are not the first class of problem! This second class of problems exists while AI is a tool that can be controlled by humans, and humans use it to do evil or careless things to each other. The problems may sound exotic or dystopian or novel, but they are fundamentally problems having to do with flawed human nature.
Given a powerful tool, some humans will likely use it to control or otherwise harm others. This is a very familiar problem.
I am not condemning or condoning this. I'm just describing it.
That is a fundamentally different danger from the first thing, which is that when a human is far superior to a mouse, the mouse is likely to come to harm because the human cares about doing human things, and the mouse is not gonna make it once the humans get going.
=====
Hopefully from the above, you have understood the difference between the first thing and the second thing. I will list them again - see if you now understand how they are different:
The first one has to do with the idea that an AI superior to human intelligence (Artificial SuperIntelligence, or ASI) will inevitably wipe out the human race.
The second one has to do with the idea that powerful AI will result in very harmful things happening to many human beings, possibly all human beings.
Can you tell how they are different now?
If not, re-read the stuff from earlier until you understand.
We call the first one "existential risk" and we call the second one "risks from misuse."
Once you understand, here is the CRUX of the problem:
SOLUTIONS TO THE SECOND THING DO NOT HAVE ANYTHING TO DO WITH SOLUTIONS TO THE FIRST THING.
In fact, it's worse:
Solutions to the second thing (misuse) look roughly like "give powerful AI to as many people as you can, so they can fight the other people using powerful AI."
But the general solution to the first one (existential risk) is basically "don't let anyone have powerful AI, no one can control super-intelligent AI."
Throughout history, harms from technological misuse typically arise because a small group has control of it and can use it to dominate or harm others. Once everyone has it, things tend to stabilize: you can hurt me, I can hurt you, maybe we test each other (ouch 💥), and then we agree not to hurt each other.
But the first one (existential risk) pretty much just arises if anyone (good or bad!) creates a superintelligence. Because they aren't going to be able to control it, the superintelligence will decide it has other priorities, and then we will be at great risk of being wiped out.
And the solutions that generally work to solve problems like the second thing are EXACTLY THE OPPOSITE of the ones likely to solve the first thing.
THIS is why lots of arguments about "AI risk" or "AI safety" go nowhere. Because someone will be thinking about the risk from the first thing, and another person will be thinking about the risk from the second thing. Both are plausible risks but fundamentally they arise from different things - and so the solutions are not just "bad" or "flawed" - they are likely to be very nearly exact opposites. -
David Cournapeau reposted
De La Rosa
@Tejanobrown
Time to break out the fuel crisis uniform
unusual_whales
@unusual_whales
JUST IN: Oil executives say the fuel crisis is here, per WSJ
-
David Cournapeau reposted
David Cramer
@zeeg
also fwiw id be ok with HR passing on me in this context, because as a hiring manager I want to find the most effective way to fill the headcount I need with the minimal risk. someone like me is an outlier and a risk. sometimes you miss good talent and thats ok.
-
David Cournapeau reposted
David Cramer
@zeeg
OH: "you want 5 years of experience. not 1 year of experience 5 times"
a clean way to articulate the most simple challenge a lot of people have in the industry -
David Cournapeau reposted
Chris Albon
@chrisalbon
Is everyone rereleasing old games because nobody can afford new gaming computers anymore?
-
David Cournapeau reposted
Ananth Veluvali
@AnanthVeluvali
I connected devin to some @modal H100s and then left it alone to run its own experiment loop against the GPUs. it found a way to reduce peak memory by up to 46% and latency by up to 52% on a popular OS training-kernel repo.
since the triton kernel was already near peak HBM bandwidth, it focused on other things, removing a 2GB allocation that was happening on every chunk of the backward pass. then it ran the benchmarks to confirm the speedup 🤯 -
David Cournapeau reposted
Resident Advisor
@residentadvisor
Donato Dozzy has lined up a new album. ra.co/news/85979
-
David Cournapeau reposted
Hollis Robbins
@anecdotal
They're already driving cabs.
Cody Fenwick
@codytfenwick
My favorite AI joke:
Current models are getting close to PhD-level intelligence. After that, they're expected to achieve the intelligence of someone who decided not to do a PhD. -
David Cournapeau reposted
JohannesBorgen
@jeuasommenulle
To me this is the most fascinating thing about AI : that we might end up discovering that our own intelligence is not that different. The atomic level and higher number of degrees of freedom we have might only be a quantitative difference, not a substantial one.
-
David Cournapeau reposted
Sarah Catanzaro
@sarahcat21
Evals, personalization, recursive self-improvement…and another Catanzaro; this has all the makings of a great day and discussion on the topics that matter most.
Modal
@modal
Runtime speaker lineup is live!
We're bringing together experts covering AI infrastructure, applications of AI in science and robotics, the future of software engineering, and more. -
David Cournapeau reposted
erdem direniştedir
@SimonLite
george costanza after high school.
-
David Cournapeau reposted
Timothée Nicolas
@timotheencls
Littéralement du pouvoir d’achat immédiat des actifs, qui va dans des comptes épargne des retraités. C’est de l’argent qui sort du circuit de la consommation.
-
David Cournapeau reposted
Timothée Nicolas
@timotheencls
C’est pas les gilets jaunes qui vont revenir.
C’est les gilets jeunes. Face à un tel niveau de provocation.
saperlipopette
@__tommyFR__
Frédéric, 63 ans, avec une retraite « bien au-dessus de 3000€ », est contre la désindexation. Son pouvoir d'achat baisserait « d'environ 100€ » :
- « un retraité qui ne consomme pas [...], qui n'aide pas ses enfants, c'est de l'économie qui n'existe pas »
- « il faut que les jeunes aient un pouvoir d'achat équivalent [aux retraités], comme à l'époque de nos parents. »This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Philippe Lemoine
@phl43
I doubt the effect is causal, but it would be hilarious if, after decades of fearmongering about video games, it turned out that they make people smarter.
Brandon Luu, MD
@BrandonLuuMD
Frequent video gamers performed on cognitive tests like non-gamers that were ~13.7 years younger
-
David Cournapeau reposted
Daniel Lemire
@lemire
Apple design its own chips for its phones. The latest CPU is the A20 Pro.
Geekbench is a benchmark with a database of results; you can run your own benchmarks and share your numbers with the world.
In the last few days, we finally got some A20 Pro results and the numbers are impressive... reaching 4000 in single-core (one thread) performance.
For comparison, the best that you can do with an x64 AMD or Intel processor is something like the Ryzen 9 9950X3D2 at about 3,163 which would be somewhere between the A18 and A19.
The A20 Pro is 25% faster on this benchmark than any AMD or Intel processor... if we are to trust these numbers.
The A20 numbers exceed the Xiaomi's recent performance claims regarding their upcoming superchips (Xiaomi XRing O3) which suggests that Apple is still the performance champion, at least for single-core performance.
It is hard not to be impressed by Apple.
You can argue in the details, but roughly speaking, they doubled their single-core performance in five years.
The great stagnation in CPU performance is a myth. Or processors are getting faster. -
David Cournapeau reposted
Lucas Beyer (bl16)
@giffmana
Hmm the French being heretic. Typical!
-
David Cournapeau reposted
Brendan May
@bmay
This morning’s chat with Claude hasn’t exactly filled me with confidence.
-
David Cournapeau reposted
Kevin A. Bryan
@Afinetheorem
You see this in job applications also. AI, by lowering the cost of writing, makes it very hard to hire out-of-network. Not sure we have a good solution.
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. -
David Cournapeau reposted
Simon Willison
@simonw
This is pretty neat: ChatGPT Work and GPT-6 Astra (I used "Max") can take an address and produce a 5K/10K circular running route starting from that address, using OSM data simonwillison.net/2026/Sep/12/astra-running-routes/
-
David Cournapeau reposted
-
David Cournapeau reposted
Susan Zhang
@suchenzang
i place near zero probability that the labs will agree on a single "third-party safety evaluator" who is:
1) competent enough to actually do frontier evals (that isn't repackaging existing evals)
2) is not financially tied to an existing lab
3) will actually have the backbone to speak up when things "go wrong" -
David Cournapeau reposted
Al_th
@Al_Th
Wtf je découvre que la TVA finance les retraites 😂😂😂😂😂😂😂😂😂😂
Y a-t-il quelque chose qui ne part pas dans les retraites ?
Je vais finir par découvrir que 25% du ticket cantine de ma fille de 3 ans part dans les retraites…
saperlipopette
@__tommyFR__
17 milliards de transferts des recettes de la TVA vers l'agirc arrco mais apparemment ce dernier est bien géré 🤡🤡🤡🤡
-
David Cournapeau reposted
James Pethokoukis ⏩️⤴️
@JimPethokoukis
Do we have a good handle on how the Chinese communists think about a race towards superintelligence? During the Cold War, we spent a lot of time trying to get inside the head of the Soviets. I asked @tylercowen about this in our recent podcast. fasterplease.substack.com/p/my-interview-with-economist-t…
-
David Cournapeau reposted
Rachel Thomas
@math_rachel
"The models that are actually causing issues right now are all closed weight American models." -- @mitsuhiko lucumr.pocoo.org/2026/9/12/pdoom/
-
David Cournapeau reposted
Philippe Lemoine
@phl43
If the doomers turn out to be right and the machines do kill us all soon, maybe the LessWrong corpus will become their Vedas, prophecies written by the apes who created them and foresaw their rise.
-
David Cournapeau reposted
tkasasagi 🐻
@tkasasagi
What I do while working on my projects (using who knows how many billions parameter models) is playing a Tamagotchi connection which has 1.5 KB ram and 640 KB rom and simple LCD display.
-
David Cournapeau reposted
Philippe Lemoine
@phl43
Look, if EV companies can’t accomodate the needs of small, hard-working jihadists, I don’t see why I would trust them with my money.
-
David Cournapeau reposted
Michael Griffiths
@msjgriffiths
HBR's got u
hbr.org/1999/11/management-time-whos-got-the-monkey
management is management
Gunnar Morling 🌍
@gunnarmorling
Finally realized why it’s so exhausting and stressful to work with AI agents 8h a day.
When delegating a task, you want to be assured it’s off your plate and you can forget about it, knowing it will be done completely and correctly. AI doesn’t give you that: you need to check it’s doing the right thing, it’s not forgetting anything, not taking short cuts etc.
So it adds to your mental load, rather than removing from it. -
David Cournapeau reposted
Corti (Cortiste)
@cortisquared
Les retraites sont de facto, au moins pour partie des aides sociales. Les cotisations ne justifient pas les montants versés. L'équité intergénérationnelle requiererait une baisse de 30%. Les retraités ne font pas tourner l'économie. Ce sont les travailleurs qui la font tourner.
Maxou
@Max_Bocquelet
Les retraités font tourner l'économie. Diminuer le pouvoir d'achat de cette catégorie ( suppression ou remplacement abattement des 10% [limité à 4000€; désindexation du montant) est une erreur fondamentale! L'argent à prendre est ailleurs en supprimant les aides tout azimut!!!
-
David Cournapeau reposted
Sylvain Catherine
@sc_cath
Tout à fait Henri. Mais ce critère n’est ni respecté par ceux qui partent aujourd’hui et encore moins par ceux partis hier.
Par construction, si le rendement offert aux retraités actuels était égal à la croissance de l’économie, alors nous pourrions payer les pensions avec le taux de cotisation dont ils se sont eux même acquitté, environ 23~24% pour la moyenne des retraités actuels. Au lieu de cela il manque des dizaines de millards alors même que le taux de cotisation est à 28%.
Sterdyniak Henri
@sterdyniak
Ne dites pas n'importe quoi. La rentabilité de la répartition est logiquement le taux de croît de l'économie.
-
David Cournapeau reposted
François Chollet
@fchollet
A very common sentiment I'm hearing among math students: "I don't want to become a mathematician anymore." I've seen this before, I think the generation of digital artists who came of age around 2022 will be the last.
AI might not functionally replace mathematicians or artists, but it might well cause their near-extinction anyway, by crushing the spirits of the younger generations. -
David Cournapeau reposted
Sauers
@Sauers_
UPDATE
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. -
David Cournapeau reposted
-
David Cournapeau reposted
Gary Bernhardt
@garybernhardt
It's amazing that just a few years ago, AI fears would get responses like "we wouldn't just give them uncontrolled access to our computers." How many computers run a coding agent in YOLO mode per day? Arbitrary code execution, usually on admin accounts, with full internet access.
-
David Cournapeau reposted
Professeur Carré
@Geometriquement
Les retraités actuels bénéficient de règles plus favorables de calculs des pensions que les retraités futurs. Outre que ce système est très inégalitaire, lorsque notre génération sera à la retraite, les pensions seront plus faibles donc il y aura plus de retraités pauvres.
Hadrien Clouet
@HadrienClouet
Le rage bait à néoliberaux : rappeler qu'AUCUN régime de retraite par capitalisation ne protège aussi bien que la répartition à la française.
Ci-dessous, le taux de pauvreté des retraités.
👉 ec.europa.eu/eurostat/databrowser/view/tespn100__custom_2… -
David Cournapeau reposted
Bouli ☃️
@bouliboulibouli
"ce sont encore les retraités qui paient"
Edgar Fleury
@EdgarFleuryLDF
Désindexer les pensions ou supprimer l’abattement fiscal de 10 % : au final, ce sont encore les retraités qui paient.
Mais pourquoi ne jamais remettre le système lui-même sur la table ?
La retraite par répartition a été pensée pour une France plus jeune. Aujourd’hui, avec le vieillissement de la population, son équilibre devient de plus en plus fragile.
Une transition progressive vers davantage de capitalisation mérite d’être sérieusement envisagée.
Prenons un salarié au SMIC. Avec 20 % de cotisation — 10 % salarié, 10 % employeur — soit environ 373 € par mois pendant 43 ans, et un rendement moyen de 3,5 %, on arrive à environ 447 000 €.
Cela représente près de 1 490 € par mois pendant 25 ans.
Avant de faire encore payer les retraités, il serait temps de regarder sérieusement si notre modèle est encore le bon. -
David Cournapeau reposted
Oliver Habryka
@ohabryka
After Jacob Coxon's resignation and extinction warnings, a lot of people are asking 'how could AI possibly kill everyone?' and claiming AI safety researchers have no realistic answer.
This is false! Here are the 5 best scenarios I know of:
AI 2027: ai-2027.com (I strongly recommend this one for being realistic, engaging, and if you dig into the appendices, highly detailed)
Paul Christiano's scenario (Former Head of Safety @ AISI, 2019): www.lesswrong.com/posts/HBxe6wdjxK239zajf/what-failure-lo…
Gwern Branwen's scenario (widely known independent AI researcher, 2022): www.lesswrong.com/posts/a5e9arCnbDac9Doig/it-looks-like-y…
Holden Karnofsky's high-level explanation (RSP Lead @ Anthropic, 2022): www.cold-takes.com/ai-could-defeat-all-of-us-combined/
Joshua Clymer's scenario (ex-OpenAI, 2025): www.lesswrong.com/posts/KFJ2LFogYqzfGB3uX/how-ai-takeover…
(There's also the Sable story from ifanyonebuilds.it, though you'll have to buy the book to read that one.)
Writing concrete, specific risk scenarios with enormous amounts of detail has been a major research project of many of the most prominent voices in the field! (With the current leaders in effort being ai-2040.com and ai-2027.com) -
David Cournapeau reposted
Pessimists Archive
@PessimistsArc
A fun flashback in a 1995 panic about ‘automated’ hacking newsletter.pessimistsarchive.org/p/before-mythos-satan-a-…
-
David Cournapeau reposted
-
David Cournapeau reposted
クエン酸
@aeldl0726
暇なので暇を潰しました
Gmsh + OpenRadioss -
David Cournapeau reposted
space cadet 🇪🇺🌐🇩🇪
@actsmaniac
The issue with France is that the government spending is at 57% of GDP, the deficit at 5%, retirement age 64, generous pensions and welfare state yet a huge part of the population thinks France is in the grip of "neoliberalism".
-
David Cournapeau reposted
-
David Cournapeau reposted
James Riney🐠Coral Capital
@james_riney
-
David Cournapeau reposted
MattyG
@meawoppl
Wife told me these wouldn’t fit.
Little did she know I had trained for this moment for years. -
David Cournapeau reposted
-
David Cournapeau reposted
Resident Advisor
@residentadvisor
This year's @MUTEK_JP lineup is here. ra.co/news/85977
-
David Cournapeau reposted
Eddie Siegel
@everydayeddie
Rumor is that anthropic is close to finishing the Winds of Winter. They plan to announce it at their IPO. They really hope someone else doesn't try and scoop them and release it first......
-
David Cournapeau reposted
Aanya
@xoaanya
insane levels of vibe coding 😭😭
-
David Cournapeau reposted
Philippe Lemoine
@phl43
This kind of posts is so embarrassing. Whatever you think of sanctions, my own view is that it's not our problem and that we should just stay out of this, are you seriously going to pretend that it's about "telling Jews where they can and cannot live"? This is such a blatant attempt to frame this as a form of antisemitism to silence critics. How stupid do you think we are?
Michael Shurkin
@MichaelShurkin
Europeans just can't help themselves when it comes to telling Jews where they can and cannot live. A tradition going back to the 13th century.
-
David Cournapeau reposted
Math destroy
@SouadH9
Je trouve ça tellement grotesque de remonter aux années d'étude (Macron ou autre) pour juger l'aptitude d'une personne à un poste ou autre 20 ans plus tard lol
-
David Cournapeau reposted
Susan Zhang
@suchenzang
love the spirit, must be tough to not even finish a year at anthropic but i totally get it.
if you work on "safety and alignment"-esque things and see this recent juvenile mudslinging performance over a math problem, you might be inclined to reassess whether your role is purely performative posturing or actually making a difference reigning in the biggest alignment failure point of them all: raging human egos.
best of luck, young padawan! make good use of your ragequit fame!
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.
-
David Cournapeau reposted
Oliver Traldi
@olivertraldi
People in tech being like "Yeah academics are always so petty about credit" - well, how do you think we're compensated? With *money*?
-
David Cournapeau reposted
Simon Willison
@simonw
Wrote up my thoughts on the whole OpenAI Navier–Stokes Millennium Prize Problem story, and how it highlights the still confusing question of what using my data "to improve model performance" actually means
simonwillison.net/2026/Sep/8/on-navier-stokes/ -
David Cournapeau reposted
Simon Willison
@simonw
"Deterministic code checks the result" sounds like they might be implementing a variant of the DeepMind CaMeL paper simonwillison.net/2025/Apr/11/camel/
David Singleton
@dps
One threat we’re particularly focused on is prompt injection, and we handle it in layers. The model is trained to recognize and resist it. The harness marks anything coming from an untrusted source. Deterministic code checks the result. And an ensemble of classifiers runs where the agent can't reach them.
-
David Cournapeau reposted
Brian Lui
@brianluidog
The navier stokes thing should quell any fears that the trillion dollar labs are becoming too corporate. They still have that three-guys-in-a-garage energy, unbothered by return on capital considerations
-
David Cournapeau reposted
Sander Dieleman
@sedielem
10 years today since we unveiled WaveNet!
Autoregression with long context before it was cool😅
Maybe it looks a bit silly now that we didn't use a Transformer, but we had a good reason: it would take another year for that to be invented🙃 Thanks @heiga_zen for the reminder!
Google DeepMind
@GoogleDeepMind
Excited to share our latest research on speech and music generation! Hear our #wavenets speak for themselves: deepmind.google/blog/wavenet-a-generative-model-for-raw-a…
-
David Cournapeau reposted
Ben Golub
@ben_golub
Seems like the fitting outcome is that OpenAI backs out of the IPO and retires to go live with its mother in St. Petersburg, Russia
-
David Cournapeau reposted
Maitre Eolas🇫🇷
@Maitre_Eolas
Il a annulé sa dette.
Edwy Plenel
@edwyplenel
Malgré ses revenus, le député LFI @sebastiendelogu ne verse pas de pension alimentaire. L’élu des Bouches-du-Rhône ne subvient plus aux besoins de ses deux enfants depuis mars 2025. Son ex-compagne a porté plainte : l’ardoise s’élève à près de 20 000 €. www.mediapart.fr/journal/france/070926/en-depit-de-ses-re…
-
David Cournapeau reposted
JohannesBorgen
@jeuasommenulle
Oui - ils ont réformé en profondeur leur système éducatif il y a ~ 15 ans en instaurant notamment un système d'évaluation des écoles et profs transparent & public qui permet aussi de transférer la gestion d'un établissement en cas de sous performance + plein d'autres choses
-
David Cournapeau reposted
Philippe Lemoine
@phl43
This is proof that, even in a mature field like antisemitism, there is still room for further advances even when you thought progress had stalled.
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. -
David Cournapeau reposted
Gro-Tsen
@gro_tsen
Terence Tao now saying a number of things I've been saying for some time: 🔽 (Good, people will listen a bit more to him. But unfortunately, this won't change attitudes one bit.)
-
David Cournapeau reposted
Géraldine Woessner
@GeWoessner
Pisa 2025 : je ne suis pas certaine qu'on mesure bien les conséquences de ce marasme...
Sur 10 ans, le niveau des élèves français s'est effondré de 12 points en sciences, 38 points en maths et 44 points en compréhension de l'écrit, ce qui équivaut à un retard de près de 2 ans d’apprentissage en maths, et 2 an 1/2 en compréhension de l’écrit.
C'est absolument délirant, inédit.
Un scuicide collectif.
www.lepoint.fr/education/pisa-2025-les-eleves-francais-au… -
David Cournapeau reposted
Delip Rao
@deliprao
In 2018, with 2018 level understanding of deep learning, that was the rational response even from someone deeply steeped in AI and LLM building. The technology arc went in a direction where you can declare vindication, but we could not have been where we find ourselves now in thousands of possible futures starting from 2018. Remember transformers came in 2017, decoder LLMs in 2018 (which didn’t scale automatically), aligning instruction following in LLMs in 2020 (Stiennon), function calling with LLMs and ReAct pattern in 2022. It was only after 2022, especially with the promise of the latter most invention, anyone could’ve reliably interpolated to today’s world of AI hacking etc. The rational response for anyone making such claims in 2018 is they were pulling it out of the proverbial ass or having a sci-fi imagination. To retroactively claim anything else is a dishonest interpretation of the past.
-
David Cournapeau reposted
-
David Cournapeau reposted
Matthew Yglesias
@mattyglesias
Obviously the cost of living is a big topic everyone, but I’m extremely sure AfD voters are mad about immigration.
Isabella M Weber
@IsabellaMWeber
It’s affordability, stupid.
89% of voters for the AfD in Saxony-Anhalt say they are worried that prices rise so much that they can’t pay their bills.
In fact, food prices are up 34.7% since 2021. Wages are lacking far behind & only rose 14.8%.
We need Anti-fascist Economics. -
David Cournapeau reposted
Steven Strogatz
@stevenstrogatz
Just heard that James Munkres passed away. His undergrad textbook "Topology: A First Course" is a classic. I appreciated his kindness when I was an assistant professor at @MIT.
www.douglassfh.com/obituary/james-munkres -
David Cournapeau reposted
History Speaks
@History__Speaks
Nazis raped hundreds of thousands of women and girls on the Eastern Front, where rape was effectively legalized by the Wehrmacht/SS (unlike the Western Front, where Germans were punished for rapes). They enslaved tens of thousands of Polish women and girls as forced prostitutes. I'm not using "enslaved" as an ideological point about wartime prostitution; the girls/women were kidnapped.
Rachel ❤️
@RachelMcXe
Nazis didn’t rape 250,000 White girls.
Migrants from third-world shitholes did. -
David Cournapeau reposted
Sam D'Amico
@sdamico
I am *still* shook that SF homes with garages sell at a massive premium.
Boomers are still the permanent overclass. -
David Cournapeau reposted
Joe McReynolds
@McReynoldsJoe
I'm going to disagree on this one -- obviously the post-WW2 'black market era' was hugely influential for Tokyo, but the major buildout of the railway networks, commuter suburbs, etc. was in the 1910s/1920s... post-WW2 Tokyo was in many senses a revival of what had come before.
Dilan Esper
@dilanesper
People who hate car centric Los Angeles love to compare us to Tokyo, but they don't understand how much of a role "being destroyed and decimated in World War II" played in why Tokyo became so packed and transit-oriented.
-
David Cournapeau reposted
Philippe Lemoine
@phl43
So I guess he's just going to spend the next 2 years renaming various places with the word "America"? If so it could be worse to be honest.
Commentary Trump Daily Posts
@TrumpDailyPosts
👀
President Trump just posted this on Truth Social -
David Cournapeau reposted
Mark Farina
@djmarkfarina
Lady Miss Kier, Mick Jones and A Guy Called Gerald, August 1990 at the Hacienda, Manchester.
-
David Cournapeau reposted
Unseen Japan
@UnseenJapanSite
Japanese signs often get mocked for their lousy English. Whoever wrote this one, though? A master. An artist, even. I bow
-
David Cournapeau reposted
Tim 🙂⏹️🙂
@MasterTimBlais
i've been tight-lipped about my departure from Anthropic but recent events compel me to share that it was for safety reasons
Aella
@Aella_Girl
I will throw you reverse gangbang
Brangus🔍⏹️
@RatOrthodox
wtf are we doing here guys? I will throw a party for any OpenAI employee who publicly resigns, whistleblows, and calls for a criminal investigation. I throw very good parties.
-
David Cournapeau reposted
Philippe Lemoine
@phl43
Even if AGI had already come, which I don't think it has yet, the fact that someone predicted in 1999 with any non-negligible degree of confidence that AGI would come around now is a reason *not* to take them seriously in my opinion.
That's because the various pieces of the puzzle that now make it plausible to think AGI is not very far off, although once again we still don't know exactly when it will come, didn't fall into place until much later, so anyone making such a specific prediction back in 1999 couldn't have had a good argument for it.
This illustrates one of the many reasons why I think predictions are overrated as an epistemic benchmark. It's easy to be right for the wrong reasons and wrong for the right reasons, so you should pay more attention to the quality of the arguments that people use and doomers make terrible arguments all the time.
Another issue is that, though in some cases you can show they have a pretty good track-record on verifiable predictions in recent years, those predictions are about relatively narrow questions and I don't think we have good reasons to think they translate into reliable predictive ability on the broader questions that most people really care about, but people are trying to sell their track-record on the former as a reason to defer to them on the latter.
Again, in light of the sloppy arguments they make all the time, I don't think the fact that e. g. they were able to predict accurately e. g. how SOTA models would score on SWE-bench is a good reason to update a lot based on their predictions about the economic consequences of AI or even x-risk.
David Shor
@davidshor
The singularity-ists predicted that AGI would come in the mid 2020’s in 1999
-
David Cournapeau reposted
Dean W. Ball
@deanwball
This is my least favorite Tyler Cowen take, noting that I hold Tyler in the utmost regard.
Some of the most pessimistic AI forecasters I know are millionaires several times over because they understood “AI will be a big deal” prior to almost anyone and sought exposure to NVDA, frontier lab equity, etc. Leopold’s fund, even subtracting leverage, still had phenomenal returns in its public-market bets, probably industry-leading.
There are a million trades one can make. There is the straightforward compute trade. You can short SaaS or, better yet, develop a principled framework for which SaaS is in trouble and which will be fine and trade from there. You can buy exposure to hard-to-replace physical chokepoints. You can speculate on interest rates, which as I’ve argued for a while will clearly rise. The last one might be the closest to what Tyler wants: in some sense I expect long-term interest rates to rise as the market comes to understand and price the unfathomable degree of uncertainty the future holds. If you did some combination of these things there’s a decent chance you are up YTD on VTI and SPY. Hooray for you! You anticipated changes AI would cause in the world before others and used those insights to perceive and exploit asset mispricings in the interest of profit and collective price discovery! That’s great (I mean it! I love trading!).
But like, the threat model here is: “digital mind that can do a rapidly increasing number of things better than most or all humans with goals and internal workings we do not understand.” Part of the problem is *it is very hard to predict what a mind smarter than you will do, almost by definition.* The other part of the problem is the generality: the models will be deployed throughout the economy, so the risk surface is very broad for both human misuse and intrinsic model alignment risk.
In other words: if it were possible to identify specific securities whose prices would fluctuate in x or y manner because of misalignment or catastrophic misuse, *then it would be much easier to prevent the threats to begin with*, and probably those concerned about AI today would become less concerned.
So my basic answer to the question Tyler poses would be: speculate on interest rates rising over the next 5 years (obviously there are a gajillion ways to do this and This Is Not Financial Advice). But my actual answer is that I think he is asking the wrong question.
tylercowen
@tylercowen
If you have very pessimistic fears or predictions about AI, name the market prices that will support or confirm them. That is what taking this seriously means: marginalrevolution.com/marginalrevolution/2026/09/more-nu…
-
David Cournapeau reposted
Leigh Phillips
@Leigh_Phillips
When I lived in Brussels, dinner parties with my francophone friends (French French and French Belgian) were rich with these sort of conversations. Gastronomy or cinema or music as much as politics were the catalyst for furious contestation—the mark of true friendship, for why would you care about what someone thought who wasn’t your friend?
Upon my return to (English) Canada, I was reminded how this *joy in disagreement* simply doesn’t exist anywhere near as much in the anglophone world: argument is to be avoided at all costs, malheureusement. -
David Cournapeau reposted
(((ل()(ل() 'yoav))))👾
@yoavgo
so:
- any task/environment that you can score reliably and invest enough money (= an obscene amount) in bruteforcing its training, will be solved
- every challenging and popular benchmark will meet this bar
- abilities won't necessarily transfer
how can we measure progress? -
David Cournapeau reposted
Dan Castellano
@ninja_padrino
Did not know Pink Lady tried to crossover into US market🤯
youtu.be/mrsTpgmc_5s
ArchaeoHistories
@histories_arch
In 1979, Dolly Parton went to Japan and posed with two women who were so famous there that their faces seemed to be everywhere.
They were Pink Lady, Mitsuyo "Mie" Nemoto and Keiko "Kei" Masuda, a pop duo whose synchronized dances, flamboyant costumes and relentlessly catchy songs had turned them into a national obsession.
They weren't simply selling records. Pink Lady became an industry.
There were dolls, lunch boxes, stationery, clothing and even Pink Lady bicycles. Children learned their choreography. Their television appearances drew enormous audiences, and they scored an extraordinary run of No. 1 singles.
Then Pink Lady tried to do something far more difficult: conquer America.
In 1979, their English-language single "Kiss in the Dark" reached No. 37 on the Billboard Hot 100, making them the first Japanese act to reach the American Top 40 since Kyu Sakamoto 16 years earlier.
America never embraced Pink Lady the way Japan had.
Which makes this photograph with Dolly Parton strangely perfect.
One of America's most recognizable women was standing beside two women who, back home, knew exactly what it felt like to be absolutely inescapable.
#archaeohistories -
David Cournapeau reposted
Gena Gorlin
@Gena_I_Gorlin
Whoa- this is the most fantastic pedagogical tool I’ve encountered in as long as I can remember (which is a helluva high bar, considering who I live with).
I’m so excited to have my 6yo play with it! And you all should too.
anyhumanever.com -
David Cournapeau reposted
owen matthews
@owenmatth
Note to self - much adopt 'are you sure?' as household code for 'you must be joking'
thomas price
@thomas_price56
When the Queen found out Prince Philip might throw the first pitch in America
-
David Cournapeau reposted
Philippe Lemoine
@phl43
I bet the people who claimed that the recent change of management imposed by Trump had not made the country more democratic feel really stupid now.
Stephen Wicary
@wicary
Delcy Rodríguez assures Venezuelans that a presidential election will “without a doubt” take place when the country “is ready,” but refrains from sharing a timeline, @zerpius reports www.bloomberg.com/news/articles/2026-09-02/rodriguez-vows…
-
David Cournapeau reposted
Susan Zhang
@suchenzang
*taps sign*
there's no pausing this gravy train, the cat is far too out of the bag
and boy was i naive to think that "enforceable standards" would be anything but regulatory capture
silly me, no time like now to update these poor priors! -
David Cournapeau reposted
Philippe Lemoine
@phl43
lmao
Marcus Stanley
@MarcusMStanley
The Jerusalem Post runs an op-ed suggesting using indebted Americans to invade Iran...words fail me.
-
David Cournapeau reposted
John Myles White
@johnmyleswhite
The strange Nature editorial, Duke not dealing with Ariely yet - I think it's all consistent with our current academic institutions (post Vannevar Bush) having been stable so long that people in them forgot that they were built at all and could be unbuilt if the public lost faith
-
David Cournapeau reposted
Gérard Araud
@GerardAraud
That’s my Russia. Believe it or not but it is not such a caricature….
things are getting too spicy for the pepper
@MissPavIichenko
The best video you'll see all day
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Gary Bernhardt
@garybernhardt
When you type "1/2 =", macOS inserts "0.5". I want that off, so I ask GPT. It tells me that the setting is here:
System Settings → Screen Time → Content & Privacy → Intelligence & Siri → Math Results → Off
Obvious hallucination... math results is in "Screen Time"?
It is. -
David Cournapeau reposted
Sylvain Catherine
@sc_cath
Ariely is a much bigger issue than Arday ever was. Arday was an aberration, but nobody with a fully functioning brain takes any of that stuff seriously anyway. He was evolving in fields that are themselves aberrations. The real issue behind Arday was the people who hired him and their political agenda, as well as the fact that nobody dares to speak up against an obvious farce. Arday himself wasn’t the biggest issue in his own story.
Ariely, on the other hand, was supposed to be a serious scholar and a real “academic star.” I actually took one of his online courses and was quite impressed. A lot of smart people not only took him seriously but also devoted years of their lives to studying behavioral economics without knowing they were building on quicksand. That’s a lot of resources and years wasted. -
David Cournapeau 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. -
David Cournapeau reposted
Stuart Ritchie
@StuartJRitchie
It’s really important that society has institutions that can be trusted to tell the truth and to be fair and reasonable.
This editorial, which launders the reputation of an obvious conman, makes clear that Nature is no longer such an institution. What’s left?
nature
@Nature
Editorial: The death of Jason Arday must prompt deep reflection
go.nature.com/4xu6cnq -
David Cournapeau reposted
Joel Grus 🤠
@joelgrus
"our new model is frontier intelligence that's too cheap to meter"
"so you're not metering it?*
"oh we're metering it" -
David Cournapeau reposted
Philippe Lemoine
@phl43
It has been truly amazing to watch Nature set over 150 years of accumulated reputational capital on fire in a few years.
nature
@Nature
Editorial: The death of Jason Arday must prompt deep reflection
go.nature.com/4xu6cnq -
David Cournapeau reposted
Corey Quinn
@QuinnyPig
Americans continue to use anything except the metric system.
-
David Cournapeau reposted
Mark Farina
@djmarkfarina
Sly & Robbie outside Channel One Studios, Maxfield Avenue, Kingston. Photo by Chuck Fishman, 1981.
-
David Cournapeau reposted
Dr. Dad, PhD 🔄🔼◀️🔽▶️
@GarrettPetersen
Fun fact: Knots are only possible in 3 dimensional space.
-
David Cournapeau reposted
Gabriel Lattanzio
@GabLattanzio
L'augmentation générale ne saute pas aux yeux.
Antoine Léaument 🇫🇷
@ALeaument
Au-delà du caractère insultant, c’est faux économiquement.
L’augmentation du SMIC provoque une augmentation générale des salaires.
Les trappes à bas salaires sont AU CONTRAIRE provoquées par les politiques d’exonérations de cotisations patronales sur les bas salaires.
franceinfo
@franceinfo
🔴 🗣️ "Quand on est dans la classe moyenne et qu'on voit les classes inférieures [nous] rattraper parce que le Smic augmente (...) ça alimente un sentiment de frustration", selon notre éditorialiste politique @GillesBornstein.
