Replying to @RILFhunter3
It has been assumed true for over a hundred years, and the suggested bounds on prime distribution do not assist in factorization. The global internet is already moving away from cryptography that is vulnerable to prime factorization attacks, anyway
23
Replying to @ca_tscratch
Correct, and correct. I think people are just confusing βprime distributionsβ as saying something about factorization as you note. And also perhaps overestimating the relevance of primes to modern cryptography (eg lattice/SVP based systems)
24
Replying to @sclv
Iβd hope so! Well, and affirming we are not in Crazy World, mathematically speaking. But I think itβs mostly just a misunderstanding of how βprime distributionβ relates to prime factorization (and its own relevance to crypto, which is actually fairly small)
1
1
66
Replying to @saintsoftness
Current dissections Iβve seen suggest a big theme seems to be assembling lots of ideas from all over. Itβs interesting to ask, re the mountain analogy: does a solution seem tantalizingly possible but elusive, or insurmountable without a whole new approach?
1
1
13
Replying to @stdlib @Staroxvia
There was some hubbub earlier in the year about someone using an exploited bug to post some cheeky proofs. Historically these bugs have happened in every major mechanized theorem prover, but it is of course a much larger problem now with these absurd sized proofs
1
34
Replying to @Staroxvia
A major ongoing effort by the Lean developers is to create multiple high quality implementations of the core semantics and run major proofs through them ("ron-con" and "con-leche", alongside "nandoa" etc). Every kernel bug (I am aware of) has been caught by this
2
3
106
Now you can instead have the machine do both halves simultaneously, so the "trailing" aspect is no longer a constraint and instead allows mechanical provers to be front and center. This is a reconfiguration of the social structure of mathematics above all else
1
2
34
Also, historically, "mechanical theorem provers" trailed far, far beyind actual human-level "handwritten" proofs. Thus there's this kind of idea that they serve as a sort of "insurance" to ensure human ideas are correct, or at least there was.
1
2
30
Many historic big results take tons and tons of time to verify, too. Poincare and FLT both took two years to have full proofs pieced together by the larger community. They had reasons to believe in the truth before that, different ones than some might have now. Is what it is!
1
2
26
But this isn't really how mathematicians work, like, at all. This is why someone like Levent can kind of make headway despite it being so large. So when people say this "250,000 lines" thing (wrong but ok) it's kind of at the junction of these two ideas of what "Proofs" are.
1
2
35
Now we kind of have this monkey paw thing where we have a generalized LLM spit out a bunch of proof terms, but make no mistake, entire mathematical programs/theses are dedicated to "how do we reduce 80 line proofs to 1 line?"
1
3
58
One other thing that people I think don't quite understand is that the academic history of mechanized theorem provers, such as Lean, is choc full of work on "proof automation". Proofs in tools like lean and what mathematicians call "Proof" are rather different in reality
I am having to go off private because the Riemann results are seemingly making a lot of people very anxious about "internet encryption" and various things, and it's hard not to try and clarify this as a guy who has written a fair amount of cryptographic code
1
8
550
It just really sucks seeing people dooming about this or extrapolating on it. This result is potentially monumental but not for those reasons!
8
206
I am having to go off private because the Riemann results are seemingly making a lot of people very anxious about "internet encryption" and various things, and it's hard not to try and clarify this as a guy who has written a fair amount of cryptographic code
Replying to @stdlib @saintsoftness
I say this as someone who has written a fair amount of cryptographic software myself! That said, I do think this is monumental. The highest mountain in mathematics has been reduced to a stepping stone on a forward path. To where? What is the next mountain? Hard to say!
1
1
27
Replying to @stdlib @saintsoftness
Actually, primes are actually a pretty small part of modern cryptography, except that case! Nearly all other modern constructions are safe against quantum computing, etc. I hope this doesn't come across as rude; but I've seen so much dooming over this and wanted to clarify
1
1
25
Replying to @stdlib @saintsoftness
Furthermore most modern internet standards are rapidly moving to post quantum algorithms, specifically to avoid quantum algorithms capable of "fast" prime factorization. When you connect to Google, you're already using post quantum algorithms based on other mathematical problems!
1
1
20
Replying to @saintsoftness
This is a common misunderstanding but it does not let you predict or otherwise "find" primes, you can't get a "formula for primes" out of this. Beyond that, we've assumed it's true for more than a century; it would be pretty bad engineering to put your chips on "Riemann is false"
1
2
35




