Postdoc studying dynamics, geometry and information processing in complex systems. Middling but enthusiastic musician
Joined July 2024
- Tweets1K
- Following396
- Followers124
- Likes5.7K
Eric Compeau retweeted
I think the proof was already defined this way. And it was a major blunder that prominent mathematicians played along with AI labs and pretended machine code is a proper math proof.
The response should have been: “Cute, but far below our standards and quite useless to us.”
Eric Compeau retweeted
Eric Compeau retweeted
Cornel West: Now Brother Barney is purple AND green. And even Brother Barney knows class struggle is not a struggle BETWEEN classes, but rather a struggle to eliminate class altogether.
Norm Finklestein: I’m sorry. I’m sorry. You would call this reptilian imbecile your brother?
Eric Compeau retweeted
The worker puts his life into the object; but now his life no longer belongs to him but to the object. Hence, the greater this activity, the more the worker lacks objects. Whatever the product of his labor is, he is not. Therefore, the greater this product, the less is he himself. The alienation of the worker in his product means not only that his labor becomes an object, an external existence, but that it exists outside him, independently, as something alien to him, and that it becomes a power on its own confronting him. It means that the life which he has conferred on the object confronts him as something hostile and alien.
Karl Marx, “Economics and Philosophical Manuscripts” (1844)
Eric Compeau retweeted
Really excellent account of the recent AIxMath developments from the perspective of an uncompromising ecologico-romantic sensibility.
Also made me chuckle to imagine Terence Tao dressed as Nite Owl.
“The ultimate hidden truth of the world is that it is something we make, and could just as easily make differently”
— David Graeber (1961-2020)
Link: tinyurl.com/y9a24tv9
Eric Compeau retweeted
Finally had the chance to read this properly last night, and it is delightful. In an age of unchallenged grifters, nihil is a true maverick. In my mind, I had an image of him as a mathematical cowboy, in the twilight days of the Old West, riding into the sunset defiantly, firmly defending a threatened traditional way of life.
Nihil's language may come off as abrasive or over-the-top to some readers (and indeed nihil openly admits this antagonism is deliberate; i.e. "it is the key to get people’s attention on X"), but in my view, it is absolutely crucial to have critical dissident voices in this historical moment, to at least balance the overwhelming current of the discourse... irrespective of how this critique is worded, or how well it propagates across the medium of X.
Particularly intriguing is nihil's (surprisingly civil) dialogue with Sebastien Bubeck of OpenAI, who obviously has a very different vision for mathematics. I would love to read Bubeck's response.
Another astute point concerns "intellectual property theft", which reminded me of the critical commentary of @provisionalidea (one of my fav accounts here) that periodically shows up on my feed, particularly this post [xitter.cf/provisionalidea/status…] as well as the criticism of AI safety theatrics (although I stress both writers have rather different arguments). These are subtle points that hardly receive the attention they deserve. One cannot help but feel that we are witnessing a "superficialization (or even commodification) of human intellectual work" as nihil put it.
Incidentally, I loved the Penumbra reference: "If fate frowns, we all perish.". Any other Penumbra fans in the chat?
“The ultimate hidden truth of the world is that it is something we make, and could just as easily make differently”
— David Graeber (1961-2020)
Link: tinyurl.com/y9a24tv9
Eric Compeau retweeted
>even sam altman would provide global UBI if openai reached singularity
if you believe this you have a level of gullibility and lack of reality testing in the domain of political economy that if it applied to activities of daily living would qualify you for a conservatorship
Eric Compeau retweeted
By far the most lucid account of the last few months. Read it.
“The ultimate hidden truth of the world is that it is something we make, and could just as easily make differently”
— David Graeber (1961-2020)
Link: tinyurl.com/y9a24tv9
Eric Compeau retweeted
“The ultimate hidden truth of the world is that it is something we make, and could just as easily make differently”
— David Graeber (1961-2020)
Link: tinyurl.com/y9a24tv9
Nice to read a thread that—unlike a lot of the bloviating from many public mathematicians—engages with the details of how integrating AI into mathematical practice may likely transform it.
Folks interested may also want to read Michael Harris' material on the matter (🔗s below)
TL;DR: “Every criterion for readability has been met, except for the presence of a reader.” — L. E. J. Brouwer, 1920s.
To expand:
1) First of all, I don't know and don't care to learn Lean. Working in Lean shouldn't be a prerequisite for mathematics and if it is I do not wish to engage with this skewed new vision of mathematics. I refuse to use AI to autoformalize either.
2) I would definitely read wrong proofs, if there are clearly stated and sufficiently interesting ideas. Famously, Dave Gabai spent a long time reading Valentin Póenaru's wrong proof of Poincaré conjecture and even after finding the mistake he would go on to consider it as one of the more interesting pieces of mathematics he's read. More broadly, I don't really look for theorems and proofs in mathematical texts, that's a shallow way to engage with mathematics.
3) Formalized mathematics devalued the structural need for readership. The way mathematics functioned before was this: the more interesting a given paper is, the more the paper was read by people, and exponentially smaller the chances there are of fatal errors because every mathematician has a different mental model of things from which vantage they would scrutinize the paper. After sufficiently many lateral attacks like that, theres virtually no chance of having a fatal error: the main results become like a highly connected network, hard to disconnect by taking out a few edges (isolated errors).
4) If we are limiting the domain of mathematics to "theorems and correct proofs" I don't know if many people would read at all if they can check something is correct at 0 cost. There will be a rush to build out mathematics vertically which can also be automated. Might as well leave all of it to the computer? Interesting mathematics was always born out of people struggling with a piece of writing, in turn born out of a desire to ensure validity (not in a formalist sense but in a more intuitive sense).
Thus, I posit that the formalist view of correctness that requires a Lean certificate is fundamentally at odds with "mathematics, to be understood by mathematicians".
Eric Compeau retweeted
It’s not a coincidence that the devastation at the NSF was concurrent with the OpenAI math dumps. There will be a much smaller piece of the grant money pie going to basic math research going forward.
Don’t surrender to the march of the pied pipers like Tao. Organize, and fight.
Eric Compeau retweeted
Fucking brutal
Thanks to Terry Tao for featuring me on his blog!
terrytao.wordpress.com/2026/…
Eric Compeau retweeted
Solving problems was always part of the quest, but never the whole quest
The real quest was for understanding, and solving problems used to be confirmation of that understanding
Not only has that link been severed, but much potential for understanding has gleefully been harmed
Eric Compeau retweeted
Probably the correct take on “AI can solve math now”
Well worth reading: proofsandprompts.com/2026/10…
Eric Compeau retweeted
It really is the society's burden to answer to themselves why they were or if they should continue to pay mathematicians. It's not my burden. Why did you value us so much that you threw trinkets our way? Think about it long and hard. Read the quoted tweet. Meanwhile I'll do math.
Eric Compeau retweeted
To all my mathematician friends: all the shit you're taking from this, many communities (artists) have had it a thousand times worse, while being even less responsible than us for these models. Some kind of class solidarity now is better late than never.
reading the pained responses of mathematicians, there’s a real cruelty in the way the labs have pursued this. not least because the attempted outcome in dropping so much unvalidated and poorly-written work (per the comments) is to push some of the world’s most passionate problem-solvers to be fact-checkers to a machine, as a kind of cynical cheap-as-free RL environment-by-nerdsnipe. dump a bunch of shit in public, call it solutions, get the experts to pore over it, painstakingly review and critique, scrape their efforts and use it to tweak the model, all while stoking terror about the future of the profession. it’s fucked up.