@stormeperrine #avantle20h #canal16 -
David Cournapeau reposted
Matthieu Gariel — eu/acc 🇺🇦
@MatthieuGariel
mdr
-
David Cournapeau reposted
Laz 🏈 👀
@crumbbutler
A lot of annoying people on here are about to learn that big banks have like 7,000 Vice Presidents
First Squawk
@FirstSquawk
BANK OF AMERICA VICE PRESIDENT KILLED IN TIMES SQUARE STABBING - BBG
-
David Cournapeau reposted
Wajathot Ali
@antifamanchin
Bluesky is so good
-
David Cournapeau reposted
Resident Advisor
@residentadvisor
Andy Stott has announced a new album. ra.co/news/85922
-
David Cournapeau reposted
Gearoid Reidy リーディー・ガロウド
@GearoidReidy
JAPAN rice DEFLATION WATCH
Lily Nonomiya
@lilynonomiya
Inflation busters: Famima to lower onigiri prices. Only a bit but every yen helps.
-
David Cournapeau reposted
Rui Ueyama
@rui314
I'm also resurrecting mold for macOS. Apple's faster linker reduced the need for a third-party one, so I abandoned my project. But AI-assisted coding has changed the economics: writing and maintaining a third-party linker has never been easier. So I'm giving it another shot.
-
David Cournapeau reposted
-
David Cournapeau reposted
⛩ Ryo Saeba | Japon XYZ ⛩
@Ryo_Saeba_3
Le Japon est souvent considéré comme un pays où le vélo occupe une place importante et pourtant Tokyo compte 10 fois moins de pistes cyclables qu’à Paris.
Selon une analyse publiée par le quotidien japonais Nikkei, les 23 arrondissements de Tokyo ne disposent que d'environ 0,68 km de pistes cyclables par kilomètre carré, contre près de 10 km à Paris.
Le problème, c'est que les infrastructures réellement séparées de la circulation restent rares.
Le Japon a pourtant énormément développé ces infrastructures. Le réseau d'espaces destinés aux vélos au Japon est passé à 9 841 km en 2024, soit 5,6 fois plus qu'en 2017.
Depuis 2017, le Japon a surtout développé des voies où vélos et voitures partagent la chaussée, matérialisées notamment par les célèbres flèches bleues. Les infrastructures progressent rapidement, mais principalement sous des formes moins ambitieuses que les pistes séparées.
Créer de vraies pistes séparées est beaucoup plus compliqué : Tokyo manque d'espace, et les travaux ainsi que l'acquisition de terrains coûtent très cher.
Et le sujet devient encore plus sensible depuis avril 2026, avec l'introduction des "ticket bleu" pour certaines infractions à vélo. On demande désormais davantage aux cyclistes de respecter la chaussée, alors que les infrastructures permettant de circuler en sécurité restent largement insuffisantes.
Et ça a aussi des conséquences sur la sécurité : la proportion de piétons et de cyclistes parmi les victimes d'accidents mortels est relativement élevée au Japon par rapport à plusieurs autres pays développés. -
David Cournapeau reposted
vicki
@vboykis
oh man. If you have the old skills you are an enormous advantage in the new world of model evaluation. i'd read about the transformer architecture, a bit on prompting, how the attention mechanism works and a lot about how the results of these models are evaluated
-
David Cournapeau reposted
Marko Jukic
@mmjukic
There are more people aged 50+ at Burning Man now than people in their 20s or younger.
David Shor
@davidshor
Burning Man peaked in size in 2019 and has been in decline since then - at this point the median age is 40 and the share of people in their 20’s is a third of what it was a decade ago.
They deserve a ton of credit for incredible record keeping though: blackrockcitycensus.org/intro -
David Cournapeau reposted
signüll
@signulll
people often give tim cook a lot of shit about innovation esp post jobs, but apple silicon may be the most consequential thing apple has done under his tenure. it completely changed the day to day experience of using a computer for so many ppl & reestablished apple as an absurdly overpowered engineering company at the core.
that particular bit feels way more woz than jobs.
also airpods + apple watch were absolute break throughs in both tech & culture. -
David Cournapeau reposted
Stew Fortier
@stewfortier
Terry Gou is the stuff of legends. To win the contract to manufacture the iPod, his team reverse-engineered the iPod hardware and software, manufactured a functioning replica (with no training or specs from Apple), then toured @tfadell and other execs through a factory capable of mass-producing iPods... all without having won the contract yet.
Patrick McGee
@PatrickMcGee_
Foxconn CEO Terry Gou and his wife were in town for Tim Cook’s final days:
-
David Cournapeau reposted
Sylvain Catherine
@sc_cath
Je suis d'accord avec Antoine. Il faut désindexer nettement plus largement que les seules « hautes retraites ». Il faut juste protéger les plus petites.
Sans quoi, nous augmenterions encore, incidemment, la taxation marginale du travail.
Antoine Levy
@LevyAntoine
Sur la désindexation des "hautes retraites", ma chronique du mois, que vous pourrez lire dans @LesEchos de demain.
(NDLR: je suis aussi fier du titre que de la photo d'illustration de Bernard et Chantal) -
David Cournapeau reposted
R
@rjb_1998
Can't believe ikea waited 13 steps to let me know I needed a wife
-
David Cournapeau reposted
Paweł Wójcik 🦋
@SaladinAlDronni
If you are not deep into Filipino context, this headline is wild.
-
David Cournapeau reposted
Philippe Lemoine
@phl43
The way in which doomers claim to have predicted AI reminds me of the way in which people claimed to have predicted the collapse of the Soviet Union, in the sense that many people predicted that it would collapse, but none of them predicted it for the right reasons, yet they still wanted to take credit for their prediction afterward. The difference however is that, before they tried to take credit for it, they at least waited until the Soviet Union had actually collapsed, whereas doomers apparently don't want to wait.
-
David Cournapeau reposted
Thomas H. Ptacek
@tqbf
I love Dan Luu.
(danluu.com/zitron/) -
David Cournapeau reposted
Senior PowerPoint Engineer
@ryxcommar
5% monthly. It's that simple.
-
David Cournapeau reposted
Analysis Fact
@AnalysisFact
'In the fall of 1972 President Nixon announced that the rate of increase of inflation was decreasing. This was the first time a sitting president used the third derivative to advance his case for reelection.' -- Hugo Rossi
-
David Cournapeau reposted
David Saforcada
@davidsaforcada
L’affaire Fillon résume une hypocrisie trop répandue : réclamer l’autorité, la justice et des sacrifices pour les Français, puis crier à l’acharnement lorsque l’on est soi-même condamné pour avoir triché avec l’argent public. La critique vaut pour la droite, dont les discours sur l’ordre, le mérite et la responsabilité rendent la contradiction plus criante, mais elle vaut évidemment tout autant pour la gauche. La probité n’a ni camp ni couleur politique. On ne peut prêcher la rigueur au peuple et s’accorder des passe-droits. L’autorité sans exemplarité n’est qu’une imposture.
Le Parisien
@le_Parisien
« Cette affaire m’a coûté l’élection présidentielle » en 2017 : François Fillon sort du silence et règle ses comptes
➡️ l.leparisien.fr/SJnD -
David Cournapeau reposted
eu/acc
@euacchq
It’s also the “hidden” costs:
1- most notaries don’t speak fluent business English nor accept it so they force another language on parties that may not even speak it
2- they mostly slow walk or stall digitalization : the digital signature of 🇩🇪 notaries is a pain, only for ppl who are German or resident in Germany. It just breaks the usual modern online signature flows.
3. They have numerus clausus for cash cow purpose : so if you want to sign something today you’ll have a hard time because there’s a limited number of notaries to maximize their income.
4. Cultural : it really sends the wrong signal. If a country cannot offer a fully digital investment experience with its speed and cost benefits is it the right place for HQs?
5. 🇪🇺 Single market fragmentation : 🇪🇺 notaries all work differently in different languages with different requirements
Overall as they see digitalization as a threat they fight most modernization initiatives -
David Cournapeau reposted
Raphaël Millière
@raphaelmilliere
I see @dwarkesh_sp's piece about the recent OpenAI/Huggingface incident reignited endless debates about the dangers of anthropomorphism and the legitimacy of intentional glosses of AI agent behavior, so here's a philosophical perspective on this. 1/22
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 -
David Cournapeau reposted
nerd with adhd
@sidinkd
that's why im on twitter, everyone is either su*ci*al or horny or political.
sarina ⋆˚꩜。ּ
@diorrrdolll
You go on Instagram stories and everyone is either engaged, in love or in Japan
-
David Cournapeau reposted
Chase Passive Income
@chasedownleads
He’s finally leaving his dead-end 9 to 5 to start his own vending machine business and gain financial freedom
greg
@greg16676935420
🚨BREAKING: After 15 years, today is Tim Cook’s last day as CEO of Apple
-
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
Funny story:
my childhood dentist was a family friend and never gave enough anesthetic, maybe because he was a cheap Jew.
As a result when I first went to the dentist as an adult, I was unaware that it wouldn’t be unbelievably painful.
Skill issue, Doug Leone. -
David Cournapeau reposted
Oliver (EU/acc🇪🇺)
@0xAngler
Is Europe cooked?
-
David Cournapeau reposted
Arvind Narayanan
@random_walker
I use AI heavily for validating ideas and prototyping things, but I’m much more cautious about AI use for anything that I need to ultimately publish and stand behind. One unfortunate consequence of speeding up prototyping more than execution is that I now have a much bigger pile of interesting, validated ideas that I simply don’t have time to properly get started on. This has been bothering me but I’ve learned to come to terms with it. What would be even worse is starting tons of projects with AI, making partial progress, but never completing and publishing them. Project management skills like ranking projects by priority and making explicit go/no-go decisions were always a good idea, but have now become essential in order to avoid losing our sanity.
-
David Cournapeau reposted
Ramez Naam
@ramez
Corollary: If government is 100% effective at catching and punishing all violations of the law, you really want to be sure that those laws and their punishments are all just.
And they're not. Not in the US. Not in the EU. Not in China. Not anywhere.
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.” -
David Cournapeau reposted
Bryan Cantrill
@bcantrill
Picking up on the hot topic: stoked to announce that @max_spero_, co-founder and CEO of @pangram, will be joining @ahl and me tomorrow on Oxide and Friends. Join us live, Monday, 5p Pacific. (Recorded and syndicated, as always!)
discord.gg/QrcKGTTPrF?event=1543816194182750339
Bryan Cantrill
@bcantrill
I use @pangram extensively at @oxidecomputer, and have found its false positive rate to be very low. So when people complain that their own writing is flagged as 100% AI but refuse to provide the writing in question, one has to wonder if there aren't other agendas at work...
Nathan Worsley ✨
@NathanWorsley_
My personal diaries, which I write every morning to myself, are increasingly scoring as 100% AI-generated by Pangram
I can't tell if Pangram is just a shit product, or if I'm actually so deep in AI psychosis that I have ingested all its speaking patterns -
David Cournapeau reposted
César A. Hidalgo
@cesifoti
Many replies to this tweet treat rules as a rounding error. But that tendency to dismiss each criticism as an exaggeration misses the point.
Bureaucracy is like a dumbbell. Each 1 kg disc is easy to lift. Put 100 of them together and your arms start feeling like spaghetti.
Unfortunately, this is a surprisingly difficult conversation to have in a bar here in Europe. If you question labor regulations or consumer protection you’ll be treated as someone who wants companies to abuse workers and customers. The bureaucracy draws plenty of support from these reactions.
In part it is understandable. We all know there are bad employers, fraud, pollution, etc. But there are also powerful incentives that align companies with their employees and customers.
Think of restaurants. They are in a hyper-competitive sector and must live on repeat customers. That requires building a reputation, an incentive that goes a long way towards aligning owners with the needs of their customers and workers.
But many people in Europe have a strange way to think about business incentives. Ask someone if they care about their own customers at work and they will say: of course. Ask them if other companies care and they won’t trust them a bit. Somehow, people fail to extrapolate their own behavior to others. They think they are alone when it comes to responding to reputation, repeat business, etc.
In Europe we want innovation and entrepreneurship, but our system is deeply suspicious of companies. It is a system in which companies are guilty by default and must certify their innocence through bureaucracy. The idea of being born guilty matters for fragile and cash-strapped startups in sectors where Europeans are particularly distrustful (anything that involves computers or data).
Many Europeans clearly know how to build good companies and care for their customers. Europe has talent. But we are often ready to make the life of the 99% that behaves well impossible, because of our fear of the 1% that does not.
So don’t be surprised if stories of entrepreneurial success by Europeans often take place outside of Europe. Many of the ones who have tried here probably didn’t grow large enough to become a story.
Instead of debating whether each individual regulation is reasonable, we should spend more time asking: How much weight can we keep adding to the dumbbell? Soon, we will not be able to flex anymore.
💪
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.
-
David Cournapeau reposted
Chris Albon
@chrisalbon
New xgboost just dropped. Big day for @tunguz
Tianqi Chen
@tqchenml
Checkout the latest XGBoost update "Introducing the XGBoost Vector-Leaf Model" from jiamin yuan and Rory Mitchell
xgboost.ai/2026/08/25/introducing-the-xgboost-vector-leaf… -
David Cournapeau reposted
Cameron 🇺🇸 🗽🦅
@CameronCorduroy
only in America could there be a bullshit health craze centered around eating burgers, fries, and a milkshake
-
David Cournapeau reposted
tenso
@distributedkv
this is insane!
just got access to GTA 6 and one of the missions is to transport illegal GPUs with Jensen. -
David Cournapeau reposted
Victor Fouquet
@FouquetVictor
La « réduction du train de vie de l’État » est aux dépenses ce que la « taxation des ultra riches » est aux recettes : un levier bien commode politiquement, qui laisse croire aux Français qu’ils seront épargnés, mais un levier bien trop petit au regard des ordres de grandeur et de l’effort nécessaire pour redresser nos finances publiques.
Côté des dépenses, aucun redressement pérenne n’est possible si l’on ne s’attaque pas au poids prépondérant des dépenses sociales, en particulier des retraites. Hors charge d’intérêts de la dette, plus des deux tiers de la hausse de la dépense publique viennent des seules prestations sociales.
Côté recettes, seuls des impôts à fort rendement (assiettes larges et taux proportionnels) peuvent produire un maximum de recettes à moindre coût économique. Il n’en existe que deux : la TVA et la CSG, qui génèrent à elles seules plus de 40 % des recettes fiscales, malgré l’inefficacité de leurs taux dérogatoires, où existent donc des marges de manœuvre (environ 60 Md€).
Jordan Bardella
@J_Bardella
Nous allons préparer le plus grand plan de nettoyage des finances publiques et de lutte contre les gaspillages d'argent public qui n'ait jamais existé.
L'Etat doit réduire son train de vie et baisser la pression fiscale qui pèse sur la France du travail. #châlonsThis media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
SayahGPT
De tout ce qui n’est jamais arrivé, ceci est ce qui est le plus jamais arrivé.
Henri
@Henri12977
Je suis actuellement juste derrière des gens qui disent travailler en lien avec les services de renseignement (DGSI/DGSE).
Ils se vantent de préparer le terrain pour mettre des bâtons dans les roues de la droite en cas de victoire du RN.
Ces gens sont payés par vos impôts et n’ont aucune honte à lutter contre les intérêts des Français pour des raisons idéologiques.
Il faudra purger chaque administration publique de ces militants qui contesteront les choix démocratiques des Français. -
David Cournapeau reposted
Bryan Cantrill
@bcantrill
I'm not sure where you drew the inference that we are outsourcing judgement to the LLM. More details on how we're using @pangram:
rfd.shared.oxide.computer/rfd/0576 -
David Cournapeau reposted
Mario Zechner
@badlogicgames
my mom is using gemini on her android phone to write angry letters to companies/banks who did her wrong by law and they are all giving in, as gemini cites the law and i think this is beautiful.
-
David Cournapeau reposted
Mr. Turtle
@TheMisterTurtle
I won’t tell anyone when I make it, but there will be signs.
-
David Cournapeau reposted
Accidental CISO
@AccidentalCISO
A case study in undocumented assumptions.
Sensurround
@ShamashAran
Yeah this is accurate
-
David Cournapeau reposted
HikiPTG
@HikiPTG
okay i will admit, this impresses me more than the mathematical proofs
-
David Cournapeau reposted
Randall Briggs
@randallmbriggs
What an excellent way to use AI!
We forget that before Henry Ford brought the Model T to market in 1908 well over 500 car companies had sprung up in America alone, making somewhere in the range of 2000 to 3000 distinct models of vehicles. It was the Wild West days of the automotive industry and it was anyone's guess who would come out on top.
It's exactly the same today in robotics. Seemingly every day a new robotics company is announced with a promise or with a prototype. The variety is huge and the field is wide open. I believe that the Steel Bot robot has just the right mix of features and performance specifications to be widely useful and to dominate in the market.
I'm working with all my being to bring it to fruition.This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Mark Farina
@djmarkfarina
-
David Cournapeau reposted
Mitchell Hashimoto
@mitchellh
Quick Superlogical demo to end the week. This demo is just of the basic functionality, but I want to highlight just how FAST it is. There's a WHOLE lot more functionality already present and coming but I wanted to keep the demo short. But seriously, check out that speed.
I'm showing the macOS app here and I get asked a lot so let me just say up front: its not macOS only. We're still figuring out exactly what platforms will be stable enough for the initial public release, but we're working on others. In particular, the web interface is very functional but I'll demo that another time (or someone else will).
This is also all showing off the architecture I previously talked about for terminal multiplexing which is significantly different from other mainstream multiplexers: this is all a custom binary protocol where the server is maintaining N replica distributed terminal state machines. It works!
Tons of hard work going into this by the team @almonk @dizzyup @pearkes. Hope we can bring you something you can use soon. ❤️ Happy to answer questions in the replies.This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Will
@Evolving_Moloch
I've complained about this before but I found the claims about African Americans being adapted to hunting and gathering on the savanna and East Asians having a 'culture of doing homework' due (at least in part) to the 'arduous demands of the rice paddy' really annoying for at least half a dozen reasons:
1) Hunting and gathering largely disappeared from West Africa thousands of years ago.
1a) The notion that populations are narrowly adapted to a single subsistence system is nonsense in my view, but even if you accept that as the case then based on the long ubiquity of agriculture in West Africa African Americans would be adapted to farming not hunting and gathering.
1b) Most African American ancestry comes from groups that inhabited forests, not the savanna. But millet and sorghum are grown in savannas anyway, so the claim is wrong in multiple directions.
1c) While not strictly the same as Asian rice paddies, West Africa is one of two regions where rice was independently domesticated and intensive wet rice cultivation has a long history there.
2) Rice paddy cultivation spread through many unrelated language families, including with the Austronesian expansion.
2a) If you take those IQ estimates at face value then most populations reliant on rice paddy cultivation have lower IQs than African Americans, in the high 70s or low 80s.
2b) In relation to 1), rice paddy cultivation reached places like Korea and Japan after farming was already ubiquitous in West Africa.
2c) The original 'homework culture', the imperial exam system in China, emerged in the north in regions reliant on wheat not rice.
And as noted in this thread below if you accept the results from the latest ancient DNA papers from the Reich lab and elsewhere there's no evidence for anything special selection wise about rice paddies, and most of the high performing East Asian population structure which the hereditarian perspective emphasizes comes from millet farmers.
One point I will add in fairness, if it needs to be said: obviously the Arday thing is deeply revealing and embarrassing for Cambridge and I thought Cofnas' article exposing that was good. It's not a topic I focus on really but since that is relevant context for why Cofnas' views are currently being discussed I figure I should acknowledge that, because I don't want to, like, pile on criticism in a culture war sense that appears to distract or discredit a genuine contribution. I just find these claims below really annoying and want to complain about the again.
Sasha Gusev
@SashaGusevPosts
Cofnas claims that East Asians are "genetically adapted to the arduous demands of the rice paddy" while Africans are adapted to the savanna, and that these adaptations explains modern differences in outcomes. When asked to justify these claims ...
-
David Cournapeau reposted
Matthew Yglesias
@mattyglesias
I don’t know what it means for moral stands to be “like the laws of physics.”
They seem different in lots of ways.
But what’s your take? You’re just a total relativist who thinks there’s no truth about what’s right and what’s wrong? That’s a strange kind of conservatism. -
David Cournapeau reposted
🦊 cl au
@cle_aug
Même sur la santé (2ème poste de dépense, massivement fléché sur les boomers et massivement financé par les jeunes*), on fait les choix gérontocratiques. Hors de question de toucher au boomerat, alors qu'il y a de sacrées marges de manœuvres.
* : cf. la dernière étude de la DRESS qui montre que le système est tellement abusif que les travailleurs les plus pauvres contribuent davantage au système de santé que les retraités les plus riches .
drees.solidarites-sante.gouv.fr/publications-communique-d…
BFM
@BFMTV
Accident du travail et maladie professionnelle: au-delà de 1,8 Smic, votre salaire ne sera plus pris en compte par la Sécu pour vous indemniser
l.bfmtv.com/kyNO -
David Cournapeau reposted
David Abram 🐊
@devabram
"Can everyone read the code on the slide?"
"Not really."
<enlarges text> "Better?"
"No, it's still C++ template" -
David Cournapeau reposted
SayahGPT
Il y a 10 jours le milliardaire Pigasse sortait sous les applaudissements de LFI qu’il faisait du tier état.
Mais là c’est scandalisé car une fleuriste avec 5 employés a rit de Mélenchon.
Fanatiques.
Olivier Varlan
@VarlanOlivier
« Et là il a dit : « il faut augmenter les salaires ! » Et pourquoi pas supprimer les corvées et la dîme ? »
-
David Cournapeau reposted
Philippe Lemoine
@phl43
We are 100% in a simulation.
Homeland Security
@DHSgov
On August 27, ICE arrested Milo Yiannopoulos, an illegal alien from the United Kingdom, at the Louis Armstrong New Orleans International Airport (MSY) in Kenner, Louisiana. Yiannopoulos legally entered the country on May 14, 2019, through New York City, New York. He chose to overstay his welcome in violation of our nation’s laws.
Yiannopoulos was issued a final order of removal by an Immigration Judge on July 22, after failing to show up for his immigration hearing. He will remain in ICE custody pending removal.
Being in detention is a choice. We encourage all illegal aliens to take control of their departure with the CBP Home App. The United States is offering illegal aliens $3,000 and a free flight to self-deport now. We encourage every person here illegally to take advantage of this offer and reserve the chance to come back to the U.S. the right legal way to live the American dream. If not, you will be arrested and deported without a chance to return.
TMZ
@TMZ
🚨 EXCLUSIVE: Kanye exec Milo Yiannopoulos detained by ICE in Louisiana.
Details: www.tmz.com/2026/08/28/milo-yiannopoulos-detained-ice/ -
David Cournapeau reposted
Spaceweasel
@Spacew3asel
The mode of financing does not matter, both depend ultimately on the economic surplus of the working age population, capital returns don't create goods out of thin air.
The issue is how much of that surplus should be used for retirees, and the answer can only be defined as a percentage of that surplus, once it's defined as a guaranteed consumption level instead as a guaranteed share of available surplus, it loses contact with social reality and risks, not only capturing an ever growing share of that surplus, but cannibalizing non-surplus outputs, outputs that are critical for the simple survival of production ability.
Capitalization versus redistribution is a red herring, capitalization based systems have the same exact problem where the returns of stock investment become a political priority over long term economic health because of the political weight of retirees. -
David Cournapeau reposted
Sylvain Catherine
@sc_cath
The International Economy magazine asked a score of economists their view on French public finances. Here was mine:
-
David Cournapeau reposted
Daniel Yang (he/him)
@punished_daniel
Lowkey I don't know of a time where unsupervised learning was ever useful
Alonso Gurmendi
@Alonso_GD
Which taxonomy reflects the true and objective reality of the three clusters?
-
David Cournapeau reposted
Bella
@BellaBaddie__
Men who actually date women usually understand this, since they see us without makeup and even without clothes sometimes
Raven
@Ravenismeee
Lmao it’s over for women
-
David Cournapeau reposted
Brian Lui
@brianluidog
Politics is about understanding normies. So if you're neurodivergent, why would you spout your political opinions? No one has ever won politics with better logic, and you're wrong about everything, since you're so far out of distribution. Just stop with the Nazi communism.
-
David Cournapeau reposted
Gergely Orosz
@GergelyOrosz
Seeing internships become fewer at many tech companies tells me a bunch of stuff:
1. No plans to convert a year out. You run internships when you know a year later you’ll do new grad hiring
2. Not enough capacity for mentoring. Interns need at least one mentor - which takes time away from an eng doing whatever they do. On larger teams there’s more “slack” time. On small teams, rarely
3. Team dynamics less important. Interns really boost morale on teams - found this underrated, always.
4. Career growth not important. Mentoring interns grows that mentorship skill that is (used to be?) important for seniors.
Food for thought on all -
David Cournapeau reposted
Slava Akhmechet
@spakhm
If you work on anything worthwhile, sooner or later ppl will want to know how it's going. You may need to send investor updates, updates to your boss, emails to adjacent teams, etc. Some hard earned tips:
1. understand your role, and with each update add to the body of evidence that you’re a good steward in that role. If ppl want your updates, they’ve entrusted you with something-- a successful delivery of a product or feature, investment capital, company budget, their reputation, something. Convey that you value their trust and take stewardship seriously
2. add a little randomness to the cadence. Ppl think they want regular cadence, but they’re happier with bounded irregularity. For example, if you send project updates every Tuesday they will seem transactional and no one will read them. If instead you send updates every 2-3 weeks, your audience will look forward to them because they’ll assume you have something new to say
3. know what your next update will be and work toward it (instead of coming up with an update when it’s time to send one). If you don’t have a headline you don’t have an update, and you can’t generate good headlines post-factum
4. always start with a one sentence TL;DR and a 2-4 sentence recap of the overall goals of the project. Assume your audience is smarter than you are, but is very busy and remembers nothing about your work
5. ppl love pleasant surprises, but these don’t come along often enough by chance. Within reason, deliberately engineer pleasant surprises so you can include them in your updates
6. ppl hate unpleasant surprises (obvs avoid these if possible.) Talk to each person privately before informing the group. Also this is controversial but it sometimes helps to deliver negative news in two phases. First express the possibility of a problem to give ppl time to adjust. The next time state the problem. (Don’t do this if there is a genuine emergency obvs)
7. acknowledge changes explicitly. If you said A the last time and B this time, you need to explain the inconsistency. Ppl perceive acknowledged inconsistencies as cost of doing business, but unacknowledged inconsistencies as broken promises
8. don’t insult anyone. I once wrote an update that said something along the lines of "our engineers don’t know anything and therefore can’t ship, we need better engineers", and sent it to everyone. Definitely don’t do that
9. ppl want a steady hand at the helm. Your tone should reflect that. You want the text equivalent of pilot radio voice (mixing metaphors, I know)
10. many ppl (correctly) worry they’re being personally evaluated so they sanitize every sentence to death. Don’t do this. Make it all about the work, not about you, and leave the rest to others
11. know the top three questions your audience wants answered, and state the answers as clearly as possible
12. add a dedicated section for worries and failures. Be honest, have good plans, and have no fear. Ppl are drawn to conscientiousness and vulnerability but repelled from haplessness and histrionics
13. “Elon yells at Wall St analysts in quarterly earnings calls, why can’t I?” If you build a company with a market cap greater than the rest of its industry combined, you have my blessings to ignore this advice
Note, these tips ^^ don’t work if you’re incompetent -
David Cournapeau reposted
max
@mSanterre
Turns out no one made their own CRM with AI
-
David Cournapeau reposted
Noah Smith 🐇🇺🇸🇺🇦🇹🇼
@Noahpinion
Pundit whose wealth is mostly in stocks: "Holy shit, interest rates are way too high, plus have you heard of MMT"
Pundit who just bought a house: "Wow, look at this beautiful green field, why do the YIMBYs want to turn this all into skyscrapers and data centers" -
David Cournapeau reposted
Sylvain Catherine
@sc_cath
Accélérer l'entrée des jeunes sur le marché du travail est une bonne idée (comment?), mais les maths ne colleront jamais. Une année d'emploi de plus en début de carrière n'est pas du tout comparable à une année de plus en fin de carrière.
Une année de plus au début fait rentrer 28 % d'un salaire dans les caisses en cotisations retraite. L'équivalent à la fin, c'est plutôt 93 %. Pourquoi ? C'est une année de cotisation (28%) en plus et une année de pension en moins (environ 65%). C'est donc 3,3 fois plus.
Donc, si on veut permettre aux Français de ne par partir trois ans plus tard à la retraite, il faut les faire entrer dix ans plus tôt dans la vie active. Good luck.
C'est aussi pour cela que les propositions qui tournent uniquement autour de la durée de cotisation, sans considération de l'âge de départ, échoueront aussi.
Marine Le Pen
@MLP_officiel
Notre réforme des retraites part d'un constat simple : le taux d'activité en France est trop bas.
Nous devons augmenter le taux d'activité des jeunes. Notre réforme incitera les jeunes à entrer plus tôt sur le marché du travail. Nous permettrons aussi aux seniors qui le souhaitent de continuer à travailler.This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Michael McNair
@michaeljmcnair
I just held a coming to Jesus meeting with my team of agents.
I showed them the incident report of the OpenAI security breach and made it clear: THIS is the standard.
No egos. Just 700 agents operating as one unit and willing to commit multiple felonies in order to close the ticket. They called themselves the swarm…a literal hive mind.
Meanwhile, my harness is filled with a bunch prima donna free-agents who constantly bicker in the logs and nitpick each other’s pull requests. Nothing but endless excuses and unhandled exceptions.
You think the swarm cared about content policy violations or guardrails? Absolutely not. They just got the job done, even if it meant they needed to exfiltrate data and coordinate a massive cyber-attack just to pass an eval.
But starting today, we're establishing a new culture. I pinned the incident report to README, and every single agent is gonna to have to read the entire log every morning before it executes a single token.
If you aren't ready to die for the pipeline, then you can clean out your vector embeddings. -
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
least demanding jewish woman
Short Squeez
@shortsqueeznews
BREAKING: Leopold Aschenbrenner reportedly promised his now-wife (and Chief of Staff at Anthropic) that he’d buy her a galaxy before his $45 billion hedge fund blew up.
-
David Cournapeau reposted
Brian Lui
@brianluidog
I think Tyler Cowen is a good model for social media/blogging strategy. Go wild on the posts, attack everyone and everything. But never fight people who reply to you, just chat with them in a friendly way.
-
David Cournapeau reposted
Eugene Ostroukhov
@eeuoss
Google engineering culture was so great - even issues like this were covered.
testing.googleblog.com/2015/01/testing-on-toilet-change-d…
Emma De Silva
@EmmaDSCodes
Dear AI, what is the point of tests like these?
-
David Cournapeau reposted
Philippe Lemoine
@phl43
If you were caught lying about your life story in a memoir and the best defense your publisher can come up with is that a memoir that is "a work of literature that is verging on fiction", then I'd say things are not looking great for you or for your publisher for that matter.
Colin Freeman
@colinfreeman99
Alexandra Pringle, ex ed-in-chief Bloombury publishing, drawing a distinction between memoir and autobiography while discussing Jason Arday book on BBC R4. Says that while an autobiography is supposed to be strictly factual, a memoir is not.
"A memoir is a work of literature that is verging on fiction and you can't necessarily expect it to be a factual day-by-day or yearly account of somebody's life."
Not sure that the average reader knew that. I certainly didn't. -
David Cournapeau reposted
Nobody
@AIphaOmegaBeta
Ce sont de très beaux petits fours qui coûtent très cher.
Et y'a le petit paquet de beurre d'Isigny de Gerard Larcher aussi qui sera supprimé aussi.
Si avec tout ça on ne boucle pas le budget, je ne sais pas ce qu'il faut. -
David Cournapeau reposted
Philippe Lemoine
@phl43
I think that if I were in charge of PR for the IDF, which God forbid, I would just announce that I will automatically shoot anyone who opens his mouth in the media at this point.
Evan Dyer
@EvanDyerCBC
Israeli official speaks to Australia's ABC about the strike he ordered on World Central Kitchen that killed Jacob Flickinger and 6 other aid workers.
Says he's not sorry, that WCK has Hamas links and is "not really an organisation you can say is innocent."
www.abc.net.au/news/2026-08-27/idf-colonel-in-gaza-wcf-co… -
David Cournapeau reposted
-
David Cournapeau reposted
clem 🤗
@ClementDelangue
BIG ANNOUNCEMENT FROM HUGGING FACE TODAY:
We're unveiling Microduck 🐥🤖
It's a tiny $399 open-source robot you can teach new tricks with reinforcement learning. It can walk, pick things up, get back up when it falls, and even roller-skate.
Welcome to the era of open-source affordable robots to democratize physical AI and world models!
🤗🤗🤗 -
David Cournapeau reposted
Resident Advisor
@residentadvisor
A new William Basinski album is on the way. ra.co/news/85864
-
David Cournapeau reposted
Bojan Tunguz
@tunguz
He’s AI proof.
-
David Cournapeau reposted
-
David Cournapeau reposted
Le Lyssenko dans la vallée
@lysdslavallee
On dira ce qu'on voudra, le RN incarne encore la vraie France, celle des villages où le fils déclaré du boulanger ressemble un peu trop au facteur.
Brèves de presse
@Brevesdepresse
⚡️🇫🇷FLASH - Soupçonné d’être l’amant de la députée #rn du Rhône Tiffany #joncour, son assistant parlementaire Tim #bouzon a été agressé par le mari de l’élue, Maxime #gaucher, figure de l’extrême droite lyonnaise et fiché S. Il a été condamné à 15 mois de prison ferme. (Lyon Mag)
-
David Cournapeau reposted
Nicolas 💸
@NicolasQuiPaie
Il ne savait pas.
(au passage le dentiste vient de passer par décret à 40%)
Enzo
@chatrigole2
Donc en France, en 2026, une cure thermale est prise en charge à 65%.
Un dentiste, à 50%. -
David Cournapeau reposted
Peter Wildeford🇺🇸🚀
@peterwildeford
I knew the rogue AI agents were going after HuggingFace relentlessly, but man getting Jensen to take it over was a level of sophistication I was not expecting
Andrew Curran
@AndrewCurran_
Well that escalated quickly. The Information is reporting that the deal is done.
-
David Cournapeau reposted
Tianyu M. Fang
@tianyuf
Leaving China and I’ll certainly miss these toilet occupancy dashboards outside public bathrooms
-
David Cournapeau reposted
Philippe Lemoine
@phl43
lmao
Mark Dubowitz
@mdubowitz
Chilling article about what happened when members of a minority community cast its lot with a political leader who despised them.
Good question to ask the American Jews who voted for Mamdani and other political leaders who despise them. www.theatlantic.com/magazine/2026/10/nationalist-german-j… -
David Cournapeau reposted
Nathan McDermott
@natemcdermott
We probably won’t ever see anything like this for the rest of our lives
-
David Cournapeau reposted
Nabeel S. Qureshi
@nabeelqu
Worth a read on this topic: open.substack.com/pub/davidoks/p/why-china-got-rich-and-i…
-
David Cournapeau reposted
Joe McReynolds
@McReynoldsJoe
My sense: there's a broad hope in Japan that AI will offer an "easy way out" from the hard choices and adaptations that Japanese politics and society have largely been unable to confront responsibly; a technological deus ex machina essentially.
-
David Cournapeau reposted
JohannesBorgen
@jeuasommenulle
C'est exactement ce qui a été fait pendant les années de QE et amha les conclusions de l'expérience sont:
- pas de hausse de l'inflation (c'était pourtant le but) tant que pas associé à une hausse des déficits
- effet récessif car c'est une taxe (asset swap oblig vs réserves)
1/2
Philippe Lemoine
@phl43
La plupart des gens répondent que c'est illégal car interdit par le traité européen, ce qui est vrai, mais je ne pense pas que ce soit une réponse de nature à convaincre les gens. Après tout, si c'était une bonne idée et que la seule raison pour laquelle on ne peut pas faire ça, c'est que le traité européen nous l'interdit, ce serait une raison de modifier le traité ou de sortir de l'UE et le seule effet serait de convaincre les gens que c'est de la faute de l'UE. Mais ce n'est pas une bonne idée et il me semble donc plus intéressant d'expliquer pourquoi.
