Pinned Tweet
“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
— David Graeber (1961-2020)
Link: tinyurl.com/y9a24tv9
22
53
311
55,472
Joke's on you I get paid to sleep. All hail techno-utopia. Feel free to not care, and I'll feel free to sleep while earning money.
Is this a joke? Really crazy stuff!🫨 Of course they can refuse to use AI for math at their leisure, but who will pay them for academic jobs solving problems AI has already solved? Maybe someone will, but nobody will care about them besides each other. What a sad state they’re in
0
0
23
812
Другая планета retweeted
@ivanich22893 @nihilunbounded Except they didn't. Which proves there are still many prominent mathematicians in academia who are extremist formalists.
0
1
19
685
Другая планета retweeted
@nihilunbounded dylan going electric again (this time the amp is a context window)
0
1
2
145
Tokens at the break of noon
Lakebuild shadows the longest proofs
The handmade GPUs, the prompts aloof
To understand you know too soon
There is no sense in trying!
If Codex logs could be seen
They'd put my head in a guillotine
It's alright ma! It's slop and slop only.
Lakebuild shadows the longest proofs
The handmade GPUs, the prompts aloof
To understand you know too soon
There is no sense in trying!
If Codex logs could be seen
They'd put my head in a guillotine
It's alright ma! It's slop and slop only.
2
0
8
578
Sorry to be that guy but any pair of homeomorphic closed oriented smooth 4-manifold are diffeomorphic after sufficient stabilization, so the meme is a bit strange :-)
[Proof-idea: M homeo to N implies M x R homeo to N x R, improve to diffeo using Kirby-Siebenmann smoothing theory since we are in dimension 5, embed M x 0 in N x R smoothly to obtain a smooth cobordism between M and N, use Kreck's modified surgery]
[Proof-idea: M homeo to N implies M x R homeo to N x R, improve to diffeo using Kirby-Siebenmann smoothing theory since we are in dimension 5, embed M x 0 in N x R smoothly to obtain a smooth cobordism between M and N, use Kreck's modified surgery]
Liked this one.
Any guesses for how would it look in 2028/30/32?
1
3
62
2,316
Другая планета retweeted
i think both the formalists and the intuitionists are fucking dorks send tweet
0
2
12
379
I'd be genuinely quite interested to learn what these breakthrough ideas are, rather than just hearing rumors of their existence on Twitter. I did not get a lot out of Buckmaster's talk on the Euler equations.
@Frances01896069 people who understood the NS solution told me that it contains 10 field-shaping breakthroughs as part of the solution. The idea that the models only produce results in the convex hull of human knowledge (whatever that means) is misguided.
7
3
93
4,706
Другая планета retweeted
@nihilunbounded 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.”
The response should have been: “Cute, but far below our standards and quite useless to us.”
1
2
9
783
Thanks for the question!
I don't think about the four color theorem; I try not to comment on things I don't understand (although a proof along the lines of Kronheimer-Mrowka would be awesome; see arxiv.org/abs/1508.07205)
But in general I think (and of course, sinxe I am a nobody, you can vehemently disagree) a proof should be redefined to be something that brings insight and understanding, and not (just) a verifiable collection of symbols that lead to a certain truth value of the original statement. This may require overturning a century and a half of mathematician's conception of what constitutes a proof, but I think it is a worthwhile endeavour (and I also think the alternative is paperclip universe where LLM slop rules math and very few people understand wtf is going on).
I think a computer assisted proof can do that (eg I have used SnapPy before). I think some nontrivial amount of rich mathematics is lost in this great project to pin down mathematics to the domain of dry formalism. This is most noxiously evident in LLM proofs with an autoformalized Lean certificate.
I don't think about the four color theorem; I try not to comment on things I don't understand (although a proof along the lines of Kronheimer-Mrowka would be awesome; see arxiv.org/abs/1508.07205)
But in general I think (and of course, sinxe I am a nobody, you can vehemently disagree) a proof should be redefined to be something that brings insight and understanding, and not (just) a verifiable collection of symbols that lead to a certain truth value of the original statement. This may require overturning a century and a half of mathematician's conception of what constitutes a proof, but I think it is a worthwhile endeavour (and I also think the alternative is paperclip universe where LLM slop rules math and very few people understand wtf is going on).
I think a computer assisted proof can do that (eg I have used SnapPy before). I think some nontrivial amount of rich mathematics is lost in this great project to pin down mathematics to the domain of dry formalism. This is most noxiously evident in LLM proofs with an autoformalized Lean certificate.
