Noah Snyder

[Noah Snyder @NoahJSnyder]

"This may be a graduate class, but it's still called Algebra, so we're doing the quadratic formula on the first day!" --me, apparently

[Noah Snyder @NoahJSnyder]

FWIW this is no longer true. GPT5.5 is very good at routine skein theory calculations.

[Noah Snyder @NoahJSnyder]

@littmath Is any of this about lack of “geometric intuition”?

When I’ve tried to have LLMs do anything skein theoretic, they’re just shockingly bad at even very simple geometry. Would kinda make sense, a ton of humans training data is navigating 3d space.

[Noah Snyder @NoahJSnyder]

@MattStrangelove Athletes are the only celebrities who are regularly tested for steroids.

[Noah Snyder @NoahJSnyder]

@Ognifedefingo AI is very good at *routine* proofs across all subjects very quickly. It’s still not usually very good at things that aren’t routine for someone. Like if it’s something 500 people can do it’ll do it immediately, but if only 5 can it probably gets stuck.

[Noah Snyder @NoahJSnyder]

@samth @bdquinn Doh! I did know that, I’ve even been to Juneau.

I think my ratio of Geography knowledge (3rd in PA in Geo Bee, 81% LearnedLeague) to capital knowledge (would not get more than 40ish state capitals and would do even worse on countries) is especially extreme.

[Noah Snyder @NoahJSnyder]

@samth @bdquinn Honolulu. SLC. Phoenix. Anchorage. Maybe Boise? But I don’t think they’re *that* hated.

(Atlanta of course, but already discussed.)

But yeah many of the main examples in this category (NYC, Chicago, Vegas) aren’t capitals.

[Noah Snyder @NoahJSnyder]

@samth @bdquinn Boston maybe doesn’t fit any of these? (Hated for its size and dominance, not because it’s the capital per se.)

[Noah Snyder @NoahJSnyder]

@HannaCanadia E has Banff and Drumheller.

[Noah Snyder @NoahJSnyder]

@TheGothicOwl @FailureUltra @Isaac2003_v2 It’s not the same kind of remote, but the coast along Moray and Aberdeenshire is also a lovely place to spend time.

Edinburgh to Cairngorms to Speyside distilleries to Murray coast and then the ferry Aberdeen to Orkney and Shetland is a fantastic few weeks.

[Noah Snyder @NoahJSnyder]

@fatdaddy_2 @thebigjohnnyd I go with 2013 when they killed RSS.

[Noah Snyder @NoahJSnyder]

@MNT_Condition @JoshNorris Can you remind me when he played there for us? I'm having trouble recalling. Or are you just saying he's played there for club?

[Noah Snyder @NoahJSnyder]

@NilsHeadley @seanjwestwood Nope, but the work that was done by these people will get passed down!

[Noah Snyder @NoahJSnyder]

@DeivonDrago @seanjwestwood Right, there’s two kinds of admins: the ones who help you get work done, and the ones who make more work for everyone else. If you get rid of the former and not the latter that just means you’re wasting money paying professor salaries for those professors to do administrivia.

[Noah Snyder @NoahJSnyder]

@EllynBriggs Australia’s soccer team is nicknamed the Socceroos! Because football there means Aussie Rules.

[Noah Snyder @NoahJSnyder]

@EllynBriggs Yeah, they’re going to be a fun team to adopt. Good enough to win every game, but not so good that you’re jumping on bandwagon.

Never apologize for calling Association Football by its shortening soccer. Every Anglosphere country calls their local football code just football.

[Noah Snyder @NoahJSnyder]

@rapid_rar2 @hakunamakunana @florianederer It matters, but still this is a less significant result than a strong upper bound would have been. I’m actually a little surprised that Erdos allowed a counterexample for his prize.

Still a fantastic result!

[Noah Snyder @NoahJSnyder]

@rapid_rar2 @hakunamakunana @florianederer It’s a weird one because it’s only a “full solution” to Erdos’s framing, better to think of it as the first significant progress than as a full solution. But first significant progress on an old problem is still a big deal! Maybe more a JAMS paper than Annals (compliment).

[Noah Snyder @NoahJSnyder]

@hakunamakunana @rapid_rar2 @florianederer I’m a full prof at a top 40 department that hired last year. As a best result this compares favorably to the vast majority of our junior offers. Of course it would need to be accompanied by a track record of other (less exciting but serious) work and strong letters.

[Noah Snyder @NoahJSnyder]

@rapid_rar2 @florianederer Note that the Guth-Katz result is much stronger than the OpenAI result, which is a (great!) first step. The real goal with all of these kinds of problems isn't if Erdos guessed right, it's to get the asymptotics. And upper bounds are harder and more prestigious than lower bounds.

[Noah Snyder @NoahJSnyder]

@rapid_rar2 @florianederer "The unit distance problem was the only problem from discrete geometry I included; the distinct distance problem was also a strong contender, but since this has been (almost) completely solved by Guth and Katz, the unit distance problem was a better example of a still open..."