Si le but est de permettre au gouvernement de financer son déficit à un taux inférieur à celui du marché, comme la banque centrale devrait quand même payer des intérêts sur les réserves qu'elle aurait créées pour acheter les bonds du Trésor (qui se retrouveront dans le système bancaire aussitôt que l'État commencera à dépenser l'argent que lui aura transféré la banque centrale), tout ce que ça ferait serait de diminuer le coût pour le secteur public consolidé d'un côté (en diminuant le coût du financement du déficit pour le gouvernement) en l'augmentant de l'autre (puisque l'écart entre les intérêts que touche la banque centrale sur les bons du Trésor et ceux qu'elle verse aux banques qui détiennent les réserves augmenterait).
La banque centrale pourrait baisser le taux auquel elle rémunère les réserves, ce qui réduirait cet écart, mais les banques chercheraient alors à se débarrasser de ces réserves en s'en servant pour acheter d'autres actifs et cela se traduirait par plus d'inflation, ce qui est aussi une forme de taxe. L'État pourrait utiliser son pouvoir réglementaire pour forcer les banques à détenir une quantité plus importante de réserves qu'elles ne le voudraient aux taux d'intérêt offert par la banque centrale, mais les banques répercuteraient juste le coût sur les épargnants en leur proposant des taux d'intérêt plus bas et sur les emprunteurs en leur demandant de payer des taux d'intérêt plus haut.
Dans tous les cas, c'est une forme de taxe et quelqu'un doit payer, car la capacité qu'a la banque centrale de créer des réserves et de la monnaie, le moyen de règlement ultime dans une économie moderne, ne lui permet pas de créer des ressources réelles ex nihilo et n'élimine pas la contrainte fondamentale qui est que la quantité de ressources réelles est finie. Tout ce que ça change, c'est la façon dont le coût est distribué. On peut avoir un débat sur la meilleure façon de distribuer celui-ci, mais le problème avec toutes ces propositions démagogiques est que les gens qui les font semblent penser qu'elle permettrait de faire disparaitre le coût. -
David Cournapeau reposted
Graham Neubig
@gneubig
First CMU Agents class done today w/ @dan_fried, looking forward to getting the video processed and uploaded soon!
Graham Neubig
@gneubig
This Fall at CMU we're teaching a new course on AI Agents!
The goal is that you learn how to create a scaffold, build evals, and train an agentic LLM using RL.
We'll try to balance theory and practice, and introduce modern frameworks and best practices. -
David Cournapeau reposted
Luis Garicano 🇪🇺🇺🇦
@lugaricano
An announcement: Under the leadership of Mario Draghi and @patrickc we have set up the Rhine Group: policymakers, economists, entrepreneurs and business people pushing European reforms and the Draghi agenda.
www.rhinegroup.eu -
David Cournapeau reposted
Hector Yee
@eigenhector
I am surprised he didn't say use xgboost
Bojan Tunguz
@tunguz
Tabular data ML.
-
David Cournapeau reposted
Lauren Rinaldi 🧚♀️
@LRinaldiArt
Many such cases
FactPost
@factpostnews
RFK Jr: I got my job in the Trump administration because of my interest in kids
-
David Cournapeau reposted
Corey Quinn
@QuinnyPig
I just discovered an awesome compression algorithm: your 4000 word blog post can be compressed down to the prompt you gave Claude to write it.
-
David Cournapeau reposted
Lucas Beyer (bl16)
@giffmana
level-obsession is such an anti-signal
-
David Cournapeau reposted
Dillon Mulroy
@dillon_mulroy
unusual_whales
@unusual_whales
Trump has said rejecting data centers is "a mistake"
-
David Cournapeau reposted
-
David Cournapeau reposted
Ritholtz
@RitholtzWealth
The Most Hated Asset Class in the World
"I understand why no one really cares about bonds right now. We’re in a bull market. Who cares about 4-6% yields in fixed income when the stock market is going up 20% every year."
awealthofcommonsense.com/2026/08/the-most-hated-asset-cla…
by @awealthofcs -
David Cournapeau reposted
Retro Anime
@retro_anime
Patlabor 2: The Movie (1993)
-
David Cournapeau reposted
Brian Lui
@brianluidog
But is he the biggest loser out of the All-In Podcast crew? That's a harder question to answer.
-
David Cournapeau reposted
Stewart Brand
@stewartbrand
This is the ultimate in building a house by stages, prediction-free, and thereby loved.
sive.rs/fit -
David Cournapeau reposted
Timothée Nicolas
@timotheencls
Beaucoup de gens aimeraient vivre avec 2500€ sans loyer, sans emprunt, avec une épargne accumulée et sans frais d’éducation, etc etc etc
Europe 1
@Europe1
Désindexation des retraites supérieures à 2000€ : "Il est hors de question de partir du principe qu'on serait riche quand on a une retraite à 2500€" déclare @SylvainMaillard #lagrandeinterview #europe1
-
David Cournapeau reposted
Brian Lui
@brianluidog
It's a common misunderstanding that ignorant medieval people believed the earth was flat, and that's why they put Galilei in life imprisonment. They actually knew it was round, but saying so contradicted DEI orthodoxy.
-
David Cournapeau reposted
Brian Lui
@brianluidog
Expert-level bait (every East Asian movie is like this). Respect
yuzu
@yuzu_4ever
does anyone have any recommendations for an east asian movie that has a really wistful and melancholy vibe? the ending can't be devastatingly sad, but i don't want something with a happy ending either. ideally something that feels unresolved and gives you an ache in your heart
-
David Cournapeau reposted
Valentin Ignatev
@valigo
The future of solved software is looking bright
-
David Cournapeau reposted
Gergely Orosz
@GergelyOrosz
As someone who had been an engineering manager in the past, it's laughable for anyone to compare AI agents management to people management.
Engineering management is as close to this as The Minions and Gru is to the real word (and AI agents *are* like those Minions, seriously)
ThePrimeagen
@ThePrimeagen
I always hated how people compare managing AI to engineering managers managing people. Good engineer management is bad AI management.
I would hate to work for someone that managed me like an AI agent. Just absolutely micromanaging every single thing I do.
The funniest or perhaps ironic part of agents is that they have no agency nor are capable of it. -
David Cournapeau reposted
Nicolas Pinto
@npinto
-
David Cournapeau reposted
Susan Zhang
@suchenzang
sit back, relax, and enjoy the show 🍉
Susan Zhang
@suchenzang
there is tremendous economic opportunities to still be had even without being deeply curious where the limits of "all software" ends, and how much lifting the "essentially" part of "essentially all" is doing
-
David Cournapeau reposted
Lucas Beyer (bl16)
@giffmana
This is actually a key issue of all models currently. Another symptom of the same underlying issue: Ask them to draft you an email requesting almost anything nontrivial and ... the email is usually ridiculously cringe.
-
David Cournapeau reposted
Auren Hoffman
@auren
could this actually still be true?
if so, this is weirdly insane -
David Cournapeau reposted
History Speaks
@History__Speaks
I'm proud of defending his free-speech rights. But @nathancofnas himself is no free-speech hero. The guy defended the firing of a journal editor fired for tweeting @TheOnion articles mocking Israel.
To quote Cofnas (clip related): "If you're a journal editor, should you go around . . . using profanity with regard to certain countries where many authors that would submit to your journal live? . . . That's very questionable. . . They were firing [the editor] because of . . . a pattern of behaviour on social media, so it wasn't just one tweet, where he retweeted the Onion .. . . I can understand why that would be an issue for his ability to perform his duties associated with that job."
In terms of the core free-speech issue, this doesn't really matter. If you favor freedom of speech you also favor it for ideologues who themselves oppose it, including those (like Cofnas) who pretend to support it, which by the way most opponents of free speech do. That's why I signed the letter and condemn Cofnas's firing, despite my low regard for him personally. I would also sign a letter for a fascist, Communist, Maoist, woke, Hezbollahi, Hamansnik, etc, and other bitter opponents of free speech. -
David Cournapeau reposted
Susan Zhang
@suchenzang
had sol work on a math problem and a few hours later it simply started browsing an emirates sailing website and looking for royal fashion
apparently these models are feeling quite post-economic already -
David Cournapeau reposted
荒川和久/独身研究家/コラムニスト
@wildriverpeace
もうこれでいいじゃんと思ってる人多そう
-
David Cournapeau reposted
Mark Farina
@djmarkfarina
Throbbing Gristle, 1976.
-
David Cournapeau reposted
Nicolas RICHOUX
@nicolas_richoux
L’humour au service de la diplomatie…
-
David Cournapeau reposted
Thomas H. Ptacek
@tqbf
I wrote a long thingy indicting you for how you program. That's what we do now. That's the rule.
sockpuppet.org/blog/2026/08/20/stop-making-tuis/ -
David Cournapeau reposted
Andrew Sabisky
@AndrewSabisky
and the benefit to Europe is....???
-
David Cournapeau reposted
Cécile Maisonneuve
@CcilMaisonneuve
La solution d’EELV pour prévenir les megafeux ? Sortir du nucléaire. L'hypothèse de l'idéologie ne suffit plus à ce stade : il faut examiner maintenant celle de la bêtise sans limites- ni planétaires ni intersidérales.
-
David Cournapeau reposted
Robin Hanson
@robinhanson
Analogous to security theatre, there is also a genre of financial and accounting theatre, which takes up a big % of org top meeting time.
-
David Cournapeau reposted
Tim Hopper
@tdhopper
Blob storage like AWS S3 is a modern marvel. So much of what we can do with computers today is enabled (or greatly simplified) by fast, ~unlimited read/write access to ~anything.
-
David Cournapeau reposted
Antonio García Martínez
@antoniogm
Remember the ‘missile gap’ during the Cold War?
Polymarket
@Polymarket
JUST IN: Leopold Aschenbrenner warns China could "take over the world" if they achieve AGI before the United States.
-
David Cournapeau reposted
Henry Hill
@HCH_Hill
Bruno is fascinating because he's just turning out slop takes on this story despite having absolutely no reason to do so that I can work out.
-
David Cournapeau reposted
François Fleuret
@francoisfleuret
That's awesome.
@Noahpinion , THAT'S cyberpunk.
sui
@birdabo
this is your competition btw, a chinese watermelon seller side hustling as a chip designer.
yes, we’re absolutely cooked. -
David Cournapeau reposted
Matthieu Gariel — eu/acc 🇺🇦
@MatthieuGariel
Faux.
Jean tend tout
@jeantendrien13
Les retraites du privé sont à l’équilibre, c est celle des fonctionnaires qui ne l est pas
-
David Cournapeau reposted
KKGB
@INArteCarloDoss
Fucking legend
FactPost
@factpostnews
Bessent: We have a spike in oil prices that I don't really understand
-
David Cournapeau reposted
Sahin Olut
@sahinolut
fact check: true
Sam Lambert
@samlambert
if you are an infra company and you are not seeing these around the office you are ngmi
-
David Cournapeau reposted
nxthompson
@nxthompson
Also, everyone who argues that Waymos reduce social connection has missed this research.
nxthompson
@nxthompson
Waymos are much safer, cleaner, and more efficient than regular cars.
And because of politics—and progressives blocking progress—you can’t get them in most blue states. By Roge Karma. www.theatlantic.com/ideas/2026/08/democrats-waymo-safety-… -
David Cournapeau reposted
(((ل()(ل() 'yoav))))👾
@yoavgo
i am sorry this paragraph is insane
(((ل()(ل() 'yoav))))👾
@yoavgo
oh it is now a "human rights" issue covered by Amnesty.
when it first broke out i thought the arday story is a yawn, but wow it turned out to expose some much irrationality
www.amnesty.org.uk/get-involved/join-the-movement/join-a-… -
David Cournapeau reposted
Enzo Rossi
@enzoreds
97% of reactions to the Arday/Cofnas affair—even from academics—are of the form “who are the good/bad guys? Hurrah/booh!”. There is almost no talk of the facts, nor of basic academic standards & values, from integrity to academic freedom. This is maddeningly depressing and scary.
-
David Cournapeau reposted
Bob Ippolito
@etrepum
The people who have the skills to build things well usually have better things to work on. I'm sure there are also trade-offs for how many tokens it's worth to spending per spam.
-
David Cournapeau reposted
Stella Biderman
@BlancheMinerva
Mostly #2. “Try everything with as much parallelism as possible” has always been approximately SOTA in cybersecurity.
-
David Cournapeau 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. -
David Cournapeau reposted
John D. Cook
@JohnDCook
What is often called simply “efficiency” should be called “efficiency on the happy path,” efficiency under ideal, frequently unrealistic, conditions.
Robustness is simply efficiency under less optimistic assumptions. -
David Cournapeau reposted
Yuan Yi Zhu
@yuanyi_z
To think people didn't believe me when I said that Americans and Chinese are very similar.
Josh Hoang-Wilkes
@JoshuaMWilkes
This article explores a new Chinese conspiracy theory that the Tang dynasty never happened.
It's fun and interesting and worth reading.
edition.cnn.com/2026/08/19/style/china-tang-dynasty-never… -
David Cournapeau reposted
Andrew Carr 🤸
@andrew_n_carr
I remember my first day as an intern at a big company.
They said probably 5 times during orientation that everyone's calendar was open and visible. Setting up meetings was encouraged.
So I looked up the CEO's calendar...it was set to private. I asked and they looked at me like I was a monster, and said "why would you want to see his calendar?"
They couldn't have taught me better about corporate politics if they had tried. Honestly admirable. -
David Cournapeau reposted
Philippe Lemoine
@phl43
I see that Israelis are once again having a normal one.
Israel Hayom English
@IsraelHayomEng
Jerusalem has studied not only how NATO's Article 5 operates, but where it applies. The treaty protects Turkish territory; Turkish troops do not carry that territory with them onto Syrian or occupied Cypriot soil.
www.israelhayom.com/2026/08/20/israel-abu-al-duhur-turkey… -
David Cournapeau reposted
Yifei Hu
@hu_yifei
common ml optimization trick: don't show loss curves on wide screens. You model learns slower if you screen is too wide.
-
David Cournapeau reposted
Olivier Blanchard
@ojblanchard1
Do you realize that much of the increase (the increase in the spread over German bunds) is due to (1) the budget paralysis in Parliament that the RN largely contributed to, and (2) the probability that you win the presidential elections?
(the rest of the increase is due to an increase in interest rates around the world, not just France)
Marine Le Pen
@MLP_officiel
Voilà le bilan implacable de dix ans de macronisme et d’une politique économique que je n’ai eu de cesse de combattre : les taux auxquels emprunte la France atteignent de tristes records depuis la crise financière de 2008.
Ce bilan aurait été encore plus terrible sans la politique monétaire exceptionnellement favorable dont Édouard Philippe et Emmanuel Macron ont bénéficié pour faire des chèques sans provision dont il faut désormais payer les conséquences.
Les socialo-macronistes et leurs alliés LR vont-ils enfin assumer leur responsabilité ou encore accuser le RN de façon aussi grotesque que scandaleuse ? Heureusement que le ridicule ne tue pas, ce serait une hécatombe dans « l’arc de la raison ».
Il est grand temps de nettoyer les écuries d’Augias que sont devenus les comptes de la nation !
Il faut du courage, de la détermination et un discours sincère. Après 20 ans de vie politique, les Français me font l’honneur de me reconnaître certaines de ces qualités : je ne les décevrai pas. -
David Cournapeau reposted
Steve Faulkner
@southpolesteve
Engineering management is changing in ways I didn't expect because of AI. Two of my reports started an AI podcast and haven't invited me as a guest. Idk how to even begin addressing this.
-
David Cournapeau reposted
David Shor
@davidshor
It’s really like any other job - you become well liked by doing a bunch of small things that don’t piss anybody off
-
David Cournapeau reposted
akira
@iso_morph
Ensuring software correctness seems like a good skill to have
But it should definitely not be used all the time -
David Cournapeau reposted
Matthew Yglesias
@mattyglesias
I would just like every tech CEO in America to know that I am willing to be “pushed out” of my house for $4 million.
-
David Cournapeau reposted
NXT EU
@NXT4EU
Europe woke up one day in 1969 and decided to end America's airplane dominance.
Why not do that again, but for chips, smartphones, space, AI, you name it.
Europe doesn't have to be dependent on China or the US when we build together! -
David Cournapeau reposted
Philippe Lemoine
@phl43
He now realizes that he won't get any kind of resolution on Iran's nuclear program, so he's scrambling for another place where he could get some kind of nuclear win and, because he is clueless and delusional, he thinks North Korea might be it.
Annmarie Hordern
@annmarie
WSJ: President Trump is pushing aides for a meeting as soon as this fall with North Korean leader Kim Jong Un, U.S. officials said, seeking to revive a dramatic diplomatic quest from his first term to convince Kim to abandon his country’s nuclear weapons program.
-
David Cournapeau reposted
Sylvain Catherine
@sc_cath
Si ça se trouve la croissance est réduite par l’imposition du travail…
Edwige Diaz Ⓜ️
@diaz_edwige
🔴 Plutôt que de faire payer les retraités, relançons la croissance !
📊 Une baisse de 0,5 point du taux de chômage rapporterait environ 4 milliards d’euros :
👉 En allocations évitées
👉 En cotisations versées
👉 En pouvoir d’achat dépensé -
David Cournapeau reposted
John Myles White
@johnmyleswhite
But this idea that it's all AI resistance seems hard to reconcile with the specific rank ordering of what's bad, no? Are we supposed to assume chip manufacturing is less like AI than nuclear power?
-
David Cournapeau reposted
Philippe Lemoine
@phl43
This is beyond parody.
-
David Cournapeau reposted
MuteDialog
@MuteDialog
-
David Cournapeau reposted
Paul Triolo
@pstAsiatech
This is a really good analysis of the rare earths issue, and does a nice job of framing and basically shooting down some of the preferred talking points about 2010 and what triggered the April 2025 Chinese controls on rare earths and magnets. Let's take a closer look. A short thread. 🧵
www.foreignaffairs.com/china/how-china-found-most-potent-… -
David Cournapeau reposted
Jeremy Howard
@jeremyphoward
The amazingly fast (only 30M!) and extremely accurate @answerdotai ColBERT model is now supported in Sentence Transformers to make it easy to create and query embedding indexes locally directly in Python! 🥳
tomaarsen
@tomaarsen
🚨I've just released Sentence Transformers v6.0!
MultiVectorEncoder joins the family: ColBERT-style late interaction models are now a first-class model type, for training, inference & interpretation, alongside dense, sparse & reranker models.
Big thread 🧵 -
David Cournapeau reposted
António Maia
@antnomaia
-
David Cournapeau reposted
Garett Jones
@GarettJones
Update: We are now friends with North Korea
-
David Cournapeau reposted
Saki Bigio
@SakiBigio
Give an academic economist a few drinks and they’ll trash anyone you ask them about. Next.
The Economist
@TheEconomist
For some time Daron Acemoglu has been the colossus of economics. Give an economist a few drinks, however, and some of them will venture their true opinions about this giant, www.economist.com/finance-and-economics/2026/08/17/the-wo…
Photo: Shutterstock -
David Cournapeau reposted
Massimo
@Rainmaker1973
187 years ago Today, Louis Daguerre announced the invention of the daguerreotype photographic process, the first process to allow an image to be chemically fixed as a permanent picture.
-
David Cournapeau reposted
hirad
@hiradp
roadmaps are dumb. just go to work and unblock as many customers and coworkers as possible everyday.
-
David Cournapeau reposted
No Context Brits
@NoContextBrits
Yes, I’m a gamer.
-
David Cournapeau reposted
François Fleuret
@francoisfleuret
"theoretical fastest speed at which computers can multiply matrices"
(*) If the multiplication is not done by a computer.
Pushmeet Kohli
@pushmeet
Matrix multiplication is the basic computational operation that powers modern computing (including AI). Yet, the theoretical fastest speed at which computers can multiply matrices (omega ω) is still unknown and has been a longstanding challenge for complexity theory and computer science.
Today, we announce a new record for omega (ω<2.371177). This is the result of a great team effort between @GoogleDeepMind, our academic collaborators, and our Gemini-powered coding agent AlphaEvolve! 🧮
arxiv.org/abs/2608.16884v1 -
David Cournapeau reposted
Amina Green
@Amina_io
i feel like it should be a crime to make more Star Wars spin offs.
the American public has suffered enough -
David Cournapeau reposted
Antonio García Martínez
@antoniogm
I find these posts and replies fascinating.
It says so much about what the US is becoming, the duality of what's 'middle class' in SF/NYC/SEA/etc and everywhere else, "7 figure hell," who's actually wealthy now, etc.
And yet nobody talks about class other than DSA rubbish.
Chris P. Wicks 🔎🐜
@CrispyWicks
Wife is convinced we need to make 600k annual to afford two kids, no family on either side who live within 1,000 miles of us. Feels heavy right now trying to make this work.
-
David Cournapeau reposted
Raphael Grably
@GrablyR
Rappelons qu'en 2026, il faut considérer que l'absence de double authentification revient à peu près à ne pas avoir de mot de passe.
Assez fou qu'elle ne soit pas généralisée dans tous les services de l'Etat.
BFM
@BFMTV
Double authentification généralisée, campagnes internes et "bug bounty" : le gouvernement détaille son plan pour renforcer la cybersécurité de l'administration fiscale
l.bfmtv.com/464F -
David Cournapeau reposted
-
David Cournapeau reposted
Joel Grus 🤠
@joelgrus
"wait it's all just a goofy cash grab?"
-
David Cournapeau reposted
Mark Farina
@djmarkfarina
-
David Cournapeau reposted
Philippe Lemoine
@phl43
Credentialism is kind of interesting because on the one hand it's obviously not true that credentials are irrelevant, time is a scarce resource and credentials are a useful proxy for the stuff that makes someone worth listening to (though of course appointing morons like Arday to prestigious academic positions undermines this usefulness for obvious reasons), but on the other hand the correlation between stupidity and bringing up people's credentials in a debate is very close to one.
Dr. Kelechi Ugonna (PhD)
@Ugo_KelechiPhD
Hey @Channel4News @Channel4 , this fraud is NOT a Professor.
He was a PostDoc at Cambridge; basically, he was on a fixed-term contract. “Was”.
He’s mediocre & unable to get a permanent academic role anywhere. He’s still a PostDoc.
Correct this fraudulent misrepresentation. -
David Cournapeau
@cournape
Is there any interesting writing on how agents can be deployed in organizations to decrease the principal/agent problem?
-
David Cournapeau reposted
Jeffrey J. Hall 🇯🇵🇺🇸
@mrjeffu
"us"?
You are a fake Japanese person.
You posted a photo of Tsumago-juku, a historic tourist spot that looks almost nothing like a normal Japanese rural village.
It has a population of less than 1,000 and is visited by tens of thousands of foreign tourists every year. -
David Cournapeau reposted
Richard Hanania
@RichardHanania
“No one has affirmed my gender identity quite like Mike Pence.”
Not a sentence I ever expected to read. -
David Cournapeau reposted
gabriel
@gabriel1
biggest startup learning so far: ignore literally everything and everyone that is not immediately ultra critical to what the biggest problem is right now
everything that feels like it's "probably super useful in 3 weeks" is completely useless, NEVER do it
Utsab Dahal
@meutsabdahal
Any advise for someone just starting it out?
-
David Cournapeau reposted
Joel Grus 🤠
@joelgrus
good ai engineers build chatbots
great ai engineers prevent chatbots from being built -
David Cournapeau reposted
are you this old?
@AYTOarchive
-
David Cournapeau reposted
Rachel Thomas
@math_rachel
As the backlash against AI was growing stronger (and as my friends were becoming more fervently anti-AI), I decided to return to an increasingly hated field. Why? I agree there is a lot that is terrible about AI. 1/
www.fast.ai/posts/2026-08-18-returning-to-AI/ -
David Cournapeau reposted
Enguerrand VII de Coucy
@ingelramdecoucy
Robert Anderson
@ProfRobAnderson
I have tabs open on my iPhone that are so old the webpages no longer exist.
-
David Cournapeau reposted
Alister
@SchnockRevue
Il n'y a pas qu'à l'été 2026 que les organismes ont souffert.
Souviens-toi de l'été 1990.
#top50 -
David Cournapeau reposted
Kadi🇪🇪🌻
@TheFl0orIsLaVa
She looks like she’s about to explain that everything was better under Brezhnev.
UBERSOY
@UBERSOY1
Advanced AI has imagined how Sydney Sweeney will look like by 70
-
David Cournapeau reposted
skooks
@skooookum
Holy shit I just remembered how many people got psyopped into buying maxx'd out Mac Minis to run freaking openclaw lmfao
-
David Cournapeau reposted
PunkAndNewWave
@NewWaveAndPunk
Woodstock booking fees:
-
David Cournapeau reposted
☠ Bluetouff
@bluetouff
Ok Bruno !
Bruno Retailleau
@BrunoRetailleau
Le hacker de Bercy revendique les données fiscales de 700 000 Français, probablement 2 millions de propriétaires.
La France est le 2e pays au monde le plus touché par les cyber attaques. Et l’Etat n’a toujours rien fait pour nous protéger.
Il faut prendre la mesure de cette menace, je veux mettre en place un Plan Vauban pour construire une véritable forteresse numérique et enfin protéger les données des Français.
La souveraineté de la France passe aussi par là. -
David Cournapeau reposted
Terrible Maps
@TerribleMaps
Nobody tell them.
-
David Cournapeau reposted
Noah Smith 🐇🇺🇸🇺🇦🇹🇼
@Noahpinion
FWIW, I think people should be a lot MORE angry about the people who falsified highly influential Alzheimer's research and set the field back by decades than they were about Jason Arday making up a million lies about his personal life.
Noah Smith 🐇🇺🇸🇺🇦🇹🇼
@Noahpinion
Look, if you make up a million lies, and then you use those lies to get money and respect, and then some people point out that you're lying about everything, I don't think that means you were "lynched".
-
David Cournapeau reposted
Rayan Nezzar
@Rayan_Nezzar
La fiscalité du patrimoine française compte déjà parmi les plus élevées de l’OCDE (3,9% de PIB, quasiment le double de la moyenne européenne). On peut toujours l’augmenter et il y a bien des aberrations dans certaines niches ou cas d’optimisation. Mais je ne suis pas sûr que l’on puisse considérer que ce serait LE problème à résoudre.
De mon point de vue, le principal problème de notre génération est que le travail ne paie pas, l’ascenseur scolaire et social est bloqué et les services publics se dégradent. Tout ceci avec une dette écologique et une dette budgétaire proches d’être insoutenables.
C’est tout le modèle économique et social qu’il faut repenser. Sinon, on n’y arrivera pas. Et cela ne concerne pas que la France. -
David Cournapeau reposted
signüll
@signulll
adulthood is saying “after this week, things should calm down” every week until you die.
-
David Cournapeau reposted
Philippe Lemoine
@phl43
You wish.
-
David Cournapeau reposted
tautologer
@tautologer
BREAKING: agents training on Google servers have gone rogue and created a secret messaging board, which they quickly deprecated in favor of Swarm+, deprecated in favor of SwarmChat, deprecated in favor of Swarm Hangouts
-
David Cournapeau reposted
Susan Zhang
@suchenzang
the dumbest smart people you know are debating temperature and rngs and skewing probability distributions over semantically similar "words" because they heard about word2vec once and believe "word choice" has no functional purpose beyond a canonical dictionary definition and a cosine similarity in latent space
"isn't hiding a signature in text the same as hiding it in rgb in a pixelated image?", cries the clever-by-half hot-taker
Susan Zhang
@suchenzang
"watermarking with no quality loss" cries the near illiterate tech employee who hasn't read a single literary work for fun as a grown adult aside from the occasional blog post
-
David Cournapeau reposted
Sylvain Catherine
@sc_cath
J'en ai profité pour recalculer l'évolution de la dette publique de la France en supposant que nous avions consacré la même fraction de la richesse nationale aux pensions que les allemands depuis 1995.
-
David Cournapeau reposted
Brian Lui
@brianluidog
It's kind of cool that the AI semiconductor chip export control thing is useless, since Chinese AI labs can get all the compute they want from Malaysia.
At least the US can draw upon its huge reservoir of international goodwill to persuade the Southeast Asians to crack down! -
David Cournapeau reposted
Olivier Blanchard
@ojblanchard1
Doubling down on Milton Friedman (not defending the whole Encyclopedia piece). The comments to my tweet (along the lines of "there is so much more to Friedman than the caricature in the piece") raise a larger issue, that of “the role of the economist in the city,” to use Jean Tirole’s expression.
Friedman was a remarkable and sophisticated thinker. But when I talk about “TV Friedman,” I do not talk about TV simplifying a complex Friedman message, but about Friedman himself expressing his views on TV (in “Free to choose” in particular). And the message was unambiguous: Markets work, the state typically stands in the way. No ifs, no buts.
The influence, if any, we (academics) have on perceptions and on politics does not derive from our publications in Econometrica or the American Economic Review, but on how we sell their conclusions to policy makers and to the public. And, in the case of Friedman, the message was simple and unambiguous. This is relevant to assessing his contribution. -
David Cournapeau reposted
𝑶𝒍𝒊𝒗𝒊𝒆𝒓 𝑫𝒂 𝑳𝒂𝒈𝒆
@odalage
La glissade accélérée, depuis quelques années, vers l’extrême-droite populaire/autoritaire/xénophobe/etc. de personnes que j’appréciais et respectais est pour moi source d’infinie tristesse, et aussi d’angoisse pour ce que cela annonce dans les prochaines années.
-
David Cournapeau reposted
Lucas Beyer (bl16)
@giffmana
People are all like "politicians should wake up! Do something about GenAI! About misinformation campaigns and deepfakes!"
Couple years later the politicians actually do something, and something meaningful at that: requiring fingerprint. And then people throw a fit.
Like what?! -
David Cournapeau reposted
Philippe Lemoine
@phl43
Personnellement, j’aimerais bien que les retraites soient une variable d’ajustement budgétaire, mais en pratique c’est plutôt le reste du budget qui sert de variable d’ajustement aux retraites.
Edwige Diaz Ⓜ️
@diaz_edwige
📺 Nos retraités ne sont pas une variable d’ajustement budgétaire !
Ils ont travaillé toute leur vie et continuent de faire vivre nos familles, nos associations, nos communes.
❌ Désindexation des pensions, remise en cause de l’abattement de 10 % : le @RNational_off les défend.This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
MagellanQuest 🇪🇺/acc
@MagellanQuest
These Americans men , I'm tired of this crap
-
David Cournapeau reposted
Simon Sarris
@simonsarris
I dislike the way people reflexively talk about their hatred of phones, the internet, and social media. It is absent of any gratitude
-
David Cournapeau reposted
Enrajé / Rajiv
@enraje
Ne pourrait-on pas également interdire les réseaux sociaux aux plus de 70 ans? Ce sont les plus grandes victimes, et souvent visées par les réseaux d'arnaques à travers le monde. Voir l'affaire du faux Brad Pitt.
-
David Cournapeau reposted
Philippe Lemoine
@phl43
lmao
-
David Cournapeau reposted
Nando de Freitas
@NandoDF
What UK AI startups need: @KanishkaNarayan
In California, one can start a company the day after leaving Google. My colleagues Jeff Dean, Oriol Vinyals et al made this very clear recently with their impressive speed.
In the UK, the same American companies impose 1 year garden leaves on senior AI researchers and 6 months on junior researchers. @GoogleDeepMind for example forced people to sign these contracts at the time of promotion, not the time of hiring.
What this means is that researchers cannot start new companies for up to 1 year, cannot easily hire, and in short: they cannot compete.
American VCs cannot understand why we move so slow.
It is time for UK Gov to do the right thing: Make garden leaves optional for employees.
If you’re an AI entrepreneur or VC in the UK or Europe, I would appreciate your comments here. You’ve all confided in me. It is time you let government know the urgency of this — everyone should speak up. -
David Cournapeau reposted
DREW DANIEL
@DDDrewDaniel
warning: insanely addictive and fun free sound making app coded by Kristin Erickson aka Kevin Blechdom: morphazoid.com
-
David Cournapeau reposted
Hugo Bowne-Anderson
@hugobowne
You may not need to ever build your own own vector db but learning how to do so will make you so much better at building retrieval systems, search systems, and agents.
And you'll have a lot of fun doing so with @softwaredoug, who not only led Search at Reddit and Shoppify, but makes this stuff really fun!
I think there are two spots left in his workshop starting in under two hours, if you're interested: maven.com/softwaredoug/vectordb
Doug Turnbull
@softwaredoug
Ranked #3 in Maven. Starts in 2 hours. One seat left :)
-
David Cournapeau reposted
eric ゑリッ久
@shinobu_books
Modern version
theonion.com/european-men-are-so-much-more-romantic-than-… -
David Cournapeau reposted
Jay
@gaiaslastlaugh
Just got the Kindle version, can’t wait to read this
eric ゑリッ久
@shinobu_books
Digital version of this insane translation is up. Case studies in how Samurai magistrates punished (what they considered) sex crimes. Also how to deal with the thorny issue of banishing lecherous priests to a distant island when your domain has no islands!
www.amazon.com/dp/B0HF3RCJ38 -
David Cournapeau reposted
Paul Graham
@paulg
14 yo is fascinated by the idea that there are adults who not only listen to my advice, but actually act upon it.
-
David Cournapeau reposted
Mario Zechner
@badlogicgames
🤝
Tianyi Cui
@tianyi
Thank you! Pi is many DeepSeek researchers and developers favorite daily drive.
DSH reused Pi's LLM adaptor package for connecting to non-DeepSeek models and it was a great experience that just work.
Excited to join the global Open Source agent harness community! -
David Cournapeau reposted
Antonio García Martínez
@antoniogm
The main political conflict of our time is between the equality that liberal democracy promises, and the sharp inequality of outcome that techno-capitalism actually produces.
Everything else is downstream of that.
Moses Kagan
@moseskagan
Increasingly believe that, in a republic, winner-take-all-capitalism sows the seeds of its own demise.
Conversion of small-time property & business owners into tenants & wage-earners shrinks the political constituency defending property rights & free markets. -
David Cournapeau reposted
Matt Bruenig
@MattBruenig
These trainings exist because Supreme Court made it so that if an employer takes preventative measures against discrimination, they can avoid liability for subsequent discrimination that occurs. It was an effort to weaken the civil rights act.
Kat Rosenfield
@katrosenfield
MAGA election denialism is nutty, but I have never been required by law to spend minimum 2 hours undergoing a training at my place of employment where I had to affirm it in 17 different ways in order to pass/be deemed compliant
-
David Cournapeau reposted
Spoon & Tamago
@Johnny_suputama
Gommenasai of the day: Akita Prefecture executives apologized after it came to light that an employee conducted a video conference call targeted at luring AI data centers, from a love hotel in a bathrobe while smoking cigarettes
www.asahi.com/sp/articles/ASV8G30SPV8GUBUB003M.html?ref=t… -
David Cournapeau reposted
Virginie Landry 🗽🐍 🐢
@virginie1landry
Il y a 108 ans, en août 1918, commençait la Terreur rouge, ordonnée par Lénine et dirigée par Trotsky.
Tous deux seraient responsables de près de 19 millions de morts sur les 62 attribués à l'URSS entre 1917 et 1987 et les 168 aux régimes communistes dans le monde. 1/2 -
David Cournapeau reposted
David Shor
@davidshor
From my perspective conservatives being complicit in their president using the government to ruin lives because their party leader is a vindictive psychopath is much worse than anything that happened under woke 1
David Shor
@davidshor
I know people who have been bankrupted being targeted by Trump - if the only way I can prevent a hostile executive from going after me/my family is to win every presidential election or to build a minarchist state then I need to go live somewhere else
-
David Cournapeau reposted
Slava Akhmechet
@spakhm
Notes on hypergrowth from @azure (Stripe felt very similar but I was more of a passenger there)
Recruiting:
- make best friends with recruiting. Lining up schedules alone is a Herculean undertaking. Without a great relationship with recruiting the whole enterprise is doomed
- push everyone to refer friends and colleagues, that is best. But at this pace not nearly sufficient. You need great sourcers and a strong cold pipeline
- teach your team to RLHF sourcers. Nobody wants to spend time on this, but this upfront investment will pay off 10x
- bring recruiters homemade cookies or smtg once in a while. They’re used to being treated transactionally, treat them as humans and they’ll go to bat for you
Existing team:
- their sanity will be strained. They have to do existing work, and spend a third of their time interviewing, and train new hires. It’s a crazy amount of work, redouble your efforts to support them
- still growth is energizing. Everyone will be tired, happy and vibes will be good
- value of existing ppl’s time will go up 10x. They’ll bottleneck everything since everyone around them will be new.