@nihilunbounded Okay, so I’m curious, do you also consider the four color theorem to not be proved? Mathematicians have been coming up with computer assisted proofs long before Lean. If you think these are not valid proofs, why aren’t they? And if they are, how is a Lean certificate different?
7
1
50
4,360
Другая планета retweeted
@nihilunbounded я пришёл дать эту leanslop [x4] ahaha, haha, haha...
Чандрам Брамбачандра Чандра Бендрам. [x4] woah...
{co stało się, stało się to co miało się stać
dopiero w jej משכניו, nosz kurde mać!
I gdy szła pod oblokiem, wolnym krokiem
Jakiś pajac podbił mi limo pod okiem} [x4]
Чандрам Брамбачандра Чандра Бендрам. [x4] woah...
{co stało się, stało się to co miało się stać
dopiero w jej משכניו, nosz kurde mać!
I gdy szła pod oblokiem, wolnym krokiem
Jakiś pajac podbił mi limo pod okiem} [x4]
0
1
2
177
Другая планета retweeted
@nihilunbounded "ZOMG — FIVE MILLION barrels of PROOFS produced in _JUST_ 87 DAYS! PROOF ABUNDANCE! What are you COMPLAINING about?!"
1
2
18
748
Другая планета retweeted
When I transferred to Stony Brook in the 90s from Community College, I spoke to Tony Knapp about math interests involving popular books at the time like Escher, Godel, Bach: Eternal Golden Braid.
He told me point blank that Godel's work was not real math. I doubted him at the time, but always valued his advice.
Nowadays I know my doubt was justified. Godel says everything about today's math.
He told me point blank that Godel's work was not real math. I doubted him at the time, but always valued his advice.
Nowadays I know my doubt was justified. Godel says everything about today's math.
2
1
12
691
Другая планета retweeted
"20 GB of Lean? My 32-bit computer without Intel PAE can only address up to 4 GBs. This is not a proof, this is theology! GET THE FUCK OUT OF HERE! DIE! DIE!!!! DIE!!!!!!!! I'M GOING TO FUCKING KlLL YOU (IN REAL LIFE)!" ~ Alcuin (ultrafinitist, user of old Pentiums.)
I'm 2/3rds of the way in. Truly incredible essay.
Those who confidently proclaim mathematics is a science should remember it was basically a disaster zone of conflicting cults for the better part of 2000 years.
So if anyone comes at me with a 20gb Lean file I do not for a single moment hesitate to say "This isn't a proof of anything by any reasonable standards except those set by mathematicians at most 100 years old, so up yours."
Screw you 🫵 and your 🫵 Lean certificates. I don't consider your theorem to be proved.
0
1
30
980
I'm 2/3rds of the way in. Truly incredible essay.
Those who confidently proclaim mathematics is a science should remember it was basically a disaster zone of conflicting cults for the better part of 2000 years.
So if anyone comes at me with a 20gb Lean file I do not for a single moment hesitate to say "This isn't a proof of anything by any reasonable standards except those set by mathematicians at most 100 years old, so up yours."
Screw you 🫵 and your 🫵 Lean certificates. I don't consider your theorem to be proved.
Those who confidently proclaim mathematics is a science should remember it was basically a disaster zone of conflicting cults for the better part of 2000 years.
So if anyone comes at me with a 20gb Lean file I do not for a single moment hesitate to say "This isn't a proof of anything by any reasonable standards except those set by mathematicians at most 100 years old, so up yours."
Screw you 🫵 and your 🫵 Lean certificates. I don't consider your theorem to be proved.
Mathematics is 100% dead 🪦
A cause for celebration 🎉
Read my eulogy and weap 😭
Link👇
10
12
170
11,366
Are you a young up and coming slopmatician in Math 2.0? Are you struggling to explain to your friends and family what a “Lean certificate” is? Worry not, for @doomslide is here to explain Lean in one simple paragraph that catches the imagination of anyone from age 7 to 70 who might otherwise question your sanity!
“A proof assistant is a program that checks every step of a proof line by line as a blind head would but with no need of a head. Lean, the checker that mathematicians use now, is built like a turducken, with the small program stuffed inside Brouwer and Brouwer stuffed inside Hilbert, so that proofs can show, as Hilbert’s did, that something exists without producing it.”
“A proof assistant is a program that checks every step of a proof line by line as a blind head would but with no need of a head. Lean, the checker that mathematicians use now, is built like a turducken, with the small program stuffed inside Brouwer and Brouwer stuffed inside Hilbert, so that proofs can show, as Hilbert’s did, that something exists without producing it.”
Mathematics is 100% dead 🪦
A cause for celebration 🎉
Read my eulogy and weap 😭
Link👇
2
3
34
2,356