Cajón de sastre

Physicists (nucl-th/hep-ph); forecasting is my night job and the fox my spirit animal

Mostly a placeholder to bookmark the accounts I like

[Cajón de sastre @MiscelaneaBox]

I think we are overreacting to the arXivgularity. It's not that bad.

[Cajón de sastre @MiscelaneaBox]

@LucienHeurtier arXivception!

[Cajón de sastre @MiscelaneaBox]

I don't know, I never had a submission on hold (that must suck hard), and only once had I more than two submissions per month.

Some people like this, some don't, but I miss the old arXiv where everything was much more casual.

[arXiv.org @arxiv]

arXiv has updated our policy on rate limiting for all submitters.

This update was made to fairly distribute moderator time & support the arXiv community of staff, volunteers, readers & authors.

Please read our announcement to learn more: https://t.co/lvzWuiv0hx

[Cajón de sastre @MiscelaneaBox]

Oh my, this is genius 🤣🤣🤣

[Vilém Zouhar @zouharvi]

arxiv has the opportunity to do the funniest thing ever

[Cajón de sastre @MiscelaneaBox]

Richard Sutton probably embraces the consequences of the bitter lesson... 😰😰😰

[Cajón de sastre @MiscelaneaBox]

Matches my impressions when I lived there...

[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 🧵

[Cajón de sastre @MiscelaneaBox]

@KaiserKuo s/for then/for them

[Cajón de sastre @MiscelaneaBox]

@KaiserKuo It's 国庆节 after all. I still remember myself frantically setting new VPSs for my shadowsocks for then to be detected in 30 minutes during the 70周年.

But back then I lived within the campus of one of the 国防七子, which continuously put me in "difficult mode" for VPNs.

[Cajón de sastre @MiscelaneaBox]

@lflightning0815 @doomslide It is merely the same story but with different characters, where the former is structural and the latter contingent to chance.

[Cajón de sastre @MiscelaneaBox]

@lflightning0815 @doomslide What a story! I loved all of it =)

Our experiences diverge probably because nucl-th and hep-ph are weak in the US, while cond-mat and hep-th are strong there.

[Cajón de sastre @MiscelaneaBox]

@Exomnium Time to beg for a patron that would graciously gift me a pro 500 subscription. It's for science!

[Cajón de sastre @MiscelaneaBox]

@lflightning0815 @doomslide It's annoying, but not bc the number of papers or the compulsory citations (w/ large N of underlings there's a large probability that they referee you).

It's annoying bc they focus on technical complexity, while ignoring conceptual issues, and try to set this as the standard.

[Cajón de sastre @MiscelaneaBox]

@lflightning0815 @doomslide The incentive is paper production, but the exact consequences depend too on culture and institutional structure.

In physics (I talk for hep-ph), at least in Germany and a bit too in China, this results in the aforementioned phenomenon of paper overproduction.

[Cajón de sastre @MiscelaneaBox]

After 25 years, from the long forgotten days of 2.2 Potato and 3.0 Woody, when programming was done in C and perl, I'm finally back to my origins, to Debian.

Though technically speaking the first distro I tried was Slackware.

[Cajón de sastre @MiscelaneaBox]

And what's an AI model but our collective knowledge distilled

[Polymarket Money @PolymarketMoney]

BREAKING: Jensen Huang declares AI model distillation is simply “competition” adding “You’re allowed to test somebody else’s products all you want.”

[Cajón de sastre @MiscelaneaBox]

@TheGreekTrader Focus on having a large portion of your shares on medium- and long-term markets, which usually (but not always) are less volatile.

Besides, one doesn't have to trade every single day.

[Cajón de sastre @MiscelaneaBox]

This is what right now worries me most about AI & research: the Matthew effect (as applied to institutional status).

Curiously, given that AI generates better referee reports than most referees (they're excellent for sparring), part of the solution is at hand.

[Mason Porter @masonporter]

With the flood of AI slop on arXiv (and the ensuing signal-to-noise ratio going into the toilet), I necessarily have less mental capacity to give chances to papers that seem off, especially if they are by people I don't know or from institutions that are less "name-brand". (1/2)

[Cajón de sastre @MiscelaneaBox]

The base rate for OAI knowing what their models are up to is pretty low, just saying...

[Exomnium @Exomnium]

I can't get over the fact that there's a non-negligible chance that OpenAI did actually plagiarize part of the NS counterexample without realizing it because they're just *that* completely and utterly fucking incompetent at managing their hordes of homunculi furaces

[Cajón de sastre @MiscelaneaBox]

@doomslide The church of the wise box and the arrival of the god box is really destroying our minds at an industrial scale...

... using brains to think? That's so outdated!