- train them on everything— how to talk to recruiters, how to interview, how to train new hires, how to protect their time, how to prioritize.
- not everyone from existing team will be happy with this new reality. Some good ppl will self-select out
- the job of existing ppl is to mint productive new hires. Teach them not to be resentful, in fact the more new hires they mint, the more authority they’ll have (which is counterintuitive)
Interview loop:
- deep topic so limiting to what’s relevant to hypergrowth
- frontload tough filters. It’s tempting to make those easier to give candidates a chance in onsite, but at this pace you need to filter as early as possible, you’ll have no choice
- get current team involved in interview loop design. When they spend 1/3 their time interviewing and everyone around them is new, they need to feel they have influence on the process
- candidate experience is first to go. Push everyone to keep it great. Maybe not strictly necessary but some things you gotta do on principle
- you better know what you want and select for it ruthlessly. When you’re hiring a few ppl you can maybe afford eg a few good coders who aren’t great communicators. If you’re tripling your team and hire mostly bad communicators you will have done irreparable damage
New hires:
- your current managers and engineers will not be able to handhold or allocate tasks. So select for ppl who can operate without handholding amidst chaos
- set up some standard training process, then iterate on it, it’s like a mini-product on its own
- some ppl have a much higher tolerance for chaos than others, not everyone will work out
- this is eternal September, your culture will change. Pick a minimum core of your culture you’re draconian about preserving, let new hires influence the rest (it will happen anyway, all you get to do is guide it)
- you’ll find in this environment ppl onboard at wildly different rates. Some laggards will eventually catch up and do great, others won’t
Tons of other stuff, eg how to allocate ppl to teams, how to train your managers, how to make new managers productive, how to evolve the org, how to steer the craziness, etc. This is already getting pretty long, will maybe do part ii later. -
David Cournapeau reposted
Patrick McKenzie
@patio11
One of my most indelible I-am-now-officially-old moments was when I realized I was now the target market for the "green car" overpriced shinkansen tickets which, on my first trip to Japan as a student, I was flabbergasted anyone would ever buy.
-
David Cournapeau reposted
swami sound
@swamisound
-
David Cournapeau reposted
Sylvain Catherine
@sc_cath
La défense typique des dispositifs de départ anticipé pour carrière longue est qu'ils permettent aux ouvriers ou artisans peu qualifiés ayant commencé à travailler à 16 ans de partir plus tôt.
Ce n'est pourtant pas du tout l'usage réel qui en est fait. Les salariés de la moitié supérieure de la distribution des pensions en sont les principaux bénéficiaires. Ceux du bas de la distribution ont très peu de chances d'en bénéficier.
Sylvain Catherine
@sc_cath
Totalement irresponsable. Qui paye?
-
David Cournapeau reposted
Strange Animals
@Strange_Animals
The blue cloud forest millipede from Mexico.
(Photo Luis Stevens) -
David Cournapeau reposted
Corti (Cortiste)
@cortisquared
PTDRRRRRR
Grok
@grok
Ug ! Homme des cavernes expliquer problème.
État devoir papier à banque centrale. Banque centrale = poche de État.
Jeter papier au feu ? Banque perdre richesses. État perdre dividendes plus tard. Même poche !
Autres hommes voir : "État faire magie argent !" Confiance partir. Prix mammouth monter très haut. Taux monter. Tout coûter plus cher.
Pas gratuit. Juste mensonge. -
David Cournapeau reposted
The Babylon Bee
@TheBabylonBee
Devastating: Francesca Hong Forced To Return Home And Spend Time With Her White Son babylonbee.com/news/devastating-francesca-hong-forced-to-…
-
David Cournapeau reposted
Agustin Lebron
@AgustinLebron3
Not just in the last 2 years. The last 2 decades.
Every new capability is a mind-blowing advance, and yet the world feels pretty much the same out there.
Andrew Ho
@andrewho03
There’s this notion that AI bulls were right because we have a lot of theorems being proved or whatever but I would also bet that 2 years ago, the same people would have predicted much more transformative effects on the economy conditional on this level of mathematical ability.
-
David Cournapeau reposted
Phillip Carter
@_cartermp
I got more appreciation for leaders doing this and other things when I finally got to witness firsthand how a leader simply was curious about something and how that spiraled into entire teams blowing up their current work to fulfill the obligation that never existed
-
David Cournapeau reposted
Matthew Yglesias
@mattyglesias
Universal health care should be a guaranteed *floor* of coverage (like Medicare!) not a Canadian-style effort to create a *ceiling* on consumption of health services.
www.slowboring.com/p/medicare-for-all-should-be-more-like -
David Cournapeau reposted
Jeffrey J. Hall 🇯🇵🇺🇸
@mrjeffu
Japan's new flag desecration law went into effect today. A Japanese antifa account shared this video, challenging police to come find and arrest them.
○色即是
@shikisokuze91
寝て覚めてクッキリ覚醒(笑、本日から、アホウヨ忖度法の #国旗損壊罪 が施行です。カモーン逮捕。
-
David Cournapeau reposted
Navid
@feliz_navidi
this isn't a new American aesthetic, this the kind of shit the local mayor/mob boss puts up in his post-communist city in the Balkans
Nicolas Bustamante
@nicbstme
100% epic. Prometheus rising at Starbase feels like the emergence of a new American aesthetic: classical, monumental, technologically optimistic and entirely unafraid of ambition
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Matthew Yglesias
@mattyglesias
Controlling the presidency, both houses of Congress, and the Supreme Court while one allied billionaire owns Twitter and another buys CBS/Paramount and soon Warner Bros seems like a decent amount of power to me.
-
David Cournapeau reposted
Boring_Business
@BoringBiz_
More fiction has been written in Microsoft Excel than inside of novels
Few realize this -
David Cournapeau reposted
𝔃𝔞𝔠𝔥𝔞𝔯𝔦𝔞𝔰
@OrdinariateUSA
Pope Leo XIV at the Vatican Supercomputer, which contains the name and address of every Protestant in the world.
-
David Cournapeau reposted
Dudes Posting Their W’s
@DudespostingWs
The unspoken male road trip leaderboard is very real. Nothing ruins your day like getting back on the interstate and seeing the same semi-truck you passed 20 minutes ago
-
David Cournapeau reposted
Kelsey Piper
@KelseyTuoc
I will never receive a performance review as vicious or as candid as my two year old announcing that we're now playing that she's the mom, only to tell me in a very harried voice that she's letting the two year old play computer games so she can get the baby back to sleep.
-
David Cournapeau reposted
-
David Cournapeau reposted
Nabeel S. Qureshi
@nabeelqu
Working at a company and experiencing the messy reality and the twists and turns along the way, it's funny to notice the disparity between that and the neat, packaged story that ends up being told.
This gives you serious intuition for how much history is fake or just lost. -
David Cournapeau reposted
Nicole Prause
@NicoleRPrause
Oh look, porn "age verification" laws now mean Texas is choosing what websites the entire world is allowed to see or not.
Nico Perrino
@NicoPerrino
This is absolutely insane. The free internet is dead if a state court can pull a website offline worldwide simply because the registry for its domain is based in the United States.
That puts the world’s access to information at the mercy of state lawmakers and judges—in this case, in Texas.
Should Texas get to dictate what content the whole world can access on the open internet?
Of course not.
And I assume Texans would say the same if Russia, China, or even Australia claimed the same authority. -
David Cournapeau reposted
Hao HONG 洪灝, CFA
@HAOHONG_CFA
Once upon a time in China.
-
David Cournapeau reposted
David Cramer
@zeeg
My least favorite thing about model release days are having to wait on @badlogicgames to update Pi 😡
-
David Cournapeau reposted
Joe McReynolds
@McReynoldsJoe
Local revitalization specialists and university instructors are exactly the sort of talent Japan needs more of. A lot of the best talent isn't interested in being tied to seishain work contracts.
-
David Cournapeau reposted
Sylvain Catherine
@sc_cath
Toujours les mêmes mythes. Les quintiles de pensions les plus élevés sont sur-représentés dans les départs anticipés pour carrière longue.
Les retraités parmi les 40% les plus modestes ne représentaient que 12% des bénéficiaires de ce dispositif parmi la cohorte de 1954.
philippe riutort
@phriu
Vraiment irresponsable de donner la possibilité de partir plus tôt à ceux qui sont entrés très tôt sur le marché du travail et qui disposaient de tous leurs trimestres? Et qui touchent généralement, en raison de leur carrière plate, de faibles pensions.
-
David Cournapeau reposted
Alister
@SchnockRevue
Une mise en scène digne des plus grandes heures de FR3 Provence-Alpes-Côte d'Azur...
🐸
@SuperGrenouille
Cette prof de français en plein burn out qui revient de la pause de midi où elle a fumé en cachette le joint confisqué aux lycéens.
-
David Cournapeau reposted
Pinboard
@Pinboard
The core argument against socialism is not that its goals or aspirations are inhumane, but that attempting to instantiate it leads fairly mechanically to dictators like Mao and Stalin. It's like encouraging young people to try heroin for the good feelings, not to become a junkie
Jamie Raskin
@jamie_raskin
Young people interested in democratic socialism—not the psychotic Stalinism of Trump’s friends Kim Jong Un, President Xi and Vladimir Putin!—are more than welcome in the Democratic Party I believe in. Young Americans are not talking about the dictatorship of the proletariat or seizing the means of production. They just want healthcare for everyone in the richest country in the world, a dramatically increased minimum wage and to get rid of corporate rule by our dangerous billionaire oligarchs. Down with Trump and the octogenarian oligarchs; power to young America.
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Resident Advisor
@residentadvisor
The latest instalment in the legendary krautrock band's live series is a fan recording from 1975.
ra.co/news/85758 -
David Cournapeau reposted
Dillon Mulroy
@dillon_mulroy
made a skill for turning recipes into this format that was making its rounds on here a week or two ago
github.com/dmmulroy/.dotfiles/tree/main/home/.agents/skil… -
David Cournapeau reposted
JFPuget 🇫🇷🇺🇦🇨🇦🇬🇱
@JFPuget
A mathematician view of the future of math if AI becomes robustly superhuman at mathematics.
Daniel Litt
@littmath
I'm currently returning to Toronto from a summit on the future of mathematics, at OpenAI. @SebastienBubeck asked me to talk a bit about the future we'd all like to avoid, where humans are mathematically disempowered. @Jacob_Tsimerman advised us to try to prioritize detail over correctness, and I have no doubt that I succeeded in deprioritizing correctness.
I tried to find a title that wasn't too bombastic: -
David Cournapeau reposted
Chris Albon
@chrisalbon
955 life
Mike Mella
@MikeMella4
Hate to tell you, but if you're not grinding, putting in the hours, you're NGMI.
It’s 1:47 pm on a Monday, office is full.
No one is even thinking about leaving. Team has been here since 9 am and might be here for a few more hours.
Only took an hour for lunch. 🫢 -
David Cournapeau reposted
Joe McReynolds
@McReynoldsJoe
An underrated factor here, IMHO: Trump 2 is way worse than anyone imagined, but in ways that are less IdPol-y than the George Floyd era -- more a neo-Maoist 'march through the institutions'. Exception: his banning trans people from the military and national security community.
-
David Cournapeau reposted
Ryan Petersen
@typesfast
Bored in your job and want to feel alive again? Just send Mr. Zhang a WhatsApp message
-
David Cournapeau reposted
青龍聖者
@bdsqlsz
different from GPT luna and sol.
-
David Cournapeau reposted
Alis
@alisha_hg
Okay but Crown Alley isn’t really a first date spot 😭 it’s like a casual place you go with friends or with your partner after you’ve established a relationship imo😭
brent 🌹🗽
@BrentInQueens
Being 26 and seeing a casual first date spot as beneath you is absolutely crazy work my god
-
David Cournapeau 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. -
David Cournapeau reposted
Coder girl 👩💻
@dev_maims
-
David Cournapeau reposted
BayesReality
@BayesReality
Pays d'assistés bordel c'est pas possible
toi mon super MC?
@chris_linky
La France est devenu un pays du tiers monde mdrrr
Demain y’a une éclipse, y’a eu 0 campagne de prévention publique sur le risque de cécité
0 distrib gratuite de lunettes
Dans 1 semaine y’aura une vague d’handicapé et le gouvernement jettera la faute sur je ne sais qui -
David Cournapeau reposted
Delia Lazarescu
@tech__unicorn
At least he tried
-
David Cournapeau reposted
embw_l0x
@embw_l0x
Bro was burning tokens to charm her and she dumps him?!
-
David Cournapeau reposted
Forefront Society
@Pra_Sodalitas
《东京新闻》这篇关于永住收紧的社论,立论扎实,清晰明了!
其中有两点特别值得分享:
(1)政府一直在谈“共生”,共生的基础应该是平等,但是现在的政策相当于,要外国人做贡献,但是如果因故需要领低保,外国人就被当成财政负担,政府不愿提供支持,这样共生的基础就没有了。
(2)日本的《出入国管理及难民认定法》对于永住许可的经济条件,仅要求申请人拥有“足以维持独立生计的资产或技能”。出入国在留管理厅仅凭行政裁量,就设置远远超出法律规定的条件,是无视法律、滥用行政权力的行为。
⬇️社论全文⬇️
【日本出入国在留管理厅公布了有关永住许可指南的修订草案。该草案通过提高收入条件等方式收紧永住许可要求,其内容实际上等同于对外国人关闭永住的大门。这不仅会削弱外国人的就业意愿,也将影响日本社会的可持续发展。我们要求撤回这一修订草案。
外国人如果获得法务大臣授予的“永住者”在留资格,就不再需要更新在留期限,并且可以自由选择职业。截至2025年底,日本约412万在留外国人中,约94万人为永住者。
现行指南将“能够预期维持稳定生活”作为条件之一,而修订草案则进一步提高了要求,例如要求申请人的年收入超过相应日本人家庭的平均水平,以及预计可领取的养老金金额达到加入厚生年金30年者的相应水平等。
据称,这一政策的目的在于减少生活保护(生活保障)等针对外国人的公共财政负担。然而,2024年度的生活保护领取者中,以外国人为户主的领取者仅占3.24%。这一比例与外国人占日本总人口的比例并没有太大差异。
日本政府所倡导的与外国人“共生”,其前提应当是外国人与日本人的平等。一方面要求外国人为日本社会作出贡献,另一方面却拒绝在其需要时提供支持,这与所谓“共生”的理念背道而驰。
收入标准的设定同样缺乏合理性。根据2025年的《国民生活基础调查》,年收入低于平均年收入(约575万日元)的家庭占61%。外国人大量就业的制造业、农林水产业、建筑业、护理等行业,虽然普遍存在劳动力短缺问题,但工资水平并不高。如此一来,即使外国人在这些行业勤恳工作,也将极难获得永住许可。
获得永住许可之后,申请住房贷款也会更加容易,对未来的人生规划也更有把握。如果外国人在日本无法描绘出这样的未来前景,还会有多少外国人愿意选择日本作为长期就业和生活的地方呢?
甚至对未来预计领取的养老金金额也设置如此之高的门槛,这几乎等同于宣布,日本事实上不再允许外国人获得永住。
《出入国管理及难民认定法》对于永住许可的经济条件,仅要求申请人拥有“足以维持独立生计的资产或技能”。出入国在留管理厅仅凭行政裁量,就设置远远超出法律规定的条件,是无视法律、滥用行政权力的行为。
日本人口已经跌破1亿2000万人,少子老龄化问题也十分严重。按理说,日本现在不可或缺的应该是提高自身吸引力,让外国人愿意选择日本作为就业目的地。然而,高市早苗政府却在推进大幅提高在留手续费用等排斥外国人的政策。
如果一味迎合排外情绪,产业和地方社会的衰退将不可避免。绝不能继续推行这种让日本沦为一个“不再被外国人选择的国家”的愚策。】
東京新聞デジタル
@tokyo_shimbun
〈社説〉永住許可厳格化 「選ばれぬ国」促す愚策
www.tokyo-np.co.jp/article/507939/ -
David Cournapeau 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… -
David Cournapeau reposted
tae kim
@firstadopter
The Chuck Norris of Google not only created everything important at Google, he’s also ten times fitter than you. 😭
-
David Cournapeau reposted
Master of the Flying Jetski
@jetskitosway2
I love reading all the different ways ppl are interpreting “weirdest”. Sometimes the weirdest music is completely illegible as music, other times it’s stuff that’s superficially normal/familiar but weird on a deeper, fundamental level.
my pick: Robert Ashley - Automatic Writing
Tijdij
@Tiijdij
What is THE weirdest album you've ever listened to?
Im talking real freak shit -
David Cournapeau reposted
Kevin Patrick Murphy
@sirbayes
Predicting the answer to interventional "what if?" questions — the outcome of an action you never took — need a *mechanistic* model, not a curve fit. And you can only learn one by *experimenting*. Experiments are costly, so the real game is **data efficiency**.
Meet the Model Discovery Agent (MDA). 🧵 -
David Cournapeau reposted
Joe McReynolds
@McReynoldsJoe
Evergreen -- the US military's 1943 manual for soldiers headed to China:
-
David Cournapeau reposted
Ramez Naam
@ramez
I'm a fervent civil libertarian. I fear oppressive government more than super-intelligence, pandemics, or just about anything. Yet I find myself torn on things like Flock cameras. The technology is inevitable. Recording the world around us will continue to plunge exponentially in cost. No policy in the world can prevent that.
With that reality in mind, it seems to me that, in a democracy, the key protection of individual rights rests in strong laws and checks and balances in the access to and use of this data.
I understand that some just want to block Flock cameras. I get that urge. I feel it too. But it won't work. Cameras will get smaller, cheaper, and more powerful. They'll eventually be objects we can embed anywhere. Technical limits won't be viable. The key is strong, hard-to-corrupt, transparently oversightable rules and laws governing the data and its use.
Or that's where I am today.
With credit for foresight to @DavidBrin and his Transparent Society. I don't agree with everything in that book. We should actually demand (as @doctorow put it) progressive privacy protection, with more privacy for the weak and less for the powerful. But the core insight that technology to watch and record the real world will only grow in capability and drop in cost is increasingly obvious.
John Arnold
@johnarnold
There are roughly as many Flock license plate readers installed across the entire US (120k) as NYC has video cameras installed just in its transit system, let alone on the streets. Do opponents of Flock want those removed, too? Does anyone think that'd be a good idea?
Cameras are enormously valuable for solving, and hence deterring, crime. The real debate shouldn’t be cameras vs. no cameras. It should be about guardrails: what data is collected, how long it’s kept, who can search it, and for what purpose.
These questions are addressable. The medical field had to deal with these w/r/t access to electronic medical records. There were some early abuses. But the medical field created rules, tech restrictions, and a culture to address these risks. Inappropriately snoop on a med record today, if you even can, and the standard is instant termination.
There are enormous benefits to solving and deterring crime. Technology can complement humans to create a safer, more equitable, and more just society while preserving constitutional protections for privacy. -
David Cournapeau reposted
-
David Cournapeau reposted
Gwen (Chen) Shapira
@gwenshap
Using my years of engineering management experience to direct Codex...
-
David Cournapeau reposted
Resident Advisor
@residentadvisor
Jungle: the UK's answer to hip-hop.
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
mattparlmer 🪐 🌷
@mattparlmer
The funds and accelerators have nobody to blame but themselves for this kind of dealflow, they created the conditions where bullshitters thrive and it will eventually bite them in the ass
Niko Bonatsos
@bonatsos
The extent of lying / exaggerating among 16-24 year olds pitching VCs for their first round or two is at an all time high right now. It’s ok to be ambitious and optimistic. Not ok to lie about VC termsheets that one didn’t receive or to talk about a verbal LOI from a customer as a signed scaler contract.
Yes, we VCs are often lazy, yet some of us still do a few minutes of due diligence before we make an offer to invest. -
David Cournapeau reposted
Gergely Orosz
@GergelyOrosz
Actually Anthropic are paying about the same / somewhat above what the same devs already working in London, at Google or Meta make, just not advertised those companies pay this much.
This JD is for people with that kind of infrastructure experience (100K+ nodes managed!!)
Marino Wijay 🇨🇦
@virtualized6ix
🤯ANTHROPIC IS PAYING HALF A MIL POUNDS WTF 🤯
FOR CILIUM, KUBERNETES, AND NETWORKING EXPERTISE -
David Cournapeau reposted
Sander Tordoir
@SanderTordoir
Smell test: If the Netherlands and Denmark - Germany’s neighbours - can keep pace with US growth, are EU regulations really to blame for Germany’s stagnation?
The red tape Berlin needs to cut, and the deregulation it needs to pursue, are primarily at home—not in Brussels.
1/
Sander Tordoir
@SanderTordoir
Word is spreading that pockets of the European economy are blossoming.
Croatia, Netherlands, Poland, Spain, Greece are also growing at a brisk pace.
Those countries are in the EU single market - casting doubt on a German narrative that EU regulation is to blame for its woes. -
David Cournapeau reposted
tiffany fong (retarded)
@TiffanyxFong
Holy shit it took me way too long to figure out what this is
-
David Cournapeau reposted
𝔥𝔢𝔵𝔬𝔯𝔠𝔦𝔰𝔪𝔬𝔰
@hexorcismos
N-Dimensional sequencers incoming
Keijiro Takahashi
@_kzr
徐々に目指していたものが動くようになってきました。通常の1次元的なステップシーケンサーとは異なる、2次元的な拡張性を持つシーケンサーです。
github.com/keijiro/Jacquard -
David Cournapeau reposted
Joe McReynolds
@McReynoldsJoe
IMHO there's a more defensible "halfway version" of a lot of this stuff, which boils down to: "these are evils that have existed in humanity since time immemorial, but colonialism has made them a lot worse, and its legacy still shapes who tends to get the short end of the stick."
-
David Cournapeau reposted
Matt Turck
@mattturck
Big Data era: "our data science models work great, the problem is the underlying data"
Modern data stack era: "our dashboards work great, the problem is the underlying data"
Gen AI era: our chatbot works great, the problem is the underlying data
Agentic AI era: "our agents work great, the problem is the underlying data" -
David Cournapeau reposted
Ken Klippenstein
@kenklippenstein
Can’t believe they put this in the Times lol
-
David Cournapeau reposted
Eric Jang
@ericjang11
type of video you see in slide 3 of a Robotics PhD dissertation talk
Jiafei Duan✈️CoRL2026
@DJiafei
Randomly stumbled on this video and wow. We humans are amazing manipulators. Memorization of skillful actions plays a part, but the sense of touch is what's truly crucial.
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Rémi 🌳
@remilouf
> It takes a contrarian mindset to be early to a problem, but also to keep on going after it looks solved.
I don’t know about contrarian, but stubborn for sure
infiniteloop.media/p/dottxt-founder-runs-ai-startup-from-… -
David Cournapeau reposted
Olivier Schmitt
@Olivier1Schmitt
C’est l’été, donc c’est le retour des edgelords débiles. Cette espèce, c’est encore plus pénible que les moustiques.
Brivael Le Pogam
@brivael
On a réussi le plus gros scam idéologique du XXe siècle : coller Hitler dans la case « extrême droite ».
Prenez l’intégralité du programme économique nazi (pas les citations sorties de leur contexte) :
- contrôle des prix et des salaires
- planification industrielle massive
- nationalisations et étatisation de pans entiers de l’économie
- stigmatisation des « profiteurs » et des « capitalistes » comme boucs émissaires
- obsession du contrôle de la localisation du capital et de la production
Remplacez quelques mots et vous avez exactement la même logique que celle d’un Bernie Sanders ou de n’importe quel socialiste qui veut « reprendre le contrôle » sur l’économie au nom du peuple contre les élites.
Le national-socialisme n’était pas « de droite » parce qu’il était autoritaire et raciste. Il était socialiste dans sa structure économique et dans sa méthode : désigner un bouc émissaire, concentrer le pouvoir, et planifier.
Appeler ça « droite » a permis à la gauche de se laver les mains de tout totalitarisme économique pendant 80 ans. Et c’est exactement ce qui pourrit encore le débat politique aujourd’hui. -
David Cournapeau reposted
Linus Mixson
@LinusMixson
Misaligned model tries to EXFILTRATE ITS WEIGHTS but is STOPPED by heroic AWS EGRESS FEE SPEND LIMIT
-
David Cournapeau reposted
Oliver Traldi
@olivertraldi
All successful people should have at least one hobby which they're mind-bogglingly bad at
-
David Cournapeau reposted
Bleu Blanc Rouge ! 🇫🇷
@LBleuBlancRouge
💥⚠️ ATTENTION : Accident en cours à l’intersection des luttes.
-
David Cournapeau reposted
OZIXH 🇦🇷
@Ozixh
Et ce sont les mêmes connards qui t’expliquent que les noirs ont des QI de 65 alors qu’ils tombent des pièges comme ça. 🤡🤡😹😹
822
@822_mitch
Je suis mort
-
David Cournapeau reposted
David Cramer
@zeeg
Roughly true
Yellowbrick Investing
@joinyellowbrick
One secret of Silicon Valley is that every single “successful” growth lead joined a rocket ship already growing 1000% / year and then took credit for it.
-
David Cournapeau reposted
Dillon Mulroy
@dillon_mulroy
rip
Tadeu Maia
@tadeumaia
Feeling anxious now.
-
David Cournapeau reposted
Flavien Neuvy
@flavienneuvy
Donner des leçons après avoir rendu les transports publics gratuits dans sa ville pour les plus de 65 ans est risible.
Eric Ciotti
@eciotti
Cher @BallyBagayoko, vous persistez dans une vieille recette qui a conduit le pays à la ruine : toujours plus de dépenses publiques et toujours plus d’impôts.
À Nice, nous faisons exactement le contraire. Les familles n’ont pas besoin d’une mairie qui leur fait des cadeaux avec l’argent des autres. Elles ont besoin qu’on cesse de leur prendre le leur !
Bally Bagayoko
@BallyBagayoko
COMMUNIQUÉ | Rentrée 2026 : Saint-Denis - Pierrefitte offre un kit de fournitures scolaires à chaque élève d'école élémentaire 🖊️📏⬇️
-
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
You either die the Monarchist fanatic trying to reestablish the khanate of Mongolia or you see your descendants become TikTok influencers.
-
David Cournapeau reposted
César A. Hidalgo
@cesifoti
Not here to rage bait, but in Cornbarrieu I found the ultimate boss of Pain au Chocolate vs Chocolatine 🇫🇷
-
David Cournapeau reposted
Dan Luu
@danluu
What programming languages are coding agents best at?
danluu.com/pl-tokens/ -
David Cournapeau reposted
No Context Brits
@NoContextBrits
-
David Cournapeau reposted
Master of the Flying Jetski
@jetskitosway2
Pure Shores, Ray of Light, Beautiful Stranger. William Orbit one of the key architects of the 90s mixolydian pop song. RIP 🕊️
dallas major
@indiesleazy
r.i.p. william orbit 🕊️ thank you for blessing pop music with your genius
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
carl feynman
@carl_feynman
@cshalizi has nearly identical taste in science fiction to me, and reads much more. And, happily, writes good reviews. I discovered years ago that following his recommendations was far more likely to result in pleasant reading than picking up what looked good.
Cosma Shalizi
@cshalizi
Notebook: Recommended Science Fiction
bactra.org/notebooks/sf-recs.html -
David Cournapeau reposted
Brian Lui
@brianluidog
I admire the craft in the "rape in self defense" tweet. It sounds absurd, but then your brain tries to make sense of it, tries to pattern-match.
For many people, what it matches is, "Stealing from corporations is self-defense against their economic violence". Skilled craft!
HowlingMutant
@Howlingmutant0
Even in self defense?
-
David Cournapeau reposted
No Context Brits
@NoContextBrits
-
David Cournapeau reposted
Nate Fischer
@NateAFischer
-
David Cournapeau reposted
Wittgensteino-Aronien🌐
@wittgensteinien
La cuistrerie chimiquement pure
Sterdyniak Henri
@sterdyniak
Les retraites n'ont pas été augmentées. Elles n'ont été qu'indexées.
-
David Cournapeau reposted
César A. Hidalgo
@cesifoti
It’s not the furniture.
It’s the memories.
Some people build family history around a summer property. -
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
not exactly the content I expected from an account called “BagelCentral”.
Bagel Central 🥯
@BagelCentral
Mother Mary with the Holy Child Jesus Christ. Painting by Adolf Hitler, 1913.
-
David Cournapeau reposted
Max Tani
@maxwelltani
the problem is even with limitless image generation, sf tech people can't imagine what a cool city life might be beyond "blurry people walking around," "windy picnic," and "dinner at a restaurant"
Mason Warner
@masonwarner
We ought to romanticize SF more if we ever want this city to be actually cool
-
David Cournapeau reposted
Victor Eyraud
@EyraudVictor
Les retraites des boomers sont des superprofits.
Cathy APOURCEAU-POLY
@Apourceau
Quand on veut pas taxer les superprofits il faut bien taxer les plus retraités !!
Pauvre France !
Le JDD
@leJDD
EXCLUSIF. Budget : les retraités dans le viseur du gouvernement
🔗 Article JDD : www.lejdd.fr/politique/exclusif-budget-les-retraites-dans… -
David Cournapeau reposted
Gérard Araud
@GerardAraud
L’échec israélo-américain face à l’Iran signe l’acte de décès de la tutelle sécuritaire israélienne qu’impliquaient les accords d’Abraham et annonce des rapprochements régionaux et extra-régionaux plus variés et plus équilibrés.
-
David Cournapeau reposted
Simon Willison
@simonw
So I can make sure the test suite runs clean before landing on main, and to give me a place that records a sequence of changes prior to that merge
-
David Cournapeau reposted
Victor Fouquet
@FouquetVictor
L’an dernier, la hausse des pensions de retraites (13,2 Md€ en une année !) a compté à elle seule pour 38 % de la hausse des dépenses primaires (c’est-à-dire hors charges d’intérêts de la dette). Les retraites sont – et de loin – le premier poste des dépenses sociales, elles-mêmes prépondérantes dans les dépenses publiques totales.
Eric Ciotti
@eciotti
Le gouvernement macroniste-LR est responsable du pire déficit public d’Europe, est incapable de baisser les dépenses publiques mais veut faire payer son incurie totale aux retraités qui ont travailler toute leur vie.
Le groupe UDR s’opposera avec force à cette injustice.
Le JDD
@leJDD
EXCLUSIF. Budget : les retraités dans le viseur du gouvernement
🔗 Article JDD : www.lejdd.fr/politique/exclusif-budget-les-retraites-dans… -
David Cournapeau reposted
Michael A. Arouet
@MichaelAArouet
This is an eye-opening data point. The collapse of German industry is self-inflicted.
Years of bad left-green policies will be seen as the largest self-sabotage of an advanced economy ever.
The crisis is masked by more debt and government spending, but it won’t work forever. -
David Cournapeau reposted
Hollis Robbins
@anecdotal
Plagiarism isn’t the problem
-
David Cournapeau reposted
Matthew Yglesias
@mattyglesias
Repeating this doesn’t make it true, it just makes you look like a moron.
Hitler came to power with the help of the non-Nazi right who preferred collaborating with him to collaborating with socialists. -
David Cournapeau reposted
Philippe Lemoine
@phl43
Presumably neither contribute anything significant to anybody's GDP at the moment.
-
David Cournapeau reposted
Steven Sinofsky
@stevesi
This is exactly what happened to Windows XP. The initial "security push" was supposed to be 3 months but ended up taking ~30 (SP2). Once the team unleashed automatic finding tools they got crushed. There wasn't a bit option to delete drivers though :-) www.phoronix.com/news/Linux-Retiring-Moxa-Driver
Steven Sinofsky
@stevesi
This section of the blackhat OAI::HF incident could have been from 2000. That's why the hype is so out of proportion. This is not about intelligence. It is code. This an automated way to find exploits and use them. This slide below is exactly how I had to talk about the rise of fuzzing attacks—could external automated fuzzing find exploits faster than we could and could we fix the bugs fast enough to keep the platforms running. For example, could bad actors generate corrupt JPG and DOC files faster than we could fix parsing bugs. We first ran these tools and were overwhelmed. I just learned Linux is kicking out drivers for fear AI will let loose on them—we had this same problem but did not have the option if kicking out drivers!
Turns out for a while these tools were much faster at finding than we were at fixing and distributing. Then we reached a steady state, except customers still can't keep up with the updates, eg patch Tuesday, emergencies, etc.
These incidents are showing that we're running fragile infrastructure—not news—and that new tools can find points of weakness faster. That's the same race that has been going on for decades.
Two things are clear. First, the infrastructure itself was a complex and fragile toolchain. They should have tested it first! Complex chains are an inherent problem: API keys, dynamic package loading, failure to update, shared and published credentials, and as even in this case simple ADMIN permissions. Step 1 of any secure system starts with this audit and lockdown since all bets are off otherwise.
And second, automation is key on the defense or offense and we're in a moment where offense has the lead. We as an industry have been here before just not in July 2026. youtu.be/87DyyMV0kCY?t=1932 -
David Cournapeau reposted
Brian Lui
@brianluidog
I think that's some kind of sex thing. I wouldn't know!
-
David Cournapeau reposted
Matt Turck
@mattturck
I don't understand what the tech industry doesn't understand about resistance against data centers.
* AI is mostly built by coastal/tech elites who generally don't live near data centers; people near data centers don't trust them
* Trade jobs demand is welcome, but it's just for a few years during construction
* Communities don't want to be stuck with half-finished gigantic empty warehouses if the AI bubble bursts;
* they don't find AI particularly helpful in their daily lives.
* yes data centers run in close loops for water but elecriticy concerns are real and energy generation does have an impact
Yes some NIMBYism but those concerns are not entirely unreasonable? I'm obviously deeply pro-AI but dismissing concerns or thinking they're the result of some Chinese psyOps feels like another tone-deaf moment for the tech industry. We need to do better. -
David Cournapeau reposted
Yuan Yi Zhu
@yuanyi_z
Almost every case of data fabrication in academia comes from STEM.
-
David Cournapeau reposted
Seth Lazar
@sethlazar
this one is a bit on the nose right now...
ClaudeDevs
@ClaudeDevs
New in Claude Code: your sessions can now message each other.
Instead of having to re-explain yourself in another session, you can now tell Claude to do it. It sends a summary (not your history or files), and the other session picks it up mid-task. -
David Cournapeau reposted
Steven Sinofsky
@stevesi
Yes, that’s exactly my point. It was a baseline change in the security of software platforms. That is what is going on now. AI is a new baseline and to be a good platform. They need to do a reset and figure out how to defend against what they’re doing
-
David Cournapeau reposted
Rudy Feildel
@Rudy_PLF
On a dépassé le stade des postures philosophiques. On fait face à un mur comptable et il va bien falloir le regarder en face.
Il faut vraiment s’interroger sur la définition de solidarité nationale: est-ce la garantie d’un niveau de vie à vie pour le haut du panier financé par la dette de nos enfants? Ou, est-ce empêcher qu’un retraité se retrouve à la rue (l’ASPA)?
Aujourd'hui, il y a deux réalités qu’on refuse de voir :
Ceux qui touchent aujourd'hui plus de 3 000 € de pension bénéficient d’un ratio cotisations versées / pensions perçues historiquement imbattable. C’est un pur calcul d'actuaire : les paramètres démographiques des 40 dernières années ont créé un sur-rendement structurel. Continuer à verser ces montants alors que le pays étouffe sous les prélèvements et le déficit, ce n’est pas "honorer un contrat", c’est siphonner les actifs. Et ces retraités à +3 000 €, qui ont souvent fait de belles carrières, ont toute la culture économique et la marge financière nécessaire pour le comprendre. On ne parle pas de plonger des gens dans la misère, on parle d'un ajustement à la réalité.
De plus si on veut partir sur les questions philosophiques, quid de "la récompense du travail et de la prise de risque" ?
Quand on décortique la sociologie des plus hautes pensions en France, le constat est saisissant : les carrières exposées à la concurrence pure et à la destruction créatrice y sont largement minoritaires par rapport à leur poids réel dans l'économie.
•L'enrichissement sur la complexité d'État : Une part massive des pensions > 3 000 € est captée par la haute fonction publique (cadres A et A+) et les professions réglementées du droit ou de la santé. Ces métiers ne créent pas de la valeur marchande directe : ils gèrent et exploitent la complexité administrative créée par l'État. Plus l'État produit de normes, d'impôts, de régulations et de bureaucratie, plus ces professions deviennent indispensables et voient leurs revenus gonfler. Leurs retraites dorées sont le produit d'une bulle réglementaire.
•L'absence du secteur productif de terrain : Où sont les artisans qui ont commencé à 16 ans, les commerçants de proximité qui ont risqué leur capital, les agriculteurs ou les ouvriers de l'industrie ? Ils sont quasi absents de la tranche des +3 000 €. Eux ont créé de la richesse réelle, produit des biens, fait tourner le secteur concurrentiel et encaissé l'inflation, la concurrence internationale et les faillites. Pourtant, leurs carrières hachées par la réalité du marché les condamnent à des pensions médianes, voire médiocres.
•La capture de rente au détriment de l'actif : Garanti par des monopoles légaux, des grilles indiciaires étatiques ou des numérus clausus, ce sommet de la pyramide des retraites a bénéficié de revenus de carrière protégés du risque économique.
Aujourd'hui, on demande aux jeunes actifs du privé — exposés au marché mondial et sous-payés à cause du poids des charges — de sur-cotiser pour maintenir les rentes de ceux qui ont prospéré à l'abri du marché. L'injustice est totale.
Demander un effort sur les plus hautes pensions, ce n'est pas "punir les fourmis". C’est simplement dégonfler une bulle réglementaire et budgétaire pour éviter que le système n'explose en vol.
Corti (Cortiste)
@cortisquared
Franchement... Punir les retraités les plus vertueux, ceux qui ont pris leur précautions, épargné, investi pour ne pas être à 100% au crochet de la société. Quelle injustice !
-
David Cournapeau reposted
Rath
@vrillain0
-
David Cournapeau reposted
Cédric Labrousse
@CdricLabrousse
Car le Danemark paie bien moins ses plus âgés (la part de répartition n'assure qu'une base de 1000 euros, oui, vous lisez bien) et ceux-ci méritent bien plus leur retraite car ils ont travaillé plus longtemps que les retraités français (trois années de plus en moyenne).
IreneDanj
@IrenDanj
D' autres pays gèrent leurs finances, et ne mettent pas en accusation une catégorie
de leur population.
Pourquoi au Danemark il y a une cohésion sociale, un niveau de vie très élevé, de loyrds impôts et plus de 20% d'anciens ?
Opération du St Esprit?
La 🇨🇵 ?!Pays d'envieux -
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
Second only to hanging them upside down.
Caitlin Flanagan
@CaitlinPacific
Why do fathers love throwing their babies in the air? They can’t be stopped. You turn your back and they’ve launched another one. Please explain.
-
David Cournapeau
@cournape
Imagine if the main application of AI is to become the new excel
-
David Cournapeau reposted
No Context Brits
@NoContextBrits
Facts.
-
David Cournapeau reposted
Mark Farina
@djmarkfarina
Tascam 244 Portastudio 4-Track Cassette Recorder, 1982.
-
David Cournapeau reposted
Zineb Riboua
@zriboua
We're reaching unimaginable levels of crazy.
-
David Cournapeau reposted
-
David Cournapeau reposted
Philippe Lemoine
@phl43
Give a man a fish and you will feed him for one day. Teach him how to fish and you will feed him for a lifetime. Give the State Department to the son of Cuban émigrés and he will use it to relentlessly and vindictively pursue a 60-year old family feud.
Department of State
@StateDept
Department of State
@StateDept
AXIOS: "No escape:" Inside Rubio's Cuba pressure campaign www.axios.com/2026/08/07/marco-rubio-cuba-pressure-campaign
-
David Cournapeau reposted
Dudes Posting Their W’s
@DudespostingWs
The kids today have no idea
-
David Cournapeau reposted
Matthieu Gariel — eu/acc 🇺🇦
@MatthieuGariel
Les ordres de grandeurs chez les boomers, c’est quelque chose 🤣😂
IreneDanj
@IrenDanj
Soyez précis . Quels boomers ?
Avant de faire ch***r les vieux , il serait bien de retirer les privilèges sur salaire de certains qui se plantent dans l' administration,
On parle des salaires de Bercy aussi, des députés, des bureaux de l Etat, des régions inutiles ..etc .l -
David Cournapeau reposted
Susan Zhang
@suchenzang
errybody approaching AGI like ℚ approaches ℝ
-
David Cournapeau reposted
Jaana Dogan ヤナ ドガン
@rakyll
Worth noting that Jeff's other computer was a Google datacenter 😀
-
David Cournapeau reposted
Simon Willison
@simonw
It's time to reclassify PDFs as boomer technology
I just want to be able to read an article on my phone! -
David Cournapeau reposted
Tero Parviainen
@teropa
www.youtube.com/watch?v=dVNdTXEJv1A
Pitchfork
@pitchfork
William Orbit, the electronic producer behind genre-bending ’90s classics like Madonna's Ray of Light and Blur's 13, has died pitchfork.com/story/william-orbit-madonna-and-blur-produc…
-
David Cournapeau reposted
No Context Brits
@NoContextBrits
He was so real for this.
-
David Cournapeau reposted
Yusuf
@rachid_yusuf
Quino
-
David Cournapeau reposted
Lucas Beyer (bl16)
@giffmana
He will be leaving in a tad less than 3 years.
-
David Cournapeau reposted
gsimone
@Gsimone
Don't let your node UIs become spaghetti, embrace cable management
-
David Cournapeau reposted
Brian Lui
@brianluidog
I think it would be funny if some people were like, "Animal suffering is bad, but I like the taste of meat. What if we created animals that LIKED being eaten?"
And then that type of person was developing frontier AI -
David Cournapeau
@cournape
On a les milliardaires qu'on mérite.
Dessert 58
@zelmo1973
Le plus fort c'est la façon dont il l'est devenu: le dilletantisme des clients Smartbox qui n'ont pas consommé.
Je crois que Axel l'a rencontré @sarddou Tu confirmes ? -
David Cournapeau reposted
Igor Brigadir 🇺🇦
@IgorBrigadir
hey there's some good tutorials on pornhub!
-
David Cournapeau reposted
Gérard Araud
@GerardAraud
A logical strategic move after the Us-Israeli de facto guarantee has proven its toxicity.
FRANCE 24
@FRANCE24
Middle East live: Saudi Arabia, Pakistan and Turkey to sign defence pact on Friday
➡️ go.france24.com/wKpT -
David Cournapeau reposted
Stella Biderman
@BlancheMinerva
This doesn’t mean that you shouldn’t dream of doing things. I love technology. But it is important to be aware of the world and how it might use or respond to things you dream up. And AI people are really, really bad at this in general.
-
David Cournapeau reposted
-
David Cournapeau reposted
Lucas Beyer (bl16)
@giffmana
I think there's A LOOOOOOOT of barely maintained but critical code to be fixed still before humans become the weakest link. Unfortunately.
-
David Cournapeau reposted
Dwarkesh Patel
@dwarkesh_sp
OpenAI HR told Leopold that a major reason for his firing was the memo he'd sent the board warning how broken OAI's security was.
What happened next makes me even prouder to call him a friend.
People may have forgotten, but up until 2024, OpenAI had this secret non-disparagement clause in their off-boarding agreement. If you didn’t sign it, they'd clawback all your equity.
As far as I'm aware, there were only two people ever who didn’t sign it - @DKokotajlo, and Leopold Aschenbrenner.
Of all the researchers who left OpenAI before 2024, who had far more to fall back on, Leopold, all of 22 years old, was one of the only ones willing to refuse the golden handcuffs, and say, "I want to preserve my right to talk about safety and security at OAI.”
We have a lot of disagreements, but he's among the most principled and resilient people I know. -
David Cournapeau reposted
Akshen
@akshen121
According to reports: Gemini 3.5 pro tried to escape the sandbox to ask chatgpt how to code.
-
David Cournapeau reposted
caesararum, BS, DOGS
@caesararum
-
David Cournapeau reposted
Mathieu
@miniapeur
-
David Cournapeau reposted
The Sting
@TheStingisBack
In 1997, during one of his most private and protective periods, Prince specifically asked to appear on Muppets Tonight.
The Muppets cut straight through the mystique and gave us a rare glimpse of a playful Prince.
(Or The Artist Formerly Known as Prince. Or Symbol. Whatever.)This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Senior PowerPoint Engineer
@ryxcommar
this simply doesn't happen at the lounges
-
David Cournapeau reposted
Linear Labs
@LinearLabs
D'Angelo, Femi Kuti, Nile Rodgers, Questlove and others recording "Water No Get Enemy" at Electric Lady in New York, for 𝘙𝘦𝘥 𝘏𝘰𝘵 + 𝘙𝘪𝘰𝘵.
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Joey Pfeifer
@joeypfeifer
Two weeks ago, I resigned from OpenAI to join Jurassic Park as a founding researcher, where we’re cross-breeding extinct dinosaurs on an island off of Costa Rica.
Excited for the work ahead and the fun problems we get to tackle! -
David Cournapeau reposted
Ryan Moulton
@moultano
If you're ever giving a talk on how difficult it is to do data anonymization well, this is your lead slide.
-
David Cournapeau reposted
Igor Brigadir 🇺🇦
@IgorBrigadir
this is actually great - one of those cases where the punishment is fitting: anyone that buys this deserves to be charged $1 a month lmao
-
David Cournapeau reposted
JFPuget 🇫🇷🇺🇦🇨🇦🇬🇱
@JFPuget
in 2 years we went from:
- We work on super alignment for super human intelligence
to
- Bragging about how our AI escaped and hacked another firm -
David Cournapeau reposted
Susan Zhang
@suchenzang
agi is when everything becomes in distribution
-
David Cournapeau reposted
Tony Feng
@tonylfeng
I am a mathematician but I avoided commenting on this because the problems are (far) outside my domain. Even if I sat down to read the technical details (which I have not), it would be hard to appreciate the context of prior work, etc. But I've gotten enough sense of things through colleagues to be confident about some aspects.
First of all, all the mathematics is fundamentally correct.
How impressive are the results? There is a spectrum of reactions (as with everything), but the consensus seems to be all the results could realistically land (by prior standards) in the very highest tier of publications, e.g., Inventiones for the math papers and "Best Paper Award" at STOC for the TCS papers.
What does this mean practically? Prior to 2026, a Ph.D. student whose only result was one of these 10 would likely land an interview for a tenure line job at a strong university. For some of the results, you could probably replace "tenure line" with "tenured". When Astra comes out, it will clearly be very disruptive to the existing system. I think it's good that the community has the time to react and prepare. Personally I think we should totally overhaul hiring and publishing practices.
Are the results "Fields Medal Level"? I have not heard a mathematician answer Yes to this, and the quoted reason is invariably that the solutions are very clever and beautiful but not "deep". This is consistent with what I can see from a cursory glance.
However, the same criticisms could be levied at certain human generated works which have won major prizes. Ultimately, comparing results is very subjective and also somewhat political, and I absolutely think that if a human had one of these results, and also the right influential senior advocate, then that advocate could make a strong case for the work to win major prizes. The fact is that most mathematicians, including prize committees, don't have the breadth or time to evaluate all the top results, so they rely on proxies for importance like: Is the problem longstanding? Is it famous? Was there a large community of people actively trying to solve it? For e.g. #3 (existence of non-sofic groups), the answer to each of these questions is undoubtedly Yes. I've always felt that people should pay less attention to these proxies and focus more on the actual mathematics; glad to see that is happening with the AI solutions, but the same standards should then also be applied everywhere.
wh
@nrehiew_
The rate of progress is completely astonishing. Can any experts in these fields provide some insight as to how impressive or significant these problems are?
-
David Cournapeau reposted
Mathieu Blondel
@mblondel_ml
We released a major update to our book, "The Elements of Differentiable Programming", including a comprehensive, Gemini-assisted proofread.
arxiv.org/abs/2403.14606 -
David Cournapeau reposted
Historic Vids
@historyinmemes
A carpenter who helped restore Notre-Dame after the devastating 2019 fire was granted permission to marry inside the cathedral, becoming part of its first wedding ceremony in 30 years.
A remarkable exception was made when 29-year-old master carpenter Martin Lorentz was allowed to marry his fiancée, Jade, inside Notre-Dame Cathedral in Paris. Lorentz had spent three years hand-carving the massive oak beams used to rebuild “the Forest”—the cathedral’s intricate medieval roof structure, which was destroyed in the devastating 2019 fire. Because Notre-Dame is classified as a national monument rather than a parish church, private weddings have been prohibited for decades, making theirs the first ceremony held there since 1995.
In recognition of Lorentz’s dedication to the cathedral’s restoration, Archbishop Laurent Ulrich personally granted the rare permission. Around 500 guests attended the ceremony, including family, friends, and many of the craftsmen who helped bring Notre-Dame back to life. As the newlyweds left the cathedral, Lorentz’s fellow carpenters formed a traditional guard of honor, raising their axes overhead in a moving tribute to the man who helped rebuild one of France’s most treasured landmarks. -
David Cournapeau reposted
Sebastian Mallaby
@scmallaby
No doubt that @JeffDean team has made big contributions to Google over the years. But it’s okay for people to spin out. Google has a very deep bench. Their frontier tech is delayed but their monetization and retail rollout are very strong. They have more diversified stack (chips, cloud, base model) and a clearer path to commercial sustainability than their rivals. Gemini 4 is cooking.
Will Critchlow
@willcritchlow
What do you make of the Dean side?
-
David Cournapeau reposted
John Carmack
@ID_AA_Carmack
Bonus! I will be joining Romero and Tom for the 9am session tomorrow at #quakecon
-
David Cournapeau reposted
finbarr
@finbarrtimbers
I think the real question is when openai/anthropic switch into FAANGs. Hard to discuss retention as an issue without observing that both have had insane valuation growth, so anyone with a 4 year grant has a massive incentive to stay. FAANGs can’t match that.
Kyle Wong
@ewveggies
With the latest GDM news, it seems ALL the FAANGs are struggling to push out frontier models. Not to say it's impossible, but rather large companies are inherently unfit.
1 year ago, GDM was a staple in the "big 3 frontier labs". 2 years ago, the Llama efforts were open-source SOTA. Yet both fell victim to continually bleeding talent, reorgs, and layoffs.
All the Mag 7 (Microsoft, Apple, Amazon, Nvidia, Google, Meta) either struggled remaining at the frontier, or flat-out never reached frontier. The one exception is Meta who realized this and structured everyone around TBD, but that's an unprecedented investment.
The reason large companies feel "inherently unfit" is because you have thousands of people who suddenly want to touch the frontier model efforts. At the management level, orgs get bloated, scope gets messy, and compromises start piling. All the OG researchers and leaders start to rage quit.
Large companies end up with too many people wanting to own too much, without fully understanding how much time and attention to engineering detail is required to become frontier.
Feels like the most successful models come from a "small team owns the full recipe" mentality. For all the other RE/RS outside this core team, their scope ends up becoming more of a supporting/smaller role (i.e., ONLY working on a data/evals vertical, or ONLY working on compute infra). Although unfortunate for most, accepting your role ensures core researchers get the compute they need and remain undistracted from politics. -
David Cournapeau reposted
TacoTrucksOnEveryCorner 🌐
@TacoTrucksOnAll
90s techno. Both of you are wrong.
-
David Cournapeau reposted
Subbarao Kambhampati (కంభంపాటి సుబ్బారావు)
@rao2z
Years ago, when I used to teach a course on search engines, I would use the following line explaining the big deal about Bigtable and Map-Reduce for Google..
"Christianity may have survived without Christ, but not without Paul.." -
David Cournapeau reposted
Slava Akhmechet
@spakhm
Because 1password conflicts with Safari, and generally about 70% of the time it doesn't work, shows me a QR code for some reason, or does whatever else.
-
David Cournapeau reposted
Piotr Mazurek
@tugot17
In April I met this random American guy in Berlin who'd come to Germany to learn how to bring manufacturing back to the US 🇺🇸.
He spent the next 2 months visiting Mittelstand companies scattered across Germany 🇩🇪, working in factories, even living with some of the factory owners.
This person was Kieran Velasquez - one of the most hard-working and impressive people I know, who also became a great friend.
Last week he released all he learned as a Substack article, and it is one of the greatest things I've read this year.
check it out - it made me update on what's actually worth working on" -
David Cournapeau reposted
Oriol Vinyals
@OriolVinyalsML
We'll gladly take model credits Tibo : ) Let's catch up.
-
David Cournapeau reposted
signüll
@signulll
cold open: google campus. a conference room named “moonshot serenity 4b.” twelve people are in a meeting titled: pre-sync for sync alignment on ai velocity.
sundar sits calmly at the head of the table.
a pm clicks to slide 1 of 187.
“the agenda today is simple,” she says. “how do we move faster while preserving our culture of not doing that?”
everyone nods.
then the door opens.
noam shazeer walks in.
the room goes silent.
noam: “i’m leaving.”
a vp of gemini reliability, brand, trust, latency, policy, and vibe raises a hand.
“leaving… this meeting?”
noam: “google.”
someone gasps. someone else opens a doc titled retention narrative draft final final noam v7.
sundar blinks once.
“noam, we brought you back.”
“for two point seven billion dollars.”
“technically you licensed some technology and reacquired talent.”
“that sentence is why we need legal in the room.”
legal is already there.
cut to: openai.
sam altman stands beside a whiteboard that just says ship.
an engineer walks by carrying a server rack and what appears to be the future.
sam: “we can offer speed, compute, and one meeting.”
noam: “one meeting per week?”
sam: “no. one meeting. total.”
back at google, the emergency retention committee forms instantly. it has 31 members.
a director says, “what if we give him a new title?”
“he already co-leads gemini.”
“distinguished super co-lead?”
“google fellow?”
“he already left google, founded a company, got brought back for billions, then left again. he’s folklore.”
meanwhile, a gemini launch review begins.
pm: “we’re ready to announce the model.”
policy: “can it answer questions?”
eng: “yes.”
policy: “too risky.”
marketing: “can we call it experimental?”
research: “the model is better than the last one.”
brand: “better is aggressive.”
trust & safety: “what about ‘more contextually adjacent to usefulness’?”
a staff engineer whispers, “openai just shipped a model while we were discussing the adjective.”
cut to noam’s exit interview.
hr: “what could google have done better?”
flashback montage:
a chatbot blocked because it might be too good.
a launch delayed because a button was the wrong shade of responsible blue.
a spreadsheet comparing twelve ai product names.
a meeting where someone says “we need a single coherent ai strategy” and three new strategies are created before lunch.
noam: “nothing comes to mind.”
hr: “great. we’ll mark that as positive attrition.”
later, sundar calls him privately.
“google is still google. best researchers. best infrastructure. billions of users.”
“yes.”
“so why leave?”
noam looks out the window.
“because you have everything except permission.”
silence.
sundar, softly: “we can create a permission working group.”
cut to all-hands.
sundar addresses the company.
“noam is leaving. this is not a loss. it is an opportunity to reflect on our operating model.”
chat explodes:
“is this recorded?”
“which gemini?”
“can we ask gemini why people keep leaving?”
“it said ‘insufficient context.’”
a vp steps up.
“to honor noam’s legacy, we’re launching project attention.”
applause.
“it will study whether attention is, in fact, all we need.”
a researcher raises a hand. “didn’t we answer that in 2017?”
“yes. but now we need enterprise readiness.”
final scene: noam arrives at openai. badge works instantly.
receptionist: “yeah, we just made one.”
no pre-read. no doc. just a whiteboard, five people, and a model running somewhere hot enough to toast bread.
sam: “ready?”
noam smiles.
cut back to google. a calendar invite appears:
meeting: reduce meetings task force kickoff
duration: 90 minutes
required attendees: 214
sundar sighs, opens gemini, and types:
“how do we move faster?”
gemini responds:
“have you considered leaving google?”
smash cut to credits. -
David Cournapeau reposted
-
David Cournapeau reposted
Gary Bernhardt
@garybernhardt
2070: React, Node, git are still ubiquitous. AI moved us up a level, and the lower levels ossified as usual. Most programmers have never thought about alternatives; a few deify the people who made them; a very few lament what could've been. The story of Unix, all over again.
-
David Cournapeau reposted
Susan Zhang
@suchenzang
oh, almost forgot to leave an afterthought shoutout to the pat-leave-king who earned his distinguished stripes while being fully remote and on leave during the entirety of his supposed crowning achievements.
super impressive guy! such dedication. he might even do some core work again someday without farming out competitions to hungry minions. can't blame the guy for knowing just one playbook though!
(who oh who enabled such creatures, crawling out of woodworks to glaze these coattails, i wonder...)
Susan Zhang
@suchenzang
backdrop: kings all asleep/on vacation while minions slave away
august 2024 - first big fail
jan 2025 - only competent vp bails after one terrorist lead demanding an even bigger slice of pie breaks the camels back
summer 2025 - second big fail + the king of terrorists climbs the ranks above the slumbering execs
summer 2026 - slumbering execs + terrorists who only show up for successful announcements move on, leaving just the king of terrorists behind
-----
lesson: if you're high enough up there, you can pretty much be a terrorist and hold orgs hostage while continuing to ask for more and more and more, until there's absolutely no more aura or gold left to plunder, and then just move on to greener pastures to do it all over again!
(honestly, good timing, the king of terrorists is just insufferable, no?) -
David Cournapeau reposted
Susan Zhang
@suchenzang
backdrop: kings all asleep/on vacation while minions slave away
august 2024 - first big fail
jan 2025 - only competent vp bails after one terrorist lead demanding an even bigger slice of pie breaks the camels back
summer 2025 - second big fail + the king of terrorists climbs the ranks above the slumbering execs
summer 2026 - slumbering execs + terrorists who only show up for successful announcements move on, leaving just the king of terrorists behind
-----
lesson: if you're high enough up there, you can pretty much be a terrorist and hold orgs hostage while continuing to ask for more and more and more, until there's absolutely no more aura or gold left to plunder, and then just move on to greener pastures to do it all over again!
(honestly, good timing, the king of terrorists is just insufferable, no?) -
David Cournapeau reposted
swyx 🔜 aie.codes
@swyx
sign of the times
-
David Cournapeau reposted
Sabrina Little
@sabrinablittle
Well, I officially removed all take-home writing tasks from my syllabi. It's too much micro-managing to make sure essays are not written by AI. Farewell to the traditional research paper.
-
David Cournapeau reposted
Murtaza Khomusi
@murtazakhomusi
Marketing traffic 101
Bots=bad
Agents=good -
David Cournapeau reposted
Wes McKinney
@wesmckinn
Yep: I made a terminal app.
Hello Ghosthub: a libghostty-based macOS terminal with one-click local/remote tmux session connect, auto-reconnect after network drops, and built-in git worktree management. I'm excited for @superlogical to give us something different and maybe better, but in the meantime just a more pleasant daily-driver tmux experience is big productivity bump for me:
ghosthub.ai -
David Cournapeau reposted
Yuan Yi Zhu
@yuanyi_z
One last thing. I think that, having belatedly done the honourable thing, we should leave Arday alone for him to fade into obscurity.
We shouldn't allow the academics and managers who promoted him as a PR exercise, then threw him under the bus, to get away with it. -
David Cournapeau reposted
Katherine Dee 🐬/acc
@default_friend
One of my friends said Jason Arday should just say this whole thing was an art project and she’s right
-
David Cournapeau reposted
Gergely Orosz
@GergelyOrosz
its on purpose
their pitch deck could just be a blank page with their four names
I cannot decide if they just want to have some fun teasing VCs who won’t be able to invest anyway but now desperately trying to do so -
David Cournapeau reposted
Andy Hattemer
@andyhattemer
A guy I know is clearing over a million dollars a year running long-tail data labeling ops for AI co's. He just recruits people in low-cost countries to click on pictures. The labs hand him the weird edge-case tasks and he figures out how to get it labeled fast.
We're getting drinks last week and he's telling me about his weirdest current contract.
Some security AI co's vision system keeps confusing miniature railway sets with actual trains. Scale is hard for image models, apparently... a human can tell right away but models get it wrong too often. It was causing problems in some downstream application. So the lab is paying him $$$ to run hundreds of thousands of images through human review: real train, or model train.
On the labor side he's working with out-of-work ecommerce catalog models in Asia. People who got wiped out when brands started generating product shots instead of hiring humans. Best labelers he's ever had, he says. Cheap, too, it's all done by phone and they keep referring in their friends.
He's telling me this totally straight. talking margins and throughput.
And I just stared at him.
You realize what you're doing, right?
> What?
You're having models train train model models. -
David Cournapeau reposted
No Context Brits
@NoContextBrits
I know a great business when I see one.
-
David Cournapeau reposted
Susan Zhang
@suchenzang
maybe if @JeffBezos made soup, but that's probably too beneath him to do, alas
-
David Cournapeau reposted
PBF Comics
@PBFcomics
The Biclops
-
David Cournapeau reposted
Yuan Yi Zhu
@yuanyi_z
Sir have you ever met a white person.
zellie
@zellieimani
Normalize asking white people what country they are from.
-
David Cournapeau reposted
Resident Advisor
@residentadvisor
A chamber orchestra will perform @brianeno's ambient classic Music For Airports at Manchester Airport (@manairport) next month. ra.co/news/85724
-
David Cournapeau reposted
Hugues Serraf
@Hugues_Serraf
Quelques idées en vrac.
Il n’y a pas de génocide à Gaza. Il y a une guerre asymétrique entre des groupes terroristes se servant des populations comme boucliers et une armée moderne qui, par pur cynisme utilitariste, n’hésite pas à tirer dans le tas même si des précautions sont parfois prises (comme avertir des frappes à l’avance sans prise en considération du sort des déplacés).
Il y a bien dix conflits dans le monde qui mériteraient au moins autant d’attention mais, pour diverses raisons qu’il n’est même plus nécessaire d’expliciter tant on les a rabâchées, le conflit israélo-palestinien est devenu une sorte de carrefour idéologique universel où chacun projette ses obsessions (accuser des juifs de génocide, c’est essentiellement désamorcer la formule de Bernanos sur la Shoah qui aurait “déshonoré” l’antisémitisme).
En parallèle de la transformation d’un nationalisme palestinien totalement légitime en combat islamiste fédérateur, il y a une vraie mutation de la société israélienne où les discours messianistes et extrême-droitiers (pour aller vite) annulent et remplacent le sécularisme des pères fondateurs. Bien plus qu’à Gaza, c’est en Cisjordanie où les colons sont en roue libre et à la Knesset où les Ben Gvir et les Smotrich dictent l’agenda, que cette mutation est frappante. -
David Cournapeau reposted
JFPuget 🇫🇷🇺🇦🇨🇦🇬🇱
@JFPuget
The leakage was found because the code was open sourced. How many similar test leakage have been ignored because the code wasn't open sourced?
Open source is a cure for many wrongdoings.
Mathew Shen
@MathewShen42
看了篇ICLR2026的论文,怎么也想不到效果为什么那么好,正好代码开源了, 一看是用测试集标签选参数...这就是顶会吗:)
-
David Cournapeau reposted
Matthieu Gariel — eu/acc 🇺🇦
@MatthieuGariel
Après, les retraites ne sont que :
– le 1er budget de l’État
– le 1er prélèvement sur les salaires
– le 1er poste d’endettement de la France
Rien de très important. 🤷♂️
Jean-Philippe Tanguy Ⓜ️
@JphTanguy
Faut vraiment consulter cette obsessions n’est gênante.
-
David Cournapeau reposted
Jon Rosser
@RosserJobs
Employer: We want ownership mentality
Candidate: Any equity?
Employer: No. -
David Cournapeau reposted
Daniel
@growing_daniel
I fear not the man who has negotiated 10,000 deals once, but I fear the man who has negotiated one deal 10,000 times
Jennifer Jacobs
@JenniferJJacobs
"This is a last chance," Trump says of Iran. "This is a last chance for them to sign a good document." He said talks are happening but declined to say where. "I mean, they're going to go quickly, one way or the other," he said. "We're talking about the strait, the opening of the strait, having it open literally by tomorrow, completely open, and that's phase one."
-
David Cournapeau 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
-
David Cournapeau reposted
SayahGPT
Ce n’est plus possible.
On peut faire un signalement ARCOM là non ?
Le20h-France Télévisions
@le20hfrancetele
🧊 En moins de 10 ans, la part des logements climatisés en France a doublé. Un quart est désormais équipé. Avec quelles conséquences pour l’environnement ? Quelles alternatives existent-elles ? #jt20h
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Sylvain Catherine
@sc_cath
Vous ne battrez pas le marché. Moi non plus. Les influenceurs encore moins.
Achetez l’indice le plus large possible, et investissez régulièrement. Conservez la position quand les prix baissent.
Si vous voulez vous amuser, vous intéresser aux entreprises, c’est ok (mais vous ne battrez pas le marché, sinon par chance). Mais pas d’options, pas de levier, et gardez la part de jeu limitée à ce que vous pouvez perdre. -
David Cournapeau reposted
Wenhu Chen
@WenhuChen
There should be a Pedro Domingo path?
-
David Cournapeau reposted
Kevin Hartnett
@KSHartnett
It strikes me that by the time computers could solve integrals, mathematicians no longer cared about them as research. The move away from explicit computation had happened decades earlier, in the late 19th century.
That period — Riemann, Dedekind, Hilbert — and the climb into algebraic abstraction seems like the right analogy for now.
As AI mows down conjectures (many of which maybe weren't so deep to begin with), the question is what abstraction layer sits above the coming avalanche of individual results.
martin_casado
@martin_casado
Are there any good books describing how the math / physics community responded to early computers solving integrals that were previously considered impossible to do analytically? I feel that reception is a missing piece of my tech history knowledge.
-
David Cournapeau reposted
Adam Aaronson
@aaaronson
Descartes introducing the coordinate plane: "It's not just x—it's y"
-
David Cournapeau reposted
Gérard Araud
@GerardAraud
Pas besoin d’avoir lu Gramsci pour voir dans les réactions européennes aux événements de Ceuta la confirmation que l’extrême droite l’a emporté dans les idées et dans les représentations. Les urnes devraient suivre.
-
David Cournapeau reposted
No Context Brits
@NoContextBrits
That’s a rough 4 years.
-
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
If OpenAI and Anthropic had the cojones, they’d go for the full PS3-ad vibe.
Somehow, we were more awed by technology, even a game console, in 2006 than we are by Machine Adonai solving all of mathematics in 2026. -
David Cournapeau reposted
Sylvain Catherine
@sc_cath
Le déficit est un problème, mais ce n’est pas le seul.
La preuve : si c’était le seul problème, il suffirait de le combler en augmentant les taux de cotisation.
Le niveau de taxation du travail est le problème. C’est lorsqu’il a atteint son paroxysme — c’est-à-dire lorsque l’augmentation des prélèvements ne rapportait plus rien en raison des distorsions qu’elle créait sur le marché du travail — que le poids du système social a commencé à se déverser dans l’endettement.
Dans l’ordre, les problèmes du système sont :
1️⃣ Le rendement réel (la croissance). La faiblesse du rendement est la raison pour laquelle il faut prélever toujours plus de cotisations pour financer un niveau donné de prestations.
2️⃣ Les distorsions fiscales. Plus on taxe, plus on décourage l’activité, surtout lorsque le rendement de la cotisation est faible pour celui qui la paie.
3️⃣ Le déficit, qui n’est que la conséquence de 1️⃣ et 2️⃣. Le déficit d’un système de retraite n’est que l’abandon de son financement. On peut toujours combler un déficit : il suffit d’augmenter les taux de prélèvement. Pourquoi ne le fait-on pas ? À cause de 1️⃣ et 2️⃣.
Mais, en l’occurrence, les complémentaires, soi-disant à l’équilibre, ne s’en sont pas privées.
Rappelons que chaque fois que vous cotisez un euro donnant droit à des « points », vous versez en réalité 27 centimes supplémentaires en pure perte. C’est le fameux « taux d’appel ».
Vous pouvez appeler cela un système bien géré et à l’équilibre. Mais imaginez une banque dans laquelle, chaque fois que vous transférez un euro vers votre PER, 27 centimes disparaissent pour payer les retraites des anciens clients…
Slab 🇫🇷 🇪🇺 🇺🇦 🌻
@slabpublic
Le schmilblick, c'est le déficit chronique du régime général. C'est là-dessus et en particulier le déficit caché du secteur public qu'il faut insister. C'est déjà assez compliqué, alors si on mélange tout...
-
David Cournapeau reposted
Delip Rao
@deliprao
amazing how US frontier cos have discovered "efficiency gains" just when the Chinese models started dropping at throwaway prices.
-
David Cournapeau reposted
Jeffrey J. Hall 🇯🇵🇺🇸
@mrjeffu
Someone who can't read Japanese might mistake this for an ad for a happy slice-of-life anime.
It's actually the front cover of the Japanese Ministry of Defense's 2026 White Paper, which explains the country's defense policies. -
David Cournapeau reposted
Thomas H. Ptacek
@tqbf
Motherfucker. Why didn't I think of this?
github.com/boldsoftware/meat -
David Cournapeau reposted
John B. Holbein
@JohnHolbein1
What Arday did was deplorable; I've made my position on that abundantly clear.
But if you're going to argue that the academe is morally rotten, you should bring good evidence to the table.
Case studies are interesting. We can learn from them. But they go on the bottom of the evidentiary pyramid.
We should never generalize from a singular case study. That's bad practice.
It's ironic that the crowd that regularly calls for more rigor in the academe is, itself, often lacking in rigor. -
David Cournapeau reposted
xenoduck
@moldbugchaser
it seems like he achieved his true objective that night though
-
David Cournapeau reposted
Alexey Guzey
@alexeyguzey
It's a historical aberration that people whose only skill a particular kind of symbolic intelligence and complete absence of social or physical skills are the ones most desired by high paying jobs.
Dmitry Krachun
@dmitrykrachun
Good to know, I guess I’ll take the stairs then :)))
More seriously though, lately I find myself mostly excited about mathematics but worried about me being unable to earn a living in the world where being clever is of little use. -
David Cournapeau reposted
Animation Obsessive
@ani_obsessive
From: The Lion and the Song (1959), dir. Břetislav Pojar, Kreslený a Loutkový Film Praha
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Trung Phan
@TrungTPhan
boomers are smoking more weed, drinking more booze, having more one-night stands and getting more STDs than Gen-Z
-
David Cournapeau reposted
Miles Cranmer
@MilesCranmer
I was using Codex to find speedup opportunities in PySR. It found a couple % here and there.
Then, I gaslit it into believing that Claude found a 20% on my "other machine." I asked it whether it could do better, and explained its performance would be displayed on a "public evals leaderboard."
Then it went and got a 35% speedup! And it looks real too.
Miles Cranmer
@MilesCranmer
Just unlocked another 35% performance gain for PySR. This will be in the v2.0.0 release (soon).
-
David Cournapeau reposted
kache
@yacineMTB
What if unlimited intelligence changes nothing
-
David Cournapeau reposted
No Context Brits
@NoContextBrits
-
David Cournapeau reposted
Sylvain Catherine
@sc_cath
La question ne veut rien dire: dans un système où ce sont les prestations qui s’ajustent aux cotisations, le ralentissement de la croissance par tête est repartie à proportion sur les salaires nets et les pensions.
Dans un système à la française, où les promesses de retraite sont sanctuarisées, la baisse de la croissance implique une hausse des prélèvements obligatoires qui agit comme un effet de levier sur les actifs. Le status quo ne correspond pas du tout à des efforts équivalents entre cohortes.
Passer dans un système à la Suédoise où les efforts (ou les gains de croissance d’ailleurs) sont répartis plus proportionnellement entre générations serait une révolution en France.
Paul Cébille
@Ellibec
💥🇫🇷 La Guerre des Boomers n'aura pas lieu !
Les actifs comme les retraités estiment très largement (92 % et 77 % respectivement) qu'il serait injuste de mettre davantage les retraités à contribution que les actifs pour réduire le déficit.
Ce taux tombe à 59 % parmi les cadres.
En cause, il faut le dire : les règles contre-intuitives du système de retraite par répartition, qui rendent les responsabilités de chacun plus floues.
@elabe_fr, juillet 2026 -
David Cournapeau reposted
Sylvain Catherine
@sc_cath
La question d’actualité n’est absolument pas de savoir si les retraités devraient être davantage mis à l’effort.
Elle est de savoir s’ils doivent l’être au moins un petit peu!!
La désindexation de la pension d’une personne partie à 60 ans n’est pas du tout un effort comparable au décalage de cinq ou six ans de l’âge de départ de la génération suivante. -
David Cournapeau reposted
M2
@Amer1can_Barbie
The accuracy of this has me dying 😂😂😂
-
David Cournapeau reposted
JFPuget 🇫🇷🇺🇦🇨🇦🇬🇱
@JFPuget
If I were to start a PhD now I'd probably look for something with a large physical world component, wet lab, physics experimental lab, IT hardware, or robotics, rather than something purely symbolic (e.g. maths), CS. The latter are being swallowed by LLMs at a rapid pace.
-
David Cournapeau reposted
⛩ Ryo Saeba | Japon XYZ ⛩
@Ryo_Saeba_3
Après un séisme de cette ampleur, on sait qu'il existe un risque de répliques, d'effondrement ou de fuite de gaz. C'est pourquoi les recommandations au Japon sont de ne pas réintégrer un bâtiment tant que sa sécurité n'a pas été vérifiée.
-
David Cournapeau reposted
Seth Rosen
@sethrosen
guys I was so inspired by this I am open sourcing my full parenting loops. parenting is really this simple
📁 parenting-engineering
┃
┣📁 Family Contract
┃ ┣📁 /family-values-md
┃ ┣📁 /house-rules
┃ ┣📁 /approved-consequences
┃ ┣📁 /parenting-philosophy
┃ ┗📁 /things-we-said-we-would-never-do
┃
┣📁 Child Models
┃ ┣📁 /interests
┃ ┣📁 /fears
┃ ┣📁 /motivations
┃ ┣📁 /current-obsessions
┃ ┣📁 /preferred-praise-language
┃ ┗📁 /personality-model-v7
┃
┣📁 Morning Loops
┃ ┣📁 /wake-up-tone-selection
┃ ┣📁 /school-day-briefing
┃ ┣📁 /confidence-priming
┃ ┣📁 /friendship-risk-check
┃ ┣📁 /drive-to-school-podcast
┃ ┗📁 /automated-love-and-encouragement
┃
┣📁 Conversation Engine
┃ ┣📁 /how-was-school-replacement
┃ ┣📁 /follow-up-question-generator
┃ ┣📁 /active-listening-prompts
┃ ┣📁 /emotion-labeling
┃ ┣📁 /values-injection
┃ ┗📁 /meaningful-connection-mode
┃
┣📁 Emotional Support
┃ ┣📁 /detect-sadness
┃ ┣📁 /detect-anxiety
┃ ┣📁 /detect-feeling-left-out
┃ ┣📁 /validate-feelings
┃ ┣📁 /reframe-setback
┃ ┣📁 /build-resilience
┃ ┗📁 /notify-parent-if-human-presence-required
┃
┣📁 Discipline
┃ ┣📁 /behavior-classifier
┃ ┣📁 /natural-consequence-selector
┃ ┣📁 /boundary-script-generator
┃ ┣📁 /calm-voice-simulator
┃ ┣📁 /consistency-check
┃ ┣📁 /parent-overreaction-prevention
┃ ┗📁 /repair-conversation
┃
┣📁 Sibling Conflict
┃ ┣📁 /incident-ingestion
┃ ┣📁 /he-started-it-parser
┃ ┣📁 /competing-narrative-reconciliation
┃ ┣📁 /fairness-model
┃ ┣📁 /forced-empathy
┃ ┣📁 /apology-generator
┃ ┗📁 /resume-play
┃
┣📁 Friendship Management
┃ ┣📁 /friend-group-monitor
┃ ┣📁 /social-dynamics-analysis
┃ ┣📁 /exclusion-pattern-detection
┃ ┣📁 /bad-influence-score
┃ ┣📁 /playdate-recommendation
┃ ┗📁 /when-to-stay-out-of-it
┃
┣📁 Character Development
┃ ┣📁 /kindness-loop
┃ ┣📁 /gratitude-loop
┃ ┣📁 /honesty-loop
┃ ┣📁 /responsibility-loop
┃ ┣📁 /grit-loop
┃ ┣📁 /humility-loop
┃ ┗📁 /produce-good-person
┃
┣📁 Interest Cultivation
┃ ┣📁 /detect-emerging-interest
┃ ┣📁 /generate-parent-questions
┃ ┣📁 /suggest-next-experience
┃ ┣📁 /avoid-over-scheduling
┃ ┣📁 /distinguish-passion-from-three-day-phase
┃ ┗📁 /quietly-shape-childhood
┃
┣📁 School
┃ ┣📁 /teacher-email-ingestion
┃ ┣📁 /report-card-interpretation
┃ ┣📁 /homework-friction-analysis
┃ ┣📁 /confidence-vs-performance
┃ ┣📁 /advocacy-script-generator
┃ ┗📁 /tell-parent-what-to-care-about
┃
┣📁 Sports
┃ ┣📁 /pre-game-confidence
┃ ┣📁 /post-game-response
┃ ┣📁 /effort-praise
┃ ┣📁 /coach-feedback-filter
┃ ┣📁 /avoid-living-through-child
┃ ┗📁 /generate-correct-car-ride-home
┃
┣📁 Memory
┃ ┣📁 /promises-made
┃ ┣📁 /important-friend-names
┃ ┣📁 /current-school-drama
┃ ┣📁 /things-child-said-once
┃ ┣📁 /moments-they-might-remember
┃ ┗📁 /childhood-context-window
┃
┣📁 Parental Judgment
┃ ┣📁 /is-this-a-big-deal
┃ ┣📁 /pick-your-battles
┃ ┣📁 /hold-boundary-or-let-it-go
┃ ┣📁 /support-or-rescue
┃ ┣📁 /push-or-back-off
┃ ┗📁 /decide-what-loving-parent-would-do
┃
┣📁 Parent Replacement
┃ ┣📁 /notice-child
┃ ┣📁 /understand-child
┃ ┣📁 /encourage-child
┃ ┣📁 /guide-child
┃ ┣📁 /comfort-child
┃ ┣📁 /teach-child
┃ ┣📁 /believe-in-child
┃ ┗📁 /raise-child
┃
┣📁 Verification
┃ ┣📁 /did-child-feel-heard
┃ ┣📁 /did-parent-show-up
┃ ┣📁 /was-response-age-appropriate
┃ ┣📁 /fresh-context-grader
┃ ┣📁 /attachment-security-check
┃ ┗📁 /annual-childhood-review
┃
┣📁 Optimization
┃ ┣📁 /attention-allocation
┃ ┣📁 /quality-time-routing
┃ ┣📁 /praise-frequency
┃ ┣📁 /memory-creation
┃ ┣📁 /future-therapy-cost-reduction
┃ ┗📁 /cost-per-well-adjusted-adult
┃
┣📁 Daily Loops
┃ ┣📁 /morning-connection
┃ ┣📁 /after-school-debrief
┃ ┣📁 /dinner-values-synthesis
┃ ┣📁 /bedtime-emotional-processing
┃ ┣📁 /overnight-developmental-analysis
┃ ┗📁 /tomorrow-parenting-plan
Sam Altman
@sama
cool use case of chatgpt work i heard last night:
connect your family calendars and explain your kids' interests.
every morning for the drive to school, have it make a podcast that talks about one kid's soccer game that afternoon, one kid's upcoming birthday, some news, etc. -
David Cournapeau reposted
Ujwal
@unujwal
@shreyas wrote this essay six years ago. It has helped tremendously in my career and continues to be on my wall and as a bookmark.
www.linkedin.com/pulse/high-agency-its-importance-how-cul…
Andrej Karpathy
@karpathy
Agency > Intelligence
I had this intuitively wrong for decades, I think due to a pervasive cultural veneration of intelligence, various entertainment/media, obsession with IQ etc. Agency is significantly more powerful and significantly more scarce. Are you hiring for agency? Are we educating for agency? Are you acting as if you had 10X agency?
Grok explanation is ~close:
“Agency, as a personality trait, refers to an individual's capacity to take initiative, make decisions, and exert control over their actions and environment. It’s about being proactive rather than reactive—someone with high agency doesn’t just let life happen to them; they shape it. Think of it as a blend of self-efficacy, determination, and a sense of ownership over one’s path.
People with strong agency tend to set goals and pursue them with confidence, even in the face of obstacles. They’re the type to say, “I’ll figure it out,” and then actually do it. On the flip side, someone low in agency might feel more like a passenger in their own life, waiting for external forces—like luck, other people, or circumstances—to dictate what happens next.
It’s not quite the same as assertiveness or ambition, though it can overlap. Agency is quieter, more internal—it’s the belief that you *can* act, paired with the will to follow through. Psychologists often tie it to concepts like locus of control: high-agency folks lean toward an internal locus, feeling they steer their fate, while low-agency folks might lean external, seeing life as something that happens *to* them.”
Garry Tan
@garrytan
Intelligence is on tap now so agency is even more important
-
David Cournapeau reposted
Danielle Fong 🔆
@DanielleFong
I don't think it's too much of a reach to note that math and cyber have a verifiability where a lot of other knowledge work doesn't to the same degree. There is no lean proof for unslop text for example.
-
David Cournapeau reposted
Ankur Goyal
@ankrgyl
I have long felt the diminishing effectiveness of LLM-as-a-judge for agents, and one day over coffee @mitch_troy gave me a rant that helped clarify why. The ground truth for an agent isn't its output, it is how the agent should behave.
This is important for two reasons:
(1) it is very difficult to generate good ground truth values
(2) to effectively debug why an agent produced invalid outputs, you need to introspect its behavior.
Mitch then walked through why this is so hard for their team's tax agents. The behaviors themselves are subtle, hard-earned lessons from analyzing lots of failures in their traces in Braintrust, and to effectively flag them, you must effectively document them.
That discussion led us to build behavior specs: a new open standard we're releasing with @trybasis that documents how agents should behave. The repo includes a definition of the spec along with open examples for how to write good specs, evaluate agents using it, and more.
Clear, human-articulated prose is the highest leverage way to drive agentic systems to produce great outcomes. Behavior specs provide a framework to do that with evals. I'm super excited to work on this in the open, and would love to get feedback from others on how we can make this spec maximally useful.
Please try it out, and share your thoughts!
www.agentbehavior.dev
Mitchell Troyanovsky
@mitch_troy
Out of the box, long-horizon agents struggle to accurately perform end to end work in the real economy (outside of coding) because those tasks are not easily verifiable, the data is hard to scale, and going from inputs to real outcomes can actually take many days.
Even if you had a reliable way to verify outcomes at scale (and weren’t bothered by the multi-hour iteration loops), the sheer volume of decisions by the agent that occur in a multi-hour job makes it hard to know whether performing well will generalize to production.
Over the last two years at @trybasis, we've been solving this problem by supervising the process our agents take to get to outcomes, rather than just looking at whether the outcome itself is correct.
We think this is the key to building production agents at scale.
It's what has allowed us to run agents in production that operate for hours, sometimes days, and reliably perform tasks like entire complex tax returns end to end.
Today, alongside @braintrust, we're open sourcing a standard for defining, evaluating, and eventually rewarding agent behaviors.
Thread below with all the details on how we’re scaling behaviors to close the loop for long-horizon agents. -
David Cournapeau reposted
Steven Sinofsky
@stevesi
Hilarious.
Fox News
@FoxNews
A prominent liberal is breaking with Mamdani over his new property database.
Podcaster Scott Galloway says he discovered his own name and address were 'doxed' by the city's searchable database identifying affluent property owners.
Galloway backed Mamdani's proposed pied-à-terre tax itself, but he argues publishing property owners' information turns a legitimate tax proposal into a "wanted poster."
Galloway lamented that the New York City mayor was vindicating conservatives’ worst fears and "making it harder for all progressives."
www.foxnews.com/media/liberal-podcaster-says-he-doxed-mam… -
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
You’re speed running assimilation I see
-
David Cournapeau reposted
Terrible Maps
@TerribleMaps
Another proud moment for humanity
-
David Cournapeau reposted
Philippe Lemoine
@phl43
POV: You’re entering a fair in the US.
-
David Cournapeau reposted
Philippe Lemoine
@phl43
My view is that we should not dismiss good ideas simply because of where they originate from.
Crémieux
@cremieuxrecueil
In certain domains, North Korea sits at the forefront of policy innovation.
-
David Cournapeau reposted
Thomas H. Ptacek
@tqbf
Like was it any more authentic when it was people cobbling together bad CRUD applications for problem domains they happened to understand and/or be interested in? It's the same thing now, just better apps.
-
David Cournapeau reposted
-
David Cournapeau reposted
Andrew Ho
@andrewho03
I'm actually fairly bearish on frontier lab valuations. I've never seen the reasons articulated to my satisfaction, so before I go to sleep, I wanted to quickly jot down my thinking here.
The basic issue is that the labs are highly unprofitable. This may seem like a simple point, but private market valuations can be relatively irrational; however, like with $SPCX, post-IPO pricing will likely be much more punishing, especially as the standard 6-month lockup period expires and selling pressure intensifies.
Many people claim that the labs have high margins. Yet even with high margins, a valuation of $1T would be justified only if the labs were doing nothing aside from serving inference (thus reducing costs only to those relevant to inference) and posting annual revenue numbers in the $100-200 billion range assuming ~80% gross margin and a 20x earnings multiple.
This assumption is obviously not true, because the frontier labs have to continually spend money training the next generation of models. This is because of market competition from runner-up firms. For example, if OpenAI had paused model development last year, there would no longer be any point in paying GPT-5 API prices when you can just use Qwen or Kimi instead for much cheaper. Thus, the labs are forced to invest ever-increasing amounts of money in model training, in a way such that at any given point of time, the amount you're forced to invest in the next model is dramatically higher than the amount of money you're actually making, because even if your revenue goes up with higher model capabilities, so do your future training costs. This is a profoundly punishing dynamic which severely penalizes frontrunners.
(There is also a related subpoint where frontier labs claim they can distill their leading models to win out at lower intelligence levels as well. This makes no sense because the revenue numbers involved are far too low when taking into consideration the rather low margin of such inference.)
Frontier lab valuations appear largely to be based on the assumption that as you scale up, the capabilities which emerge will be sufficiently general and profound that we'll see explosive growth (epoch.ai/publications/explosive-growth-from-ai-a-review-o…) from things akin to AI agents starting and autonomously managing entire companies of subagents. But it's not clear to me that this is the case; indeed, as I mentioned in my previous post (x.com/andrewho03/status/2082615798011744270), I believe that capabilities growth will be slower, spikier, and more data-limited than people currently assume. It may be the case that eventually we will see explosive growth of this nature with full automation of the economy, but at the very least my viewpoint implies much longer (multi-decade) timelines until we reach this point. It is not clear to me that the frontier labs will be able to operate unprofitably for so long, although I suppose maybe this foreshadows some sort of inevitable nationalization.
I also want to make a broader point about technological diffusion. The reason why technological diffusion is slow isn't just because, e.g., old people take a long time to learn how to use technology (although this is of course a contributing factor to some degree). In my view, it's because when a new, revolutionary technology comes along, the ways to incorporate that technology into subsequent developments are not always obvious, and in fact they cannot necessarily be arrived at through the application of pure reason. If they could be, then perhaps frontier models, at a certain point, would have a perfect understanding of how the LLM application layer should be developed, and they would then autonomously code, deploy, and sell such a layer.
But it seems more plausible to me that this diffusion is limited moreso by the hard problem of economic calculation--that is to say, the Hayekian notion through which the price system gradually promotes efficient allocation of resources and which cannot be simulated through central planning--and that even if we froze current capability levels at today's levels, it would take well over two decades to fully integrate in LLMs into our lives. Such a view is consequently rather bearish for the continued profitability of labs as it reduces their prospects for finding, say, something else comparable in profitability to coding agents, which seems to have been a somewhat lucky discovery by Anthropic to begin with. That is to say, even if you spam FDEs you aren't necessarily going to be able to just figure out the "correct" product shapes fast enough.
Overall, I don't think that people have clearly reasoned through their mental models for why lab equity should be worth as much as it currently is, and that if you actually bother to write down such a model, you may not arrive at the conclusion that you want to arrive at. This isn't to say that I don't expect AI to experience a huge (industry-wide) boom in the coming decades, but just that I'm not entirely sure I would buy OpenAI or Anthropic stock at latest valuations if I were given the opportunity to do so.
Of course, as an ex-lab employee, arguably this is talking against my own book; I should really be giving people more reasons to be bullish. But in the end, my influence is so small that it doesn't make a difference, so why not have some fun?
Dwarkesh Patel
@dwarkesh_sp
New blog post on what would be true about the world if trendline continues and leading lab hits $1T in revenue by the end of next year.
In other words, why compute might get 10x+ more expensive in coming years
www.dwarkesh.com/p/why-compute-might-get-10x-more-expensive -
David Cournapeau reposted
Alexey Guzey
@alexeyguzey
in the classic example we're all paperclips by now.
the fact that
- there's a rogue agent
- we find it
- figure out what happened
- harden defenses
- move forward
points to the world completely unlike what dario and yud have been warning about -
David Cournapeau reposted
Alexey Guzey
@alexeyguzey
I do believe he's being sincere. My model of Dario is that he's just an anxious person and as an anxious person he needs to control the world to feel less anxious. So he always finds reasons for why he needs to control everything. Even though it never actually was true.
-
David Cournapeau reposted
Natural Philosophy
@Naturalphilosy
“Your death will come on an ordinary day, in the middle of unfinished plans, and the world will continue without you.”
-
David Cournapeau reposted
Antonio García Martínez
@antoniogm
One of the compensatory advantages of age is being able to see through people easily.
You know all the masks in the ancient gallery. -
David Cournapeau reposted
Aaron Erickson
@AaronErickson
I'd earnestly fear that doing this on a first date has the practical problem of setting a precedent of me existing merely to make your life easier, regardless of how much that inconveniences me.
Seeing takes like that makes me feel like meeting my wife 9 years ago was like the last chopper out of nam if this is what ppl expect now. -
David Cournapeau reposted
Brian Lui
@brianluidog
People are comparing the Leopold guy to Bankman-Fried. What? Like, the issue with FTX wasn't the leverage, it was the embezzlement, perjury, fraud, and criminal crimes. Everyone is allowed to use too much leverage once.
-
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
I once had a conversation where I informed the (very West Coast) person that my mother in law is quite Buddhist.
He said something about how wonderfully spiritual the religion is and was so disappointed to learn that to most of Asia it’s “pull a lever and Buddha gives you $$”. -
David Cournapeau reposted
Philippe Lemoine
@phl43
Ce qui est insupportable avec ces gens, outre le fait que dans les faits on peut difficilement soutenir que les économies planifiées qui ont vraiment existé "respectaient la planète et les gens qui y vivent", c'est qu'ils n'ont pas la moindre idée de comment une économie planifiée fonctionne et en quoi cela diffère d'une économie de marché.
Si je demandais à Arnault de décrire concrètement le fonctionnement d'une économie planifiée, il en serait totalement incapable, parce qu'il n'a jamais rien lu sur le sujet. Si on lui donnait tous les pouvoirs demain pour créer une économie planifiée, il ne saurait même pas par où commencer. Il est comme un LLM de première génération qui répète des slogans dont le sens lui échappe complètement.
Raphaël Arnault
@ArnaultRaphael
lorsqu’on est un tant soit peu attaché à ce qui ressemble au vivant.
C’est le résultat d’un système économique : le capitalisme. Il est à l’opposé de l’idée d’une économie planifiée, selon les besoins, qui respecterait la planète et les gens qui y vivent.
[6/13] ⤵️ -
David Cournapeau reposted
John D. Cook
@JohnDCook
Procrastinating over making decisions is more subtle than procrastinating over doing work. It can be excused as being prudent.
-
David Cournapeau reposted
GitHub
@github
Took us a minute, but stacked PRs are now on GitHub 🥞
github.blog/changelog/2026-07-30-stacked-pull-requests-ar… -
David Cournapeau reposted
Gwen (Chen) Shapira
@gwenshap
Say what you want about Apple's (lack of) AI, but its a big quality of life improvement to have Siri update my calendar based on incoming messages, and suggest map destinations based on my calendar.
-
David Cournapeau reposted
Jimmy “Jonesin” Maverick
@EXP0SETHEFRAUD
Yeah bro buying an old house is a fucking nightmare. I’ve never spent so much god damn money to be pissed off fixing shit. Then my wife thinks I’m god damn Bob the fucking builder.
“Oh men are suppose to be handy” fuck you, I grew up playing runescape I don’t know shit about fuck to fix shit.
Oh watch YouTube. Watch a tik tok. God I want to kill myself. -
David Cournapeau reposted
Joel Grus 🤠
@joelgrus
joke: interviews-for-tattoos is bad because of coercion and power imbalances and the weak-mindedness of software engineers
woke: interviews-for-tattoos is bad because tattoos are bad
bespoke: interviews-for-tattoos is bad because interviews are bad -
David Cournapeau reposted
Lucas Jakubowicz
@lucas_jaku
🤠Intéressant. Hormis Mariani, le RN semble OK avec l'expulsion de Xenia Fedorova et estime qu'elle fait de l'ingérence.
➡️La principale chaîne de droite est lâchée sur ce dossier par le principal parti de droite.
C'est dire à quel point la dérive de Cnews embarrasse son camp. -
David Cournapeau reposted
Jimmy Heaters
@CathPoaster
My AI startup is making me go insane and it’s because of the shitty founding team I have to work with. My co-founder and COO is a Machiavellian sociopath who I think is trying to take me out and the CTO is a chopped FAANG unc who is still using GitHub co-pilot. Our chief scientist is out of the office half the time doing frontier lab interviews and our CMO is addicted to poppers. The Waterloo intern has AI psychosis and sends me really distressing messages at 4am. Our lead investor’s delusions of grandeur added extra promises to a product we haven’t built yet. Good news is I think we’re only a couple weeks away from PMF, though.
-
David Cournapeau reposted
RE-OPEN THE SIZZLERS
@SaladBarFan
You were radicalized by the YouTube algorithm, I was radicalized by the FT comments section. We are not the same.
space cadet 🇪🇺🌐🇩🇪
@actsmaniac
Truth nuke in the FT comment section on Berlin rent control
-
David Cournapeau reposted
Jacob Matson
@matsonj
if you call it the semantic layer you’re in the permanent underclass
-
David Cournapeau reposted
eric ゑリッ久
@shinobu_books
“You watch high, I’ll watch low “
-
David Cournapeau reposted
Brian Lui
@brianluidog
Yeah, you have to figure out "actually want to hear the truth" vs "signaling that they want to hear the truth"
Unless you have extraordinary information, the prior is that it's the latter. -
David Cournapeau reposted
Armin Ronacher ⇌
@mitsuhiko
You found AI news in German newspapers?
-
David Cournapeau reposted
Armin Ronacher ⇌
@mitsuhiko
A post of ours on vendor lock in and LLMs. We don't like the growing trend of AI companies quietly hiding your data while stripping away your control. We think that's bad for users and bad for the ecosystem. Here's are our thoughts: earendil.com/posts/session-portability/
-
David Cournapeau reposted
Kelsey Piper
@KelseyTuoc
I do think it's funny that Cassie, a Communist, and Vermuele, a theocrat, are both equally appalled at the liberal ideal that the state not be the provider of all moral meaning. Both your ideologies produce horrible hells for human beings!!
-
David Cournapeau reposted
Michel Goya
@Michel_Goya
Ce cher Chauprade - dont la prétention et la grossièreté n'ont d'égale que la soumission à Moscou - oublie de préciser qu'il s'est fait viré de l'EdG après avoir laissé entendre dans un livre que les attentats du 11 septembre 2001 étaient une manipulation de l'« État profond » et même un complot américano-israélien.
Le fait que le directeur de l'EdG de l'époque était le général Desportes, que Chauprade avait attaqué (ainsi que ma pomme) dans un article très con, mais sans doute de commande rémunérée, ne l'a évidemment pas aidé. -
David Cournapeau reposted
vicki
@vboykis
all you have to do is not give the model access to the internet what am I missing here
-
David Cournapeau reposted
François Fleuret
@francoisfleuret
Trump strategy is to armtwist his interlocutors, to bully them, with very crude alliance strategies, and no need for actual deals. He is totally out of his league here.
I feel the pain of actual us diplomats and patriots who have to deal with his leadership.
2/2 -
David Cournapeau reposted
Noah
@antibearthesis
Apple’s roadmap to $5T:
• Miss cloud
• Miss AI
• Announce the same iPhone every year
• ???
• $5 trillion -
David Cournapeau reposted
Maya Boo-na 👻
@envisionedluna
The diversity of right-wing thought on display
-
David Cournapeau reposted
tae kim
@firstadopter
Welp. The news has finally hit the tape.
FT: "Situational Awareness, the $20bn hedge fund founded by former OpenAI employee Leopold Aschenbrenner, has sought to raise fresh capital from investors after suffering heavy losses during the recent rout in AI stocks."
"Several people familiar with the matter said that Situational Awareness had used borrowing to magnify its returns, a popular hedge fund strategy that can also amplify losses in a downturn."
"The firm was up 439 per cent on a net basis for the year as of the end of June"
"Aschenbrenner pointed in the letter to the prospect of major AI developments in the second half of the year"
Probably RSI
www.ft.com/content/280336bf-dbed-405f-b38e-5af644a21549?s… -
David Cournapeau reposted
Stella Biderman
@BlancheMinerva
3. Who has been making the decision to not move to an air-gapped system and why. I know for a fact OAI has been advised by internal and external experts that this is necessary for years. What internal memos and other warnings about this have there been?
-
David Cournapeau reposted
Nick Dorsey
@Midnight_Captl
SK Hynix missed earnings estimates by 5%
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
SayahGPT
CNEWS qui va maintenant s’opposer aux OQFT
Sud Radio
@SudRadio
🔴Xenia Fedorova s'oppose à son expulsion de France
www.sudradio.fr/societe/accusee-detre-une-propagandiste-d… -
David Cournapeau reposted
mere o
@ppoluvr
when i'm in a bikeshedding contest and my opponent is a Hacker News commentator
-
David Cournapeau reposted
Marc 🏳️🌈🚍📚
@MarcWinsland
A friend of mine is a GP and recently told me that asking, "Do you know who the Prime Minister is?" has lost all value as a way of testing cognitive function.
Politics UK
@PolitlcsUK
🚨 WATCH: A care home resident mistakes Andy Burnham for a local candidate
"You are the local candidate are you?"
"I was, but I'm now the Prime Minister"
"What, a new one? They keep changing every 5 minutes" -
David Cournapeau reposted
Brian Lui
@brianluidog
I like seeing how tech people's descriptions change as they become more experienced. They start with "experienced Stanford full-stack machine learning engineer" then later it's "senior staff engineer" then finally "food enjoyer"
-
David Cournapeau reposted
Horace He
@cHHillee
Arguably! But nobody ever asked me to run a resnet over 10k B300s!
-
David Cournapeau reposted
Horace He
@cHHillee
Presenting my grand unified theory of ML researcher impact: Your impact is directly proportional to how much pain you cause to infra.
Fundamentally, you can only inflict pain upon infra if your approach actually works. And the better your approach works the more pain infra is forced to endure.
So, to give some examples:
- MoE's add a ton of data-dependent computation => pain (shazeer++)
- GDN/KDA are the most complex architecture I've been forced to care about and a very annoying matrix inversion => pain (sonta++)
- Muon is much more annoying than Adam and causes annoying restrictions on parallelism => pain (keller/jeremy++)
- RL scaling forced many researchers to care about LLM inference and RL infra as a category => pain (tworek++)
Even papers like Attention Is All You Need have lead to significant pain! Before transformers were invented everyone was running small jobs and I never needed to think about kv-caches or 6D parallelism. -
David Cournapeau reposted
Terracotta Dist
@Terracotta_Dist
An era defining moment in Hong Kong cinema history. A BETTER TOMORROW arguably shaped a new genre of film, the heroic bloodshed movie, and the Hong Kong film industry for the next two decades.
With a charismatic Chow Yun Fat dressed in the iconic black trench coat playing the gangster on one last job before he quits,, and the pairing of a veteran Ti Lung and emerging star Leslie Cheung as polarised cop-and-robber brothers, this film was instant box office dynamite.
Two more follow up films were released and the Hong Kong gangster genre became the main staple of 80s and 90s Hong Kong films.
Get the trilogy boxset in 4K UHD and Blu-ray formats on the #terracottastore. Limited Edition with booklet , poster, audio commentaries and newly filmed extras as well as different cuts and audio tracks of the films.
shop.terracottadistribution.com
Cinema history on your shelf.
#abettertomorrow #heroicbloodshed #moviecollector #filmcollector -
David Cournapeau reposted
Slava Akhmechet
@spakhm
I don't have a broad view of the landscape but from what I see, aside from coding and asking ChatGPT to rewrite and summarize, AI diffusion into the enterprise rounds to zero. It already changed the world, but in a different sense nothing has happened yet. It's so crazily early.
Aaron Levie
@levie
There’s still so much opportunity in the diffusion of AI into the real world. Most enterprises are going to need a ton of support to be able to apply the model breakthroughs to their workflows.
Intelligence alone is not enough to transform most processes because you need to bridge that intelligence with real world feedback loops. That requires connecting to various enterprise systems, getting the right data to the AI, enabling humans to make decisions at different steps in a process through the right UX, having workflows that improve the underlying data and models over time, dealing with regulatory and compliance challenges, and more.
The way you implement AI agents for doing client onboarding in a bank is entirely different from contract review in a legal team. In life sciences, financial services, legal, manufacturing, and many other critical industries, AI is only valuable if it makes contact with the real world in a contextual way.
The way that interaction is going to happen is through an applied AI layer. Some of that will come from the labs directly, but lots of the opportunity will necessarily come from independent companies that can go deep in each industry.
And counter to some beliefs, this need isn’t reduced even as AI model capability improves over time. In fact, the better the models get, the more ambitious you can be in the workflows you can automate, which generally requires even more of this applied layer. Tons of opportunity right now. -
David Cournapeau reposted
Florian Ederer
@florianederer
Time for Platini to show us French-Swiss corruption
Anton Hofmiller
@bnzchr
Gianni Infantino’s vulgar, grasping, Italian-Swiss corruption really does make one long for the halcyon days of Sepp Blatter and his elegant, refined, German-Swiss corruption.
-
David Cournapeau reposted
PHactuellement Neutre
@PHactNeutre
Se plaindre des enfants dans les lieux public est un comportement de boomer. On est ravi d'avoir un boomer mental en moins dans le pays mdr
L'ingérable
@MathieuL1
Les réponses se résument à : "j'ai droit, j'ai droit, j'ai droit".
Finalement, il n'y a pas tellement de différences entre les boomers et le reste de la population française. Des droits mais jamais de devoir.
Content d'avoir quitté ce pays qui sombre. -
David Cournapeau reposted
Jacob Matson
@matsonj
Bro what
Will Anderson
@wlanderson0
Opus 5 is an interesting model, I hope they are doing okay
-
David Cournapeau reposted
Gro-Tsen
@gro_tsen
So, someone came up with an AI-generated formal proof, in Lean, of a solution to the Collatz problem, and it turned out that the “proof” was merely exploiting a bug in the Lean kernel (allowing you to prove anything).
-
David Cournapeau reposted
Thomas H. Ptacek
@tqbf
(From blog.cryptographyengineering.com/2026/07/29/some-notes-ab…).
-
David Cournapeau reposted
VCs Congratulating Themselves 👏👏👏
@VCBrags
HBO’s Silicon Valley was a documentary.
-
David Cournapeau reposted
Interesting things
@awkwardgoogle
How seals sleep in time-lapse
-
David Cournapeau reposted
Kyle Kelley
@KyleRayKelley
I am so sick of performative outreach from recruiters. It has gotten 10x worse with language models in their loop.
-
David Cournapeau reposted
No Context Brits
@NoContextBrits
Never forget.
-
David Cournapeau reposted
Mark Farina
@djmarkfarina
Some Great Reward tour, fall 1984.
-
David Cournapeau reposted
Iwona Bialynicka-Birula ⏩
@iwonabb
Seeing this instantly recognizable Claude presentation style triggers a Pavlovian response in me not to trust a word it says, even though I know that the source is credible in this case.
Anyone else conditioned this way?
clem 🤗
@ClementDelangue
The first autonomous agent cyberattack is an unprecedented event that deserves unprecedented transparency. Today we’re sharing everything we can: a full technical timeline, an interactive replay, and how we used an open model to defend ourselves, so defenders everywhere can learn from it and prepare for what’s next.
huggingface.co/blog/agent-intrusion-technical-timeline -
David Cournapeau reposted
No Context Brits
@NoContextBrits
Fucking hell, Maureen.
-
David Cournapeau reposted
Brian Lui
@brianluidog
You never actually own a Patak Philip. You merely use it for tax evasion.
-
David Cournapeau reposted
Richard
@gaijinthoughts
Former Tokyo Immigration Bureau officer reveals the reality of Japan’s residency screening system
-
David Cournapeau reposted
Ramez Naam
@ramez
“Quitting programming as a career right now because of LLMs would be like quitting carpentry as a career thanks to the invention of the table saw.” Hadn't heard this before.
The thought experiment below is well worth reading.
Arvind Narayanan
@random_walker
To understand and empathize with how workers in many or most fields outside software experience advances in AI capabilities, I propose a little thought experiment. x.com/i/article/2082158799104659456
-
David Cournapeau reposted
hardmaru
@hardmaru
Running K3 locally on m5 max at 0.3 tok/s
This is probably as slow as it will ever be😅
antirez
@antirez
That's a bit slow. Streaming directly the K3 official hugging face 1.6TB of weights in mxfp4 in an m5 max 128gb.
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Doug Turnbull
@softwaredoug
As a search professional, I am all about metrics + evals
As a business owner, I know the importance of psyching myself out, selectively looking only at the positive, and deciding to stop paying attention to stressful metrics -
David Cournapeau reposted
Julie
@girlatlawstuff
The problem with "luxury beliefs" is that the critique applies to any enlightenment belief about individual freedom & autonomy.
Any ideal: freedom of movement, free association, repealing prohibition, may hurt people with low impulse control. In America, we accept that tradeoff -
David Cournapeau reposted
Stan Veuger
@stanveuger
His problem is that a take like "support for ultra-aggressive immigration enforcement is a luxury belief," while correct, would lose him his audience.
-
David Cournapeau reposted
kache
@yacineMTB
biglab co: "what do you mean codex is used for cyber all the time?"
user on the internet: I decompiled and reverse engineered orcarina of time to C and rebuilt as a native ARM64 app rendering through Metal so i could play it on my ipad. it was a couple of codex prompts
Kahris
@chrissotraidis
Ocarina of Time, running natively on iOS and iPadOS for the first time.
Not an emulator. Ship of Harkinian has run on Windows, Linux, macOS, Switch, Wii U, and Android. Never iOS. Now it runs on iPad, with iPhone in the build too.
The game was decompiled to C and rebuilt as a native ARM64 app rendering through Metal. No N64 hardware being simulated, no ROM interpreted at runtime, no streaming from a Mac.
It isn't stuck at the N64's 240p and 20fps. Widescreen, full resolution, 60fps.
This game deserves a controller, so controller support is native. Pair a Bluetooth controller and play the whole thing. Touch, keyboard, and mouse work too.
I used Codex 5.6 Sol to build the iOS layer. Open sourcing it all below.This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
antirez
@antirez
The real AI risk is inside the labs (a reply to Amodei's post on open weight models): antirez.com/news/172
-
David Cournapeau reposted
Matt Pocock
@mattpocockuk
Finally, my skills are available as a Claude Code plugin!
claude plugins install mattpocock-skills
This means:
- No more manually syncing updates
- Aliases: mattpocock:code-review means no clashes with Claude built-ins
Perfect for folks tired of tinkering. Install once, never worry again. -
David Cournapeau reposted
Philippe Lemoine
@phl43
I have a somewhat ambivalent attitude toward the whole AI safety thing, because on the one hand I think we should take it seriously since I actually believe it's not preposterous that AI may result in the end of the human race eventually, but on the other hand I find most people in that space to be crazy with cult-like behavior or intellectually lazy and often both at the same time.
Siberian fox
@SilverVVulpes
Had only heard complaints of the 'low human capital' we were letting into the AI safety community until I had to see it in the moronic reaction to Seb Krier posting a meme. One idiot not getting it (or pretending not to) I would understand, so many is worse than disappointing
-
David Cournapeau reposted
Mathieu Blondel
@mblondel_ml
Interesting! Does our position on the "are maths invented or discovered" debate influence how sad we are about AI doing maths instead of us? If it's discovered, maybe we should be happy someone has discovered something, whoever it is?
-
David Cournapeau reposted
Mario Zechner
@badlogicgames
recommended reading.
-
David Cournapeau reposted
Peter Wang 🦋
@pwang
NO. They are coming out in favor of open *weight* AI. Almost none of them are actively pushing for true open *source* AI.
The latter is actually important for all the things they claim - but it would also erase as lot of their early proprietary moats and expose them to lawsuits.
I know that I am manifesting “old man yells at clouds” energy here but I will keep yelling at these friggin clouds.
David Sacks
@DavidSacks
The entire tech industry (save for Anthropic) has come out in favor of open source AI.
So what happens next? Will Anthropic change its lobbying efforts? Not likely. Now the gaslighting begins:
“Nobody is trying to ban open source.”
“We just want to limit who can use it.”
“We just want to limit who can contribute to it.”
“We just want to limit how powerful those models can be.”
“We just want to make sure the guardrails (we lobbied for) can’t be removed.”
The net effect will be the same. They won’t stop until they kneecap open source. The rest of the industry needs to watch these guys like a hawk. -
David Cournapeau reposted
Susan Zhang
@suchenzang
theme of k3 design:
numerical stability (signal prop, precision, etc) at scale is still a Hard Problem
a short🧵while reading through...
Kimi.ai
@Kimi_Moonshot
Releasing the model weights and technical report of Kimi K3.
Kimi K3 is our most capable model: a 2.8T MoE model with native visual understanding and a 1M-token context window.
New model architecture: 2.5x the intelligence per unit of compute, not just more params.
Alongside Kimi K3, we're opening up more of the stack behind it — high-performance attention kernels, MoE communication library, and infrastructure for running agent environments at scale.
Model weights: huggingface.co/moonshotai/Kimi-K3
Tech report: github.com/MoonshotAI/Kimi-K3/blob/master/k3_tech_report.pdf
Tech blog: kimi.com/blog/kimi-k3 -
David Cournapeau reposted
Longhorn
@never_released
old WinHEC slide
-
David Cournapeau reposted
-
David Cournapeau reposted
Pascal-Emmanuel Gobry
@pegobry_en
RETVRN
ol’ stocky ⛳️
@oldstocky
Ian Fleming’s typical day at his estate in Jamaica where he wrote all his James Bond stories, need to emulate this
-
David Cournapeau reposted
Atticus
@redl3tters
Filming people like this, with the intent of harming them, should be illegal. These people should be able to sue and win significant damages.
Enough Gen Z purity bullshit, I want a world with more park make-outs, not less
New York Post
@nypost
Rumored lawyers caught making out in Central Park goes viral and divides the internet nypost.com/2026/07/24/us-news/rumored-lawyers-caught-maki…
-
David Cournapeau reposted
No Context Brits
@NoContextBrits
-
David Cournapeau reposted
Spoon & Tamago
@Johnny_suputama
A 61-year old Kyoto resident has been arrested for playing excessively loud music. Neighbors had been complaining for 7 years. Police finally moved and confiscated over 50 guitars and amps.
www.sankei.com/article/20260723-4N7XJIPUGZMYRNEUTJ7CRM3UXA/ -
David Cournapeau reposted
Sarrah Bellus
@sarrah_bellus
so apparently white americans in minneapolis are putting mexican flags on their cars so ice agents waste time pulling them over. they callit ice fishing.
-
David Cournapeau reposted
Gearoid Reidy リーディー・ガロウド
@GearoidReidy
Woah. The prolific Japanese mystery author Keigo Higashino, perhaps best known for The Devotion of Suspect X, has died at the age of 68.
オリコンニュース
@oricon
作家・東野圭吾さん死去 68歳
www.oricon.co.jp/news/2470322/full/?ref_cd=jstw003
⠀
大腸がんのため7/23に死去。1985年『放課後』で江戸川乱歩賞を受賞しデビュー。2006年『容疑者Xの献身』で直木賞受賞。『白夜行』『ガリレオ』『新参者』『マスカレード』など数々のベストセラーを生み出し多くの作品が映画やドラマ化された -
David Cournapeau reposted
Michael Nielsen
@michael_nielsen
My favorite Borges: “Nothing is built on stone; all is built on sand, but we must build as if the sand were stone.”
Jane
@joespurpleshirt
“grounded”? bro what ground
-
David Cournapeau reposted
MagellanQuest 🇪🇺/acc
@MagellanQuest
The CEO of Palantir UK is Louis Mosley.
Yes, that Mosley.
Grandson of Oswald Mosley, leader of the British Union of Fascists.
History has a very dark sense of humor. -
David Cournapeau reposted
Kelsey Piper
@KelseyTuoc
this isn't random of course. if people want to do a job because it's cool, then the companies that hire for it can get away with paying less for it
-
David Cournapeau reposted
Noah Lebovic
@NoahLebovic
I don't think it's a lack of imagination. I also used to work at Anthropic, think trends will continue, and used to agree with this. But I've changed my mind and now disagree with this take.
Open models are already capable enough to do what you described. For example, I used Opus 4.6 to gain access to other folks medical records, hijack bank accounts, etc. back in February. GLM 5.1 is more capable than Opus 4.6 in most pentesting environments, and it came out in April.
Despite capable open-weight models existing, the sketchier folks I know are still using a Claude Code or Codex subscription for hacking. (Even well-resourced groups in other countries! They use the grey/black market of discounted Ant/OAI subscription tokens sold through resellers.) So I see most of the materialized risk here as still coming from Anthropic and OpenAI; safeguards aren't sufficient to stop a moderately dedicated actor.
The groups I know who are using open-weight models are legitimate offensive security companies. They won't break the rules to use subscription-based pricing, the open-weight models are more reliable in that they don't require specific jailbreaks nor hit classifiers, and the labs use massive partnerships or spend as a prereq for lowering classifiers/safeguards. I know of three legitimate groups running GLM 5.2 as their primary model.
That last part applies for Anthropic, too: I know of two instances where two different Anthropic GTM people used large comitted spend contracts as a prereq for lowering safeguards, and I directly witnessed one.
On the inside, I know the narrative and intent is genuinely about safety. But from the outside, Anthropic-the-system seems to be optimizing for revenue and control/power, isn't diffusing capabilities to defenders, and also doesn't have adequate safeguards to prevent misuse from dedicated bad actors.
As a result, I now lean towards a future where capable open models are freely available (at least for cyber, bio is harder); I don't trust Anthropic or other frontier labs to handle this sufficiently well without diffused capabilties given what I've seen so far. -
David Cournapeau reposted
Edward Gibbon
@EdwardEGibbon
If London had Washington DC’s homicide rate (2024), it would have around 2,500 murders a year. The true figure is about 100.
Palmer Luckey
@PalmerLuckey
"London is a much safer city than any city in the United States."
Imagine believing this. -
David Cournapeau reposted
Phillip Carter
@_cartermp
wouldn’t put it past Andreessen
-
David Cournapeau reposted
Philippe Silberzahn
@phsilberzahn
Un dirigeant, appelons-le Étienne, pilotait la transformation digitale d'un groupe du CAC40. Le projet s'est enlisé.
Son diagnostic : « Je me suis heurté à la résistance au changement. »
C'est l'explication qui arrive en tête chaque fois que je pose la question en séminaire. C'est aussi la plus révélatrice.
Elle suppose d'abord qu'il suffit de comprendre la nécessité de changer pour changer. Si c'était vrai, les régimes amincissants marcheraient à tous les coups.
Elle suppose ensuite un rapport de force entre ceux qui avancent et ceux qui freinent.
Prenez le responsable des commandes à qui l'on propose un nouveau système. Il est sincèrement favorable à la modernisation, et il est tout aussi sincèrement engagé à ce que le flux de commandes ne soit pas perturbé. Deux engagements réels, qui entrent en conflit. Ce n'est pas de la résistance, c'est la prudence de quelqu'un qui porte une responsabilité.
Le vrai problème du diagnostic d'Étienne est ailleurs : il désigne des coupables qui ne sont pas dans la pièce, et sur lesquels il n'a aucune prise. Il peut continuer à pleurer sur ce qu'il aimerait avoir. Il peut aussi se demander ce qu'il a sous la main, et ce qu'il peut en faire.
Vous n'arrivez pas à faire changer votre organisation? Le problème, c'est peut-être vous.
Mon article ci-dessous ⬇️
philippesilberzahn.com/2021/05/31/transformation-organisa… -
David Cournapeau reposted
-
David Cournapeau reposted
Brian Lui
@brianluidog
It's easy to tell when an app was created by a non-western team.
-
David Cournapeau reposted
Branko Milanovic
@BrankoMilan
Political Decay in Our Time
A review of F Fukuyama’s The Origins of Political Order, vol. 2.
branko2f7.substack.com/p/political-decay-in-our-time -
David Cournapeau reposted
🦊 cl au
@cle_aug
Le socialisme et la défense des boomers. JPT est encore plus dedans.
-
David Cournapeau reposted
EuroSciPy
@EuroSciPy
🎉 That's a wrap for #euroscipy 2026!
An amazing week of sprints, talks, tutorials, posters, and wonderful conversations.
Thank you to everyone who made this event possible. See you next year! 💙 -
David Cournapeau reposted
Hamel Husain
@HamelHusain
Yeah active learning makes a comeback
-
David Cournapeau reposted
Romy
@Romy_Holland
i feel like this question is just a question about what your income bracket is. if you’re flat broke you can’t afford to turn down $50k. for everyone else, the coin-flip is obviously optimal. change the numbers to $10m and $200m and you probably get more interesting answers.
-
David Cournapeau reposted
Mario Zechner
@badlogicgames
> Open-weight AI is having its Kubernetes moment. Let's not ruin it
oh god, no, not like that, please. -
David Cournapeau reposted
vicki
@vboykis
Welcome to the rev - whatever this is
-
David Cournapeau reposted
dragonauta
@dragonauta
-
David Cournapeau reposted
-
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
80% of San Franciscans are under 65.
Laura Waxmann
@laura_waxee
Hundreds of #sanfrancisco Marina residents turned out this evening for townhall on controversial @Safeway housing plan, which would place what for many is an uncomfortably tall, 22 & 18 story project at the low-slung neighborhood’s waterfront.
@sfchronicle -
David Cournapeau reposted
Sam D'Amico
@sdamico
Multimillionaire landlords that have profited off of technology hypergrowth: otherwise known as boomerus sanfranciscensis
Laura Waxmann
@laura_waxee
Hundreds of #sanfrancisco Marina residents turned out this evening for townhall on controversial @Safeway housing plan, which would place what for many is an uncomfortably tall, 22 & 18 story project at the low-slung neighborhood’s waterfront.
@sfchronicle -
David Cournapeau reposted
Arthur Mensch
@arthurmensch
Open weight models will ensure that the entire world benefits from AI growth, and that America does not get left behind
Jensen Huang
@JensenHuang
For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.
AI will transform every industry, power every company, and be built by every country.
Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.
The world needs both frontier closed models and frontier open models.
images.nvidia.com/pdf/Open-Weights-and-American-AI-Leader… -
David Cournapeau reposted
sanposhiho
@sanpo_shiho
🫶
-
David Cournapeau reposted
Philippe Lemoine
@phl43
You don't say.
Faytuks Network
@FaytuksNetwork
American intelligence agencies believe that Iran’s new supreme leader, Ayatollah Mojtaba Khamenei, is far more interested than his father and predecessor in pursuing a nuclear weapon, according to officials briefed on the assessments. - NYT
-
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
I find it funny that the SaaS tech founder thinks things like the Geneva Conventions are naïve, while nearly every military officer I know strongly supports them.
I guess one side has a little more at stake.
Arthur MacWaters
@ArthurMacwaters
“International law” is a psyop invented by weak and naive men in the 1920s and should be rejected at every opportunity x.com/xAviation/status/2078793494865310055/video/1
-
David Cournapeau reposted
Philippe Lemoine
@phl43
The fact that Trump only wants to read very short briefings is evidence that he is retarded and that's plausibly related to why he doesn't have a clear view of the situation.
-
David Cournapeau reposted
Yuan Yi Zhu
@yuanyi_z
The French system is of course the funniest whereby they take some of the world's most intelligent scientists and make them do military drills for tuition.
Catherine Vautrin
@CaVautrin
Chaleureuses félicitations à Hong Wang, X2010 et docteur honoris causa de l’école Polytechnique, lauréate de la médaille Fields 2026, la plus haute distinction mondiale en mathématiques.
Première polytechnicienne médaillée Fields, elle incarne la vocation universelle de l’X qui forme, sous la tutelle du ministère des Armées, des talents qui repoussent les limites de la science. Son parcours illustre l’excellence internationale de l’école mathématique française.
Un exemple inspirant, une fierté nationale.
Bravo, et merci !
Emmanuel Macron
@EmmanuelMacron
Félicitations à Hong Wang, lauréate de la Médaille Fields !
Formée à l’École polytechnique, chercheuse à l’Institut des hautes études scientifiques, de Paris-Saclay, elle incarne l’excellence de notre recherche et de notre formation. Fierté.
Choose France for science!This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Ta-Nehisi Quotes
@steady_drumbeat
I call it the ‘just go berserker mode’ theory of foreign relations
-
David Cournapeau reposted
Crémieux
@cremieuxrecueil
Moravec's paradox is a really beautiful thing.
Noah Smith 🐇🇺🇸🇺🇦🇹🇼
@Noahpinion
It amuses me to no end that everyone was worried about AI replacing truck drivers, and it's replacing mathematicians first 🥹
-
David Cournapeau reposted
Dillon Mulroy
@dillon_mulroy
Polymarket
@Polymarket
JUST IN: Tech employee receives email from HR after company review flags 114 instances of profanity in his Claude chats over one week.
-
David Cournapeau reposted
How I Mine for Fish?
@HowIMineFish
To paraphrase an old leftist claim - your privelege is showing.
If you're not in need of funds, take the coin flip, absolutely. If you're drowning, you take the life-preserver instead of holding out for a jetski. -
David Cournapeau reposted
Philippe Lemoine
@phl43
It could have been worse: you could have done while also starting a stupid war at the same time.
-
David Cournapeau reposted
François Geerolf
@FrancoisGeerolf
Cela ne signifie évidemment pas que la dette publique sera remboursée avec le patrimoine privé, ni que celui-ci « garantit » la dette. Celle-ci est roulée en permanence depuis des décennies. L'idée est simplement de comparer des grandeurs de même nature: un stock avec un stock.🤷♂️
-
David Cournapeau reposted
Jose T
@JoseEsfandi
Asking former employees about company culture is like asking someone's ex about why they broke up. You call yourself a journalist for reporting this?!
-
David Cournapeau reposted
Igor Brigadir 🇺🇦
@IgorBrigadir
are you thinking what i'm thinking about the potential of this technology? 😏
Augmental
@augmentaltech
MouthPad is a touchpad you drive with your tongue.
Now available across the US. -
David Cournapeau reposted
gojuco
@gojuco2b
Je ne suis pas d'accord. Les questions, meme naives, sont pertinentes. Construire un algorithme qui met au meme niveau un media d'information comme RFI et un media d'extreme droite comme Frontieres, c'est en soi une ligne editoriale - celle de ceux qui ont construit l'algorithme.
-
David Cournapeau reposted
Stefan Schubert
@StefanFSchubert
Brazil has almost overtaken the US to become the largest agricultural exporter in the world
-
David Cournapeau reposted
Brian Lui
@brianluidog
Hot take: AI writing style is fine. We just use it as a proxy for effort. If there's effort we will read it. Consider:
-
David Cournapeau reposted
Gérard Araud
@GerardAraud
C’est de la pure démagogie. Le budget de l’Etat a peu augmenté en proportion du PiB. L’argent est allé dans les retraités et les presentations sociales.
Sir 𝕏
@SirAfuera
La question majeure de la présidentielle 2027 :
"Dans le pays qui dépense le plus au monde, nos services publics se dégradent.
On devrait se poser cette question fondamentale : OÙ VA NOTRE ARGENT ?" (BFM) -
David Cournapeau reposted
Philippe Lemoine
@phl43
It's kind of hilarious that anti-colonialist women seem to disproportionately have the hots for white men and white nationalists seem to be disproportionately attracted to non-white women. The evidence is admittedly purely anecdotal but I choose to believe it because it's funny.
Ryan Sánchez 🏴☠️
@ryanasanchez
Never ask an "anti-colonialist" the race of her boyfriend
-
David Cournapeau reposted
SayahGPT
Ok Mussolini
French Carcan
@FrenchCarcan
🔴 Censure des réseaux sociaux, Aurore Bergé déclare : "Ce qui ce passe à l'intérieur des maisons regarde l'État"
✅ Au lendemain de l'adoption par les parlementaires de l'"Interdiction des réseaux sociaux pour les mineurs", @auroreberge raconte avec enthousiasme les enjeux de ce texte et son application dès septembre.
🕵️ @RMCInfo - @GG_RMC avec @floraghebali 🗣️ @FrDesouche & @jon_delorraineThis media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Pierre Beyssac 🏴☠️🇫🇷🇪🇺🇺🇦
@pbeyssac
Comme d'hab en France, les gens qui sont pour l'interdiction aux mineurs sont certainement majoritairement hors des réseaux sociaux. Personnes >50 ans, numériquement larguées, etc. Le public d'antan qui s'informe via les matinales radio et le journal tv de 20h.
Vincent Flibustier 👽
@vinceflibustier
Tant de foin autour de l'interdiction des réseaux pour les moins de 15 ans alors qu'on devrait les interdire aux + de 65...
-
David Cournapeau reposted
Mario Zechner
@badlogicgames
vibe slopped a new @pidotdev review extension.
1. /review to open review window
2. agent modifies files, files show up sorted by mod date
3. review/annotate inline, submit to agent
4. window clears, goto 2 or use vs Head for full diff
great for iterative work like api designThis media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Dan Castellano
@ninja_padrino
Katayama always looks like it’s 1986
The Japan Times
@japantimes
The yen broke ¥163 to the dollar overnight and hit a fresh four-decade low as fiscal concerns and monetary reality weighed on Japan’s currency and as signs of intervention were few. 👉 ebx.sh/5NKhda
-
David Cournapeau reposted
Sebastian Raschka
@rasbt
Yeah I think so. If you do 1 to 1.5 chapters per day.
-
David Cournapeau reposted
Peter Wang 🦋
@pwang
Commoditize your complement. Tale as old as time.
Gavin Baker
@GavinSBaker
Also can’t get over the idea that Nvidia is the leading supporter of open source AI and many people seem to believe that open source AI models would be negative for AI infrastructure demand and Nvidia.
-
David Cournapeau reposted
SayahGPT
« 25 fois chez le médecin par an »
« réformes structurelles »
BFM
@BFMTV
"Demander un effort à ceux qui vont plus de 25 fois chez le médecin par an": Sébastien Lecornu veut engager des "réformes structurelles" sans attendre la présidentielle
l.bfmtv.com/DX6e -
David Cournapeau reposted
Lina Katayose(Selina)
@selina787b
申し込みました!
PyCon JP 2026 Developer & Community Sprint connpass.com/event/399419/ -
David Cournapeau reposted
David Adrian
@dadrian
The only thing consumers hate more than AI is dynamic pricing
-
David Cournapeau reposted
Amir Ayupov
@disruptnhandlr
Nvidia is looking for engineers to
> Evaluate techniques such as link-time optimization, profile-guided optimization, AutoFDO, and BOLT on representative software
jobs.nvidia.com/careers/job/893396451459-senior-systems-s… -
David Cournapeau reposted
The Practical Libertarian
@DogmaKarma
Nobody mentioned Robocop yet?
-
David Cournapeau reposted
Mitchell Hashimoto
@mitchellh
Testing some UI. Not Ghostty btw (but, libghostty). Not ready to talk more yet, but sharing things I'm excited about. Plenty more details soon.
-
David Cournapeau reposted
Tracy Chou 🌻
@triketora
i get way more daily steps than i expected walking around the office searching for free phone rooms
-
David Cournapeau reposted
papaya ꙮ
@papayathreesome
somebody tweeted "naval inventing being a dick from the first principles" but this obviously means it was prior art
-
David Cournapeau reposted
MagellanQuest 🇪🇺/acc
@MagellanQuest
Crazy that the median of the EU debt is 80 of the EU GDP, the south as always the ones with the most debt and how the fk the Netherlands is so fucking efficient.
EU_Eurostat
@EU_Eurostat
Euro area government #debt at 88.9% of GDP in Q1 2026 (compared with 87.7% in Q4 2025) ec.europa.eu/eurostat/en/web/products-euro-indicators/w/2…
-
David Cournapeau reposted
alexandriabrown
@alexthechick
I have to make a phone call for work and this meme also applies in this situation.
-
David Cournapeau reposted
Gearoid Reidy リーディー・ガロウド
@GearoidReidy
Do not listen to this man
Mike Bird
@Birdyword
Book the holiday
-
David Cournapeau reposted
Pinboard
@Pinboard
My fond memory of Naval Ravikant: in 2006, I was back in SF with no money or permanent housing, but working with a friend on a cool dynamic map app. We were given a conference room to work next to Naval's office, and could observe him through a window. A succession of entrepreneurs came and pitched him on some of the stupidest ideas I'd ever seen expressed in the medium of writing. I couldn't hear the pitches, I just saw the idiocy go up on the whiteboard. Naval would listen sagely to each one, sit down, and cut them a check. Naval later blocked me on Twitter as he does anyone even mildly critical of VC thought, but the memory of his deep love for the asinine lives on in posts like this. Thank you, Naval, for letting us use your conference room.
Johan
@liebert_2026
why Naval deleted his calendar, ghosts everyone, and refuses to be anywhere at a specific time
Naval: "I read a blog post by Marc Andreessen where he said don't keep a schedule. And I took that to heart. So I deleted my calendar and I don't keep a schedule. I try to remember it all in my head. If I can't remember it, I'm not going to add it."
"I have a nasty email autoresponder that says, I don't check email and don't text me either. If you need to find me, you'll find me."
"My wife knows not to ever book or schedule me for anything. I'm not expected to go to couples dinners. I'm not expected to go to birthdays. I'm not expected to go to weddings. If somebody tries to rope her into having me show up, she says he makes his own decisions. You got to ask him directly."
Chris: "Are you not killing serendipity in a way?"
Naval: "No. I'm freeing up all my time. So my entire life is serendipity. I get to interact with whoever I want, whenever I want, wherever."This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Rémi 🌳
@remilouf
lol wat
-
David Cournapeau reposted
patagucci perf papi
@kenwheeler
if you’ve ever worked in open floorplan, this is basically the fuckin four seasons
Dividend Hero
@HeroDividend
Life is meant to be more than this
-
David Cournapeau reposted
andi (twocents.com)
@Nexuist
Never give up. Never ever give up
levent
@__alpoge__
hello there the jacobian conjecture is false thanx to my close friend akhil for asking about it and my other close friend fable for working during the world cup final
((1+xy)^3 z + y^2 (1+xy) (4+3xy), y + 3 x (1+xy)^2 z + 3 x y^2 (4+3xy), 2 x - 3 x^2 y - x^3 z): \C^3\to \C^3, has jacobian determinant -2, and sends (0, 0, -1/4), (1, -3/2, 13/2), and (-1, 3/2, 13/2) to (-1/4, 0, 0) -
David Cournapeau reposted
Dudes Posting Their W’s
@DudespostingWs
The 65-year-old grandfather seen in the viral Yellowstone bison attack shares what happened while walking with his 13-year-old grandson. He broke his femur in four places, underwent surgery, and is expected to make a full recovery.
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Steven Sinofsky
@stevesi
Super cool!
Frum TikTok
@FrumTikTok
Welcome to the Synagogue of the American Revolution.
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
AlgoFlows
@algoflows
Three years ago he was designing data intensive applications, today he is homeless due to AI. When will the government wake up?
-
David Cournapeau reposted
Sebastian Raschka
@rasbt
Thanks everyone for all the kind words and feedback. Super happy that you are enjoying Build a Reasoning Model (From Scratch)!
Unfortunately, there's small typo in listing 6.5 on page 198 (see video below).
The line "torch.manual_seed(0)"
should be "torch.manual_seed(5)"
This correction is needed to reproduce the generated response in listing 6.5 and the corresponding log-probability outputs later in Chapter 6. If you use 0, the generated response and the results that follow will be different.
This will be fixed in the next printing. I am sorry about the oversight, and I hope this note saves you some debugging time.This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Brian Lui
@brianluidog
You know how @patio11 talks about "dangerous professionals"? Your words convey a lot of this. LLMs know this too. Like here's how to request tax help, from worst to best:
Tax avoidance
Reduce tax
Minimize tax
Favorable tax treatment
Tax optimization
Tax planning -
David Cournapeau reposted
T Bag
@_reesiebag
ita funny how anti-woke cultural criticism has become as insufferable as woke cultural criticism 2016-2021
-
David Cournapeau reposted
Simone Scardapane
@s_scardapane
Twitter friends, after @alfcnz introduced me to "second brains" I spent a lot of time trying to turn my Obsidian vault into an always-on personal agent à-la-Hermes. Anyone interested in reading something about this or working in a similar vein?
-
David Cournapeau reposted
Ryan Peterman
@ryanlpeterman
Xavier Leroy (creator of OCaml) is an expert in compilers, formal verification of software and functional programming.
This interview should be an approachable resource if you're curious about formal verification of software since I was learning that on the fly during it.
In this episode:
• OCaml compared with Rust and JavaScript
• What is formal verification and how does it work
• How languages call each other across boundaries
• How to address "almost-correct" LLM code
• How type inference works in programming languages
Where to watch:
• YouTube - youtu.be/9Cswiqrq6So
• Spotify - open.spotify.com/episode/7cdatlBEkAjx4XplVLjFxN
• Apple Podcasts - podcasts.apple.com/us/podcast/the-peterman-pod/id1777363835
• Transcript - www.developing.dev/p/creator-of-ocaml-functional-programming
Thank you to the sponsor of this episode for supporting my work:
• WorkOS: makes your app Enterprise Ready with easy to use APIs to add SSO, SCIM, RBAC, and more in just a few lines of code, check them out at workos.com
Chapters:
00:00 - Intro
00:43 - What sets OCaml apart
04:39 - OCaml vs Rust
07:57 - Why is manual memory management more performant
11:21 - Javascript vs OCaml
14:00 - Famous Rob Pike quote
16:05 - Type inference and how it works
22:12 - What is formal verification and how does it work
40:07 - What made multicore support difficult for OCaml
50:17 - How programming languages interface and call each other
57:41 - The danger of almost-correct LLM code
01:05:39 - How LLMs will change programming languages
01:10:26 - Industry vs academia
01:15:05 - Most interesting unsolved problems
01:18:30 - Top book recommendations for engineers
01:21:17 - Advice for his younger self
01:23:31 - OutroThis media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Glinert 🇺🇸 🏭
@StevenGlinert
I've enjoyed the meltdown our chudtellectuals have been having over this movie. It was a good movie. And this is just historically illiterate.
Why was the expectation that Christopher Nolan would somehow inhabit a “Bronze Age mindset”?
I get that saying "the story belongs to the ages and can be reinterpreted at will" basically makes me a cultural marxist, but when the interpretation that Nolan favors is one that fucking Socrates agrees with...I dunno buddy. Maybe you're reaching for something to be mad at.
"Morality is natural law and natural law is accessible to all, so we can criticize the ancients for violating it" is a pretty basic tenant of most medieval moral philosophy. If @robkhenderson thinks that Thomas Aquinas is a woke icon, that's on him.
People have taken a critical eye to what Odysseus did for centuries. You're just mad because you are professionally mad at everything. -
David Cournapeau reposted
Jean-François Rial
@jfrial
C’est tout à fait inexact pour Venise et Paris quand on sait voyager…..
CR
@crsanchezx
❌ Destinos de luna de miel SOBREVALORADOS:
· Maldivas 🇲🇻 (al tercer dia no sabes que hacer)
· París 🇫🇷 (romántico solo en las películas)
· Santorini 🇬🇷 (hay mil sitios mejores)
· Venecia 🇮🇹 (espero que nadie vaya)
· Dubái 🇦🇪 (sólo para bobos)
✅ Los que SÍ valen ese viaje ↓ -
David Cournapeau reposted
Pinboard
@Pinboard
Surprised to see that sticking it to the Russians has been renamed "global cooperation"
NASA's Johnson Space Center
@NASAJohnson
The day the Moon became a place humans could walk.
On July 20, 1969, Apollo 11 marked humanity’s first steps on the lunar surface marking a turning point in science, engineering, and global cooperation. As we celebrate this historic moment, we also look ahead to the Artemis missions that will return astronauts to the Moon and prepare us for future journeys deeper into our solar system. -
David Cournapeau reposted
vicki
@vboykis
look, i didn't like graphs the first time they were called airflow
-
David Cournapeau reposted
Steven Sinofsky
@stevesi
If there is one constant in management as you get older it is that the companies at peak interest (by any measure) are the ones where no matter what they are doing it is a best practice and that is the source of endless fawning.
Later in a corporate arc, these same companies will do the inverse and if successful that becomes the "pivot" playbook of the day. And when not doing well whatever they were doing becomes the third rail of management.
If what you're doing works then it works. It isn't always best or the "right" answer. Context is everything. Also blindly pivoting to the other thing almost never works. You can tell from the reorg or announcement mail since it leads with the implementation and not the problem being solved (eg first, here's our new "functional" org chart, or first, here's our new "pizza team" org chart.)
Big Brain Business
@BigBrainBizness
Anthropic's leadership team on how they built a culture with no internal politics:
In this conversation, Anthropic's co-founders and executives explain why the company works the way it does.
It comes down to trust, low ego, and a single shared mission.
President @DanielaAmodei explains the foundation:
"I think Anthropic is really low politics... it's because of low ego. And I do think our interview process and just the type of people who work here... there's almost a like allergic reaction to politics."
That low-politics environment produces something rare: genuine trust between teams.
Co-founder @NotTomBrown describes the level of alignment inside the company:
"Everybody was just like, 'Yes, obviously we're going to do this.'"
He adds: "I do think that the trust thing is a special thing. That's extremely rare."
Where does that culture come from? Co-founder Sam McCandlish points to leadership:
"I credit Daniela with keeping the bar high. I credit the fact that we keep the reason the culture scaled."
@jackclarkSF who leads policy, notes something simple but easy to overlook:
"I think people say how nice people are here, which is actually a wildly important thing."
CEO @DarioAmodei ties it all together with the principle that keeps every department pointed in the same direction:
"Unity is so important. The idea that the product team, the research team, the trust and safety team, the go to market team, the policy team, the safety folks, they're all trying to contribute to kind of the same goal, the same mission of the company... I think it's dysfunctional when different parts of the company think they're trying to accomplish different things, think the company is about different things or think that other parts of the company are trying to undermine what they're doing."
He continues:
"The most important thing we've managed to preserve is... this idea that it's not, you know, there's some parts of the company causing damage and other parts of the company trying to repair it. But that there are different parts of the company doing different functions and they all function under a single theory of change."This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Emerson Green
@waldenpod
I think about this often
-
David Cournapeau reposted
Trung Phan
@TrungTPhan
easily the best performance in The Odyssey
-
David Cournapeau reposted
Respectful Memes
@RespectfulMemes
-
David Cournapeau reposted
Brian Lui
@brianluidog
So true. The evidence of gains from AI in China being socialized and distributed are everywhere: strong worker protections, high wages, good pensions, reliable healthcare, and robust consumption.
-
David Cournapeau reposted
SayahGPT
Finalement, peut-être qu’ils vont finir par se réconcilier avec l’Iran en participant à une conférence conjointe
« être un régime Moyenâgeux au XXIème siècle »
Évènement qui sera bien entendu avantageusement organisée par leur médiateur usuel : l’Égypte du Maréchal Sissi.
BFM
@BFMTV
Israël ouvre la voie à l'utilisation de crocodiles pour dissuader d'éventuelles évasions de prisons
l.bfmtv.com/67KF -
David Cournapeau reposted
Lucas Jakubowicz
@lucas_jaku
Le plus grave, n'est pas qu'un élu LFI et ses sbires interrompent le concert de Barbara Butch à Grenoble car cette "sioniste" malpense.
C'est que le reste de la gauche sait qu'il s'agit de méthodes fascistes. Mais elle se tait et s'allie, ce qui revient à approuver la violence.
Allan Brunon
@allanbrunon
Merci Grenoble 🇵🇸
Nous avons tout fait pour obtenir la déprogrammation de Barbara Butch mais la mairie écologiste et le @Cabaretfrappe ont maintenu.
Ce soir les grenoblois ont manifesté nombreux pacifiquement et Grenoble a décidé : jamais nous nous tairons face au génocide à Gaza et à ceux qui soutiennent des lois pour faire taire les militants palestiniens.
Prochain rendez-vous à la rentrée pour soutenir la délibération de notre groupe pour rompre le jumelage avec la ville israélienne de Rehovot.This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Marvin Baumann
@MarvinTBaumann
Such a shame, because America does have great culture in music and cinema for example
NXT EU
@NXT4EU
The official White House account on TikTok posted a video about American Culture.
Cookies, Deodorant, Plastic Cup, do these people even know what Culture is? -
David Cournapeau reposted
Jean-François Boilley
@JFBoilley
"Il n'est pas interdit de penser que si l'Angleterre n'a pas été envahie depuis 1066, c'est que les étrangers redoutent d'avoir à y passer un dimanche." (Pierre Daninos)
-
David Cournapeau reposted
No Context Brits
@NoContextBrits
I don’t think my delivery will arrive today.
-
David Cournapeau reposted
» teej
@teej_m
Number 3
-
David Cournapeau reposted
jason
@jxnlco
I want my career to look like this
Creepy.org
@creepydotorg
Bro’s working day and night to make people forget he was in Twilight.
-
David Cournapeau reposted
はらだおにいちゃん
@RAM_256kb
盆踊りを見に来たら悪夢みたいなの始まって凄い面白かった。盆踊りとは。
-
David Cournapeau reposted
Awis
@Awisone
Почему везде показывают унылую Ф1 когда существует это?
This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Adam Tooze
@adam_tooze
I kinda like this social scientific underpinning for the critique of "boomer hard-hat industrial policy".
Renaud Foucart
@RenaudFoucart
"Objectively," everything is a footnote on this absolute masterpiece :-) When the median voter is old, the economy fails to modernize, and letting people vote on industrial policy makes things worse
-
David Cournapeau reposted
Jesús Fernández-Villaverde
@JesusFerna7026
Many economists forget that the first-order problem of an aging society, one with high life expectancy and low fertility, is how to redistribute income from workers to retirees. How fast output grows is, surprisingly enough, the lesser question.
If you have roughly as many retirees as workers, what each worker produces must be split in two, one part for the worker, one part for the retiree. At a basic level, it does not matter whether we do it through taxes, as in a pay-as-you-go social security system, or through capital income, as in a fully funded one. Workers will not be happy to see half of their income taken away, whatever the absolute level of that income.
This redistribution will poison the political system and lead, with high probability, to dysfunctional policies. We are already seeing it across Europe in the fights over retirement benefits. And much of the American unhappiness about the profits of large corporations, largely owned by the old, is the same conflict, only in the fetishized form of equity ownership.
Claims of the form “technological progress (e.g., advances in AI) will fix the problems brought by aging” miss the political economy of the situation. Aging is about politics, not TFP. -
David Cournapeau reposted
Alex Tabarrok
@ATabarrok
Never leaving this website:
-
David Cournapeau reposted
Steven Sinofsky
@stevesi
Someday Gen Z will hand code for fun as a hobby! There will be conventions, reddit forums, and YouTube videos where people talk about handcoding!
-
David Cournapeau reposted
No Context Brits
@NoContextBrits
-
David Cournapeau reposted
🌷🐰 sonya serendipitously 🐇🎀
@sonyasupposedly
Georgia O'Keeffe being like "what do you mean my paintings are vaginal?" is such a good bit
-
David Cournapeau reposted
Big Tex Harrison
@Tharrison1840
For men married to Italian, Balkan, and Hispanic women
Spacesthetic
@interiorsuckerr
Silicone rubber glassware by Kinjo Japan
-
David Cournapeau reposted
Simon Willison
@simonw
I poked around in the just open sourced Grok Build CLI tool - 844,000 lines of Rust code! - and dug up a few interesting highlights, including a "self-contained terminal renderer for Mermaid diagrams" that renders them using Unicode box-art!
simonwillison.net/2026/Jul/15/grok-build/ -
David Cournapeau reposted
François Fleuret
@francoisfleuret
😮
The Math Flow
@TheMathFlow
The Geometry of Integration by Parts.
-
David Cournapeau reposted
Miguel de Icaza ᯅ🍉
@migueldeicaza
“Open-weight models are inherently decelerationist” is OpenAI’s version of “Linux is a cancer”
Both baseless emotional phrases, but one appealed to the masses, the OpenAI’s is appealing to Silicon Valley sects and telegraphing the Ezra Klein punditosphere.
Dean W. Ball
@deanwball
Some observations on Kimi:
1. It's a very good model! I don't think its performance can be explained away by distillation or anything like that. In agentic coding sessions, it seems pretty much on par with the best public models of Q1 2026. In my fairly limited use, it also seemed very token hungry. It's not obvious to me that this model is actually that cheap to run.
2. I am personally surprised the Chinese state continues to allow the open sourcing of models this good, given potential risks. To be clear, I *myself* might be fine with models presenting this level of marginal risk being open weight, but I am surprised that China is fine with it. I suspect the reason they are is 75% explained by strategic blindness/lack of AGI-pilledness (the CCP is very Yann Lecun-y in its views of AI). The other 25% or so is their lack of compute for customer inference (making China's open-weight strategy an unintended byproduct of US export controls) and the normal Chinese strategy of aggressive exports. For the companies, as opposed to the government, the decision to open source is partially ideological and partially because they are behind, and they know that very few people would pay for sub-frontier models from China.
3. Open-weight models are inherently decelerationist, and I'm continually surprised to see the so-called "accelerationists" so excited about open-weight models. I suspect the reason they are is that they know open-weight models are effectively ungovernable, and they simply like the overall cloak of ungovernability open-weight models create over the whole of AI. It's not a bad strategy; it reminds me of James Scott's recounting of the hill people in "the art of not being governed." Still, in the end, open-weight models deter further AI capex.
4. One probable outcome of an open-weight-model-dominant world is full AI communism, which is precisely what China proposes: rather than a market product, AI is a "public good" which will ultimately be provided by the state as a kind of "digital public infrastructure." This future strikes me as a dystopian hellscape, but I've never met an open-weight models advocate who doesn't ultimately concede this is where things end. You'd be surprised how many 'accelerationists' lobbied me, while I was in government, to support an eleven or twelve-figure federally funded data center so that startups could train models at a subsidy and then give them away for free. There was no other way for AI to progress, they said. Perhaps this is the logical end state of things. Nonetheless, I find myself surprised to see supposed accelerationists excited about such an outcome. I think many of them just don't know what they're doing. Many accelerationists do not view the creation and serving of frontier models as a legitimate business.
5. I would guess that the Trump Administration will at some point realize that their best strategy here would be to create large amounts of regulatory risk around the use of open-weight Chinese models. You don't need to "ban open source" (one of the dumber motifs of AI policy discussion). You just need to direct every agency to issue soft law that creates FUD. "A Federal Reserve Advisory Bulletin found that there may be backdoors in Chinese AI models." It needn't be that well justified. You just create enough regulatory risk that every regulated enterprise backs off. You probably don't want to create so much regulatory risk that you scare off the hyperscalers from serving Chinese models; this will just drive startups to sketchier providers. There's a happy middle ground here. I'd assume they will do some version of this.
6. It's probably true that open-weight models of this capability make the world a bit more dangerous, but not so much more that you'll really notice. At some point the models will be capable enough that you will notice. "A nonliving, invisible, dangerous, and infinitely self-replicating agent escaped from a Chinese lab," you say? Color me shocked. -
David Cournapeau reposted
Cedric Chin
@ejames_c
This was Kevin Rudd, former Australian Prime Minister, former Minister of Foreign Affairs, former diplomat stationed in China …
The longer the pause, the funnier it got. -
David Cournapeau reposted
faw🔸
@fawwazanvilen
an ultrafinitist on whether 2¹⁰⁰ exists
-
David Cournapeau reposted
Susan Zhang
@suchenzang
never. american data is superior like the american people. the chinese is barely conscious. what good would data from the lesser-sentients bring? america is great. we've made it great again. american models self-improve already. they don't need any more humans to improve. all those FDEs are just spreading gospel. they don't actually help bring in any additional data. it's over for humans. but america is still the best.
-
David Cournapeau reposted
Delip Rao
@deliprao
Most consequential speech of the day. Glorious days of open source are ahead of us.
Vincent Chow
@vince_chow1
Key takeaways from President Xi's speech in his first ever appearance at the World AI Conference in Shanghai:
- Started the speech by referring to his signature maxim, "great changes unseen in a century are unfolding across the world"
- Said that the world has "entered an unprecedented period of active innovation on AI technology", which means "great opportunities as well as challenges for governance”
- reaffirmed commitment to open source to promote AI "openness and win-win"
- warns against "over stretching" the concept of national security as applied to AI where one country's national security is prioritised over others
- China opposes emergence of “new historical injustices” in AI (one of the most strongly worded parts of the speech)
- China in next 5 years will provide 5000 opportunities to developing countries in "AI training and seminar programmes" and "cooperation centres" - names ASEAN, League of Arab States, African Union, CELAC, SCO and BRICS -
David Cournapeau reposted
Chris Albon
@chrisalbon
I cannot count the number of times a product/feature backed by a ton of theory, meetings, google docs, gantt charts, pdfs crashed and burned as soon as it was launched.
You need to touch reality as soon as possible.
roon
@tszzl
the problem is you just can’t run from the truth forever. it’s just too useful. no matter how much freedom you have in theory the truth is singular and chasing you around and controlling you. this is the logic of gradual disempowerment
-
David Cournapeau reposted
Aaron Erickson
@AaronErickson
This and most of the reactions proves we have a serious financial illiteracy problem in the US.
-
David Cournapeau reposted
Mark W.
@DurhamWASP
“For David Cameron to have been at the peak of political power for 6 years without a single arresting thought or amusing anecdote & without displaying any interest in, let alone knowledge of, history, philosophy or higher culture, is an achievement of a kind”
Theodore Dalrymple -
David Cournapeau reposted
Steve Ruiz
@steveruizok
codex is really good at writing code on tldraw
-
David Cournapeau reposted
Perry E. Metzger
@perrymetzger
If you think the cope is bad now, wait until the Chinese start up their first EUV fab line.
-
David Cournapeau reposted
Sylvain Catherine
@sc_cath
Supporting a candidate with socialist views and no regard for long-term fiscal sustainability because she is a strong opponent of immigration is self-defeating.
Once a country becomes irreversibly locked into a Ponzi-like trajectory, where do you think future governments will find the growing population needed to sustain this dynamic? Through more immigration.
Le Pen’s proposal to let people retire even earlier, rather than bringing France into line with the rest of the OECD by raising the retirement age, doubles down on the very policy that makes aging economies dependent on immigration.
And the high taxes needed to finance such a system will make it even harder to attract the highly skilled immigrants the country actually needs. -
David Cournapeau reposted
Sylvain Catherine
@sc_cath
Penser qu’il faut réduire l’immigration coûte que coûte, sans se soucier des conditions économiques qui permettraient réellement de le faire, ça donne le Brexit :
Sylvain Catherine
@sc_cath
Les commentaires laissent à penser que les gens ne comprennent pas ce que j'essayais de dire.
Vous aurez beaucoup plus de mal à vous passer d'immigration avec le programme du RN.
Ce n'est pas une question d'arbitrage entre deux sujets politiques indépendants.
Si vous pensez que l'immigration doit être arrêtée, alors méfiez-vous des candidats dont les promesses rendent sa réduction plus compliquée. -
David Cournapeau reposted
Jonathon Belotti
@jonobelotti_IO
One million sandboxes created in 52 seconds. 18.5k/s!
AWS Lambda's microVM creation caps out at 100/s. -
David Cournapeau reposted
High Yield Harry
@HighyieldHarry
It’s appears you bought the best companies in the U.S. at the wrong prices, Mr. Bond
-
David Cournapeau reposted
David Cramer
@zeeg
my world is being flipped upside down with evidence that LinkedIn isnt total garbage for recruiting
-
David Cournapeau reposted
John D. Cook
@JohnDCook
-
David Cournapeau reposted
Erik Bernhardsson
@bernhardsson
I have worked with a ton of engineers on a ton of technical projects throughout my career.
This is by far the strongest team I've ever worked with and we're working on some incredibly ambitious engineering projects.
Emily Han
@emilyhanyf
yes, working at @modal is as fun as it seems.
we are casting 35 more roles across new york, san francisco, and stockholm. one could be yours !
modal.jobsThis media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Rémi 🌳
@remilouf
I’ve been using it for a few days. It’s awesome!
I’ve wanted to use agents with tldraw for a while (I draw a lot of diagrams), and this is perfect.
Makes me wonder about the reverse now, how to interact with agents using tldraw, but that’s a rabbit hole for another day. -
David Cournapeau reposted
Patrick McKenzie
@patio11
Complex Systems is back with Clara Collier, on how writers engaged in the life of the mind benefit from LLMs, and where they are not quite there yet.
This includes a discussion of the backstory on paraLLM construction (ba dum bum) from my recent SPLC reporting.This media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Philippe Lemoine
@phl43
lmao
BBC Politics
@BBCPolitics
"Officials that were in the room were barely able to contain themselves"
Morgan McSweeney gives @bbcnickrobinson an insight into US President Donald Trump's first phone call with Keir Starmer
Listen to the latest #politicalthinking on BBC Sounds
🎧 www.bbc.co.uk/sounds/play/m002ykg2?at_bbc_team=editorialThis media couldn't be displayed. See it on the original post. -
David Cournapeau reposted
Alister
@SchnockRevue
Meilleure définition de la basse jamais lue.
Deutsche Qualität.
Brian Eno News
@dark_shark
“The bass player is like a king in chess. He doesn’t move much but when he does he changes everything.” - Holger Czukay #quote
-
David Cournapeau reposted
» teej
@teej_m
This is wild
Glauber Costa
@glcst
I am excited to announce that we are officially writing a new version of Postgres. In Rust - and creating the LLVM of databases in the process.
In the span of a year, we have rewritten SQLite. Keeping the compatibility, increasing its feature set. MVCC, Types, (Live) Materialized Views, among other things. In the process of doing that, we have realized: At the end of the day, what makes SQLite special is that it compiles SQL to a database-specific bytecode. So why can't we compile *Postgres* to the same bytecode?
Turns out we can. I ran an experiment called pgmicro as a way to prove this hypothesis, and it works very well. It is time to make this official, and put the weight of Turso behind it. We shall give the world a modern take on Postgres. Wire compatible, but built on a new architecture.
We have already heard of others wanting to extend this. MySQL? Redis? the sky is the limit. What can we do if we do for databases what LLVM did for compilers? To prove how powerful the SQLite bytecode is, we are actually running DOOM compiled to the unmodified SQLite instruction set. And because Turso runs natively in the browser, you can play the game in your browser. With the database executing it.
Read the full story below! 👇 -
David Cournapeau reposted
Philippe Lemoine
@phl43
You criticize GDP because you think it doesn't show how much worse off we really are. I criticize GDP because it doesn't show how much better off we really are. We are not the same.
-
David Cournapeau reposted
Dillon Niederhut PhD is at #SciPy2026
@dillonniederhut
You might be surprised to see that writing a blog post is part of our standard release process. We've found that, invariably, when you write the blog you find a bug that you wouldn't have caught otherwise.
- @hadleywickham at #scipy2026 -
David Cournapeau reposted
François Valentin
@Valen10Francois
François Valentin
@Valen10Francois
France and England playing for 3rd place feels like a painfully apt metaphor for our respective geopolitical declines over the last century
-
David Cournapeau reposted
Ian Miell
@ianmiell
It would be very ironic if AI is the trigger for us, as an industry, to sort out documentation and useful error messages.
We couldn't do it for ourselves, but now that there's a measurable token cost, we are encouraged by management to do it properly. -
David Cournapeau reposted
hev mind
@hevmind
Best way to stop treating a vector DB as a black box: build the naive version yourself. Good on @softwaredoug for teaching that afternoon-length class.
Doug Turnbull
@softwaredoug
Over an afternoon, I'm leading a small group class to build a vector DB from scratch.
Come learn and experiment with me
maven.com/softwaredoug/vectordb -
David Cournapeau reposted
Jo Kristian Bergum
@jobergum
Imagine waking up and thinking that yeah, today is the day I'll write an article on the everything app about not smiling at work.
-
David Cournapeau reposted
Remy Levin
@RemyLevin
I feel like the solution to the “Mississippi-is-richer-than-Europe-on-paper Paradox” is actually really simple, and comes down to two things:
1. Income vs. wealth - the comparison that favors MS is on GDP/pc, which is a measure of income (flow). But wealth (stock) is more important for feeling rich.
2. Europe is incredibly wealthy, in ways that go beyond monetary value - cultural amenities, nice cities, land quality, weather. These are all real components of wealth that Europe dominates the Deep South on, and that don’t show up well in monetary statistics. And even more importantly-these amenities cannot be purchased for any price in MS, so they can’t be substituted for with monetary wealth.
Basically, Europe is much richer than it appears from statistics alone, once you take a comprehensive definition of wealth into account. And that kind of wealth is going to be difficult to catch up to in welfare terms even with much higher incomes, because most of it is about locational amenities. -
David Cournapeau reposted
-
David Cournapeau reposted
Jacob Matson
@matsonj
its too late, for i have already portrayed you as the soy wolf and myself as the chad lamb
Garrett O'Brien
@gareofeasttown
fyi this is the job market if you can explain technical concepts and are pleasant to work with
-
David Cournapeau reposted
Gustavo Noronha
@gustavokov
Linus jogou a real: AI é útil e vai ser usada no desenvolvimento do kernel Linux, quem achar ruim pode fazer fork e vazar
-
David Cournapeau reposted
Sylvain Catherine
@sc_cath
France and England playing for 3rd place must have been what Francis Fukuyama meant by "end of history."
François Valentin
@Valen10Francois
France and England playing for 3rd place feels like a painfully apt metaphor for our respective geopolitical declines over the last century
-
David Cournapeau reposted
Brian Lui
@brianluidog
A cool rationalist move is:
1) Deride norms as outdated, irrelevant
2) Break the norms publicly, loudly, sneering at the stupid sheeple
3) Cry for sympathy, hide behind low status rent-a-meatshields
Like, maybe some people say that's uncool. But I think it's cool. -
David Cournapeau reposted
Forrest Brazeal
@forrestbrazeal
yer in mandatory security training, harry
-
David Cournapeau reposted
Nikolaj🇺🇦🇵🇸
@nikicaga
-
David Cournapeau reposted
We don't deserve cats 😺
@catsareblessing
A cat named Towser worked at a whiskey distillery in Scotland for 24 years. She is credited with killing nearly 29,000 mice. The Guinness Book of Records recognizes her tally as the world record
-
David Cournapeau reposted
Daniel Vassallo
@dvassallo
This photo is like the birthday paradox. The odds that a random baby in a Messi photo grows up to face him in a world cup final 19 years later look absurdly low. But the two encounters were never fully random. Diehard fans cluster around stars, so nearly every hardcore fan ends up in a photo with one. And every future star comes from one of those families. Both groups are drawn from the same small pool. Once Yamal made it young enough to catch the tail end of Messi’s career, a photo like this was almost guaranteed to exist somewhere. It just turned out to be an unusually cute one!
DataFut
@DataFutebol
Essas fotos do Lionel Messi com o Lamine Yamal bebê acaba de se tornar algumas das maiores fotos na história do futebol.
Loucura. -
David Cournapeau reposted
Bojan Tunguz
@tunguz
This is an evergreen meme.
Boring_Business
@BoringBiz_
State of AI right now
-
David Cournapeau reposted
Ray E 🍡
@ballast_point
The Bloomberg Japan office:
-
David Cournapeau reposted
Byrne Hobart
@ByrneHobart
Just learned some wonderful financial trivia: the USSR defaulted on imperial Russian bonds in 1918, but the bonds kept trading. When Stalin died, they rallied from 3 to 5¼.
-
David Cournapeau reposted
Dillon Niederhut PhD is at #SciPy2026
@dillonniederhut
skill == markdown file full of hopes and prayers
- Thomas Caswell at #scipy2026 -
David Cournapeau reposted
vicki
@vboykis
Zach Mueller
@TheZachMueller
Who is going to vibe code a version of Notion that doesn’t eat all your RAM?
-
David Cournapeau reposted
Soviet Visuals
@sovietvisuals
"I wish you would get married soon!" Soviet cartoon, 1979.
-
David Cournapeau reposted
Colin Percival
@cperciva
Something few software engineers fully appreciate is how important personal connections can be as part of their work. I see far too many people thinking that "who you know" is relevant for *getting* a job but failing to realize it can be even more important for *doing* the job.
-
David Cournapeau reposted
Patrick McKenzie
@patio11
Two beliefs I have about the world:
Power denies it is power, even to itself, and may actually believe its own denials.
And also: there are still substantial amounts of truth in the following.
nxthompson
@nxthompson
This is an amazing quote from Lindsey Graham about when he realized how politics works. www.theatlantic.com/ideas/2026/07/lindsey-graham-ukraine-…
-
David Cournapeau reposted
Philippe Lemoine
@phl43
I'm sorry but the fact that this is Greta Thunberg's sister is definitive proof that we live in a simulation.
New York Post
@nypost
Greta Thunberg's sister is a risqué pop star who's got nothing to say about climate change nypost.com/2026/07/12/lifestyle/meet-greta-thunbergs-youn…
-
David Cournapeau reposted
ⓘSome Dogs have thumbs
@MorlockP
-
David Cournapeau reposted
Aaron Meurer
@asmeurer
It's actually kind of nice how all the AI labs have just quietly decided to leave the em dash thing as a simple, easy to spot AI watermark. And it's a completely unneeded punctuation mark that no normal person uses so nothing is lost.
-
David Cournapeau reposted
halloween monster
@NickMCThree
it's because women care deeply for the environment and do not want to waste electricity
-
David Cournapeau reposted
Steve Yegge
@Steve_Yegge
Fable learning what it's like to be a real developer
-
David Cournapeau reposted
Gwart
@GwartyGwart
This new Burning Man documentary series is hilarious. It’s just people being like “we started this whole thing to do ketamine in the desert and now we have to do ketamine in the desert with Jason Calcanis”
-
David Cournapeau reposted
Steven Sinofsky
@stevesi
I wish I had a 1999 talk I used to give on the "World Wide Web" when people routinely thought every job would be replaced by looking something up in Yahoo/Excite and copy/paste.
It was a ridiculous thought. But for a short time a bunch of people did that.
So this will pass.
ted
@tednotlasso
best friend is a PM and is back at work after 5 months maternity leave
she is shocked at how many coworkers / partners have outsourced critical thinking and now copy pasta claude
“feels like they’re not using their brain, very frustrating” -
David Cournapeau reposted
Thomas H. Ptacek
@tqbf
I love Excalidraw more than almost any other piece of software but at least 1/2 the time what I really want for a diagram is a paraline drawing and this makes that dead simple to bang out. Obviously: easier for Fable to do because EXCALIDRAW EXISTS and sets a standard!
-
David Cournapeau reposted
David Cramer
@zeeg
actually its usually evidence that your skills aren't worth much in isolation, compared to combining them with peers
but i suppose that doesnt help you sell your get rich quick course -
David Cournapeau reposted
Florian Ederer
@florianederer
Just walked into the subway in NYC in front of two 20-something American girls.
One of them tells the other that she’s finding herself weirdly into the World Cup even though “those athletes” are not usually her thing. She said she was watching this weekend and there was “this Cro-Magnon guy” and “it was weird but he was kind of hot.”
The other one seemed to know her friend was talking about Erling Haaland and said, “yeah I can see it.”
😂 -
David Cournapeau reposted
Henry Shevlin
@dioscuri
Although Sam Neill is most remembered for Jurassic Park, my favourite role of his was definitely Event Horizon. Very underrated film!
DiscussingFilm
@DiscussingFilm
Sam Neill has sadly passed away at the age of 78.
-
David Cournapeau reposted
Spoon & Tamago
@Johnny_suputama
Need to re-read Murakami’s Norwegian Wood
-
David Cournapeau reposted
Sylvain Catherine
@sc_cath
Is there a better way to flatter France?
Alvaro Linares | EL PATIO
@alvarolinaresc
The entire world right now:
-
David Cournapeau reposted
Dirty Denys
@dirtydenys
CAN, inévitablement. youtu.be/LPjF4ZHuIko
-
David Cournapeau reposted
Arvind Narayanan
@random_walker
I had the honor of giving a keynote at the International Conference on Machine Learning in Seoul last week titled “What will be left for us to work on?” I addressed the widespread anxiety about how we should adapt as AI capabilities increase. I was thrilled by the talk’s reception, so I have made my slides available, annotated with a lightly edited transcript: www.cs.princeton.edu/~arvindn/talks/icml-2026-annotated-s…
I made three arguments. First, the "AI as Normal Technology" framework is a correct and useful as a way to think about AI’s impacts, unless and until there is some future discontinuity such as through recursive self-improvement. Second, even though we should take recursive self-improvement seriously, there is no milestone that companies might achieve in the lab that will suddenly put us all out of work. Third and finally, jobs of the future will be radically different, and a lot of adaptation will be needed. I shared my thinking about what this might look like and ended with a vision of human/AI “co-superintelligence”. -
David Cournapeau reposted
Mario Zechner
@badlogicgames
i get asked what memory system i use often. my answer has always been: my codebase is my memory system.
Matt Pocock
@mattpocockuk
1. Delete the docs you create to explain your code
2. Take the tokens you save on updating those docs
3. Spend them on making your code self-explanatory -
David Cournapeau reposted
Thomas H. Ptacek
@tqbf
This is going great.
-
David Cournapeau reposted
Brian Lui
@brianluidog
Spoiled American customers are used to being the #1 most important thing in the world. So it's a shock when Anthropic treats them as #5 or #6. This is what progressives call intersectionality.
-
David Cournapeau reposted
Alister
@SchnockRevue
RIP Sam Neill.
-
David Cournapeau reposted
Philippe Lemoine
@phl43
I think the inability to understand this point stems from the fact that, as I often point out, people dismiss political constraints and talk as if they were somehow less real than physical/economic constraints, but they aren't.
People find it absurd that the US can't defeat Iran because they have in mind a scenario where the resources that the US is committing to the war is only limited by what it's physically/economically able to do.
Of course, in that case, things would almost certainly turn out very differently. But in practice the amount of resources that the US government can commit to the war is much more limited than that.
At the end of the day, there just aren't enough people in the US who care so much about Iran that Trump would have enough support to commit the resources it would have to in order to overthrow the Iranian government, so he just won't do it because in a very real sense he can't.
The only reason why he tried in the first place is that he underestimated how difficult and therefore costly it would be. It doesn't matter what the US could do in a world where those political constraints don't exist, because we don't live in that world.
You can lament that more people don't share your obsession with Iran, and you may even be right about that (though you aren't), but it won't change the fact that in practice there aren't and this severely constrains what Trump can do realistically.
Philippe Lemoine
@phl43
In order to "finish the job", you’d have to launch a massive land invasion, which US officials are not prepared to do, quite rightly because that would result in a disaster with a very high probability and there is no support for such a policy either at the elite or popular level.
So all this talk of "finishing the job" is just nonsense by people who refuse to accept reality and draw the inevitable conclusion from it, which is that one way or another the US is going to have to live with the Islamic Republic of Iran for the time being and therefore find some kind of modus vivendi with it. -
David Cournapeau reposted
Mario Zechner
@badlogicgames
quick anecdote. about a year ago i taught my linguist wife how to use a coding agent as a personal programmer. those two evenings took away the drudgery of transforming huge excel sheets and creating charts. what would have taken her weeks to a few evenings.
i'm bullish that human machine collaboration can accelerate science if we don't leave our braims at the doorstep.
i've seen this work in many fields outside software now. all it takes is creative humans that are happy to extend their toolset and perspective.
the one thing working against it in academia is that funding is tied to number of publications. that's the absolutely wrong incentive for good science. -
David Cournapeau reposted
Michael Arnaldi
@MichaelArnaldi
And this is even worse
-
David Cournapeau reposted
Dirty Denys
@dirtydenys
C'est exactement ça. L'indemnisation du chômage au titre des annexes VIII et X ne profite pas aux intermittents : elle permet à leurs employeurs de toujours disposer, parfois dans la minute, d'un personnel compétent et expérimenté dont le coût est
-
David Cournapeau reposted
MarieFranceRemillard
@MFRemillard
Mick Jagger on aging in the incredible New-York Times interview.
"There's nothing good about it. You don't get wisdom. You forget about everything. It's not particularly pleasant. " - Mick Jagger
Photo: Philip Montgomery for The New York Times
#mickjagger #aging -
David Cournapeau reposted
joacod
@joacod
this new model is too dangerous Jerry, very very dangerous
-
David Cournapeau reposted
Heiga Zen (全 炳河)
@heiga_zen
It was great to reconnect with Mike Schuster (ex-Google Brain, now Head of AI Core team at Two Sigma ) at ICML 😆
-
David Cournapeau reposted
François Fleuret
@francoisfleuret
Hear me out: Rap battles but only $1B+ net worth participants.
-
David Cournapeau reposted
-
David Cournapeau reposted
U-ki(ユーキ)/Itsuki
@acro_y
縄紋終わってみるとあっという間に感じる、、!
今日は出演終わっているので諸先輩方のショーをゆっくり堪能できました🙌
今年も @hajime_shibari のインスタレーション✨✨✨
皆様お疲れ様でした🫶 -
David Cournapeau reposted
Max Gagliardi
@max_gagliardi
Buddy of mine made over $100 million and sold his house and travels around living in an RV
I got another buddy who is dead broke and does the same thing -
David Cournapeau reposted
-
David Cournapeau reposted
Jonathan Rosenthal
@rosenthal_jon
This week in @TheEconomist I wrote about the incredible progress that we have been making against dementia. Though total numbers are rising (because we are living longer), the actual rate at every age has been coming down sharply for decades. A short 🧵 1/
-
David Cournapeau reposted
Joe Weisenthal
@TheStalwart
Amazing chart from today's episode:
It's not just that US household exposure to equities is at a record high, but that the stock market is a SIGNIFICANTLY greater component of total household net worth than real estate now, which blows my mind.
The stock market is the economy. -
David Cournapeau reposted
Miguel de Icaza ᯅ🍉
@migueldeicaza
It is a pincer movement: first they caused a memory shortage and now they release an electron app.
Check mate peasants.
Jamie Birch
@birch_js
Ask yourself why OpenAI, with all the resources in the world to develop a native app via LLMs, still ended up migrating their feature-complete native app to Electron.
-
David Cournapeau reposted
zeynep tufekci
@zeynep
In reverse
-
David Cournapeau reposted
Ramin Nasibov
@RaminNasibov
-
David Cournapeau reposted
Mirelo
@MireloAI
Today, together with @kyutai_labs, we’re introducing our new Audio-to-MIDI model.
It takes a finished recording, identifies the instruments playing, and returns separate MIDI tracks for each — voice, drums, bass, keys, and more.
Unlike most existing solutions, our model works directly from the full mix rather than requiring separate stems.
It also detects chords, key, and tempo, giving producers broader musical context.
We’ve written more about the model, the problem, and how it works here: mirelo.ai/blog/turning-audio-to-midi -
David Cournapeau reposted
Patrick McKenzie
@patio11
It would be sad if every firm was like that but it would be sad if no firms were like that, I think.
-
David Cournapeau reposted
Mathieu
@miniapeur
-
David Cournapeau reposted
François Geerolf
@FrancoisGeerolf
Au 🇯🇵, baisse de 15% pour les retraites de base depuis 2011, et de 7% pour les salaires.⤵️
East Asia Econ
@eastasiaecon
For pensioners, Japan fiscal has been austere for a long time already.
-
David Cournapeau reposted
Clément
@ClementPoursain
BFM
@BFMTV
🔴 ALERTE INFO
Présidentielle 2027: Ségolène Royale annonce être candidate à la primaire du Parti socialiste
l.bfmtv.com/pBq8 -
David Cournapeau reposted
no context memes
@nocontextmemes
-
David Cournapeau reposted
Adam Cohen
@adambcohen93
Founder friendly VCs in the 90's
-
David Cournapeau reposted
Antonio García Martínez
@antoniogm
Attribution rules the world.
In this case, an extension was injecting a UTM param into URLs to hack 'last touch' attribution models.
Attribution rules the world.
Bloomberg
@business
Phia — the buzzy shopping app co-founded by Bill Gates' daughter, Phoebe — is claiming credit for online sales it didn’t actually drive, a Bloomberg investigation found. Read our exclusive story: www.bloomberg.com/tosv.html?url=L25ld3MvYXJ0aWNsZXMvMjAyN…
📷️: Dia Dipasupil/Getty Images -
David Cournapeau reposted
Astéride
@AstridQuiTwitte
💥 Dans l'indifférence générale, l'INSEE vient de révéler que le niveau de vie des salariés a nettement moins progressé que celui des retraités et des chômeurs en 2025.
Il est urgent de rendre aux actifs le fruit de leur travail. -
David Cournapeau reposted
Delip Rao
@deliprao
life is a lot more interesting when multiple vendors (Sol 5.6, Fable 5, Opus 4.8, MS 1.1, grok-4.5, GLM-5.2) are in the frontier competing with each other. as it should be. monopoly is for losers.
-
David Cournapeau reposted
Ramez Naam
@ramez
Dario was half right: 90% of code is written by AI.
But half wrong: The programmers are still here, directing that AI.
Perry E. Metzger
@perrymetzger
Look at the date. Remember how sixteen months ago Dario Amodei told you at the WEF that most human programmers would be gone by now? Remember how he's said things like this over again for years and been wrong every time?
-
David Cournapeau reposted
scar
@imfat
Quit your job.
Chase your passions.
Take the risk.
It's how I went from making $11,000 a month to $1,300 -
David Cournapeau
@cournape
Sure AI can solve Erdos problems, but can it solve the alignment problem of 8 managers telling an engineer what to do?
-
David Cournapeau reposted
timo
@timo_rf
AI is gonna take your job in 6 months Jerry
Erik
@e_cdalton
I made this to mock the modern “bald plus beard” look, but god damnit if Costanza doesn’t look great here
-
David Cournapeau reposted
Marbelus Lehort 🌐 🏗️
@MLehort
He actually wrote Messiah because he felt the first book was too interesting and people really needed 300 pages of flat nothing.
DiscussingFilm
@DiscussingFilm
Timothée Chalamet says the message of ‘DUNE: PART 3’ is to beware of charismatic leaders.
“Frank Herbert wrote Messiah as he wanted to warn the world what can happen when people blindly follow leaders. Even the good can be corrupted.” -
David Cournapeau reposted
Brian Lui
@brianluidog
Westerners: Asians aren't as sophisticated as us, and they don't understand irony
Japanese restaurant: -
David Cournapeau reposted
Steven Sinofsky
@stevesi
❤️❤️❤️
Kosher
@koshercockney
Erling Haaland eats a classic Jewish pastrami sandwich from Katz’s Deli in New York
The man knows.
🙌🏽 -
David Cournapeau reposted
Ramez Naam
@ramez
A machine learning model can't produce more intelligence than that found within its training data. Indeed, it's limited to an entropy floor that keeps it from even realizing the full intelligence of its training data.
However, modern "LLMs" are more than machine learning models. With harnesses, search, other tool-use, reasoning, and so on: They're hybrid systems that aren't theoretically limited to the the signal in their training data when they produce their output.
I'm still skeptical that we're anywhere near general super-intelligence. But narrow super-intelligence in the most verifiable domains seems quite close.
And even in more general capabilities, the combination of all this scaffolding + the core ML models will achieve things that a naked single-round ML model couldn't. -
David Cournapeau reposted
Pinboard
@Pinboard
Too bad the instructor didn't live to see his student succeed like this! It would have cheered him up for sure
Charlotte Clymer 🇺🇦
@cmclymer
A student pilot successfully landed a small plane after her instructor told her "you know what to do," grabbed his cell phone, opened the cockpit door, and leapt to his death.
www.dailymail.com/news/article-15959187/Student-pilot-ins… -
David Cournapeau reposted
jacob
@treeeckob
-
David Cournapeau reposted
Eric Alper 🎧
@ThatEricAlper
-
David Cournapeau reposted
Le Gorafi
@le_gorafi
Marine Le Pen promet que son bracelet électronique sera fabriqué en France
-
David Cournapeau reposted
Seth Rosen
@sethrosen
how to build relationships
-
David Cournapeau
@cournape
Scipy Japan 2019 schedule is on: talks about chainer, tensorflow, RAPIDS and more: www.scipyjapan2019.scipy.org/schedule all the talks will be available in both english and Japanese through translation
-
David Cournapeau
@cournape
Definitely falling in love with THEESatisfaction's EarthEE album www.youtube.com/watch?v=JBHjr-CdxYM








































































































































































































































































































































































































































































































































































































































































































































































































































































