Ryan Beck

Imagine how much cooler the New York Jets would be if their logo looked like this

@CharlieBull0ck @NathanpmYoung Interesting, thank you! And yep point noted about the forecasting question terms, not interpreting you to be commenting on that directly!

Yeah, the real answer is "it's complicated" but I tried to simplify it as best I could. So, first of all, there's a question about whether DoW has actually *taken* the covered procurement actions discussed in that notice. DoW seems (as far as I can tell) to have given the notice of "hey we're going to carry out these covered procurement actions" but then never actually followed up by actually issuing a FASCSA order targeting Anthropic. Per FAR 52.204-30, ordinarily a FASCSA order would go up on SAM.gov identifying Anthropic as a SCR; AFAIK that has not happened. I think this uncertainty is why the D.C. Circuit's stay denial order asks for clarification about whether a covered procurement action has actually taken place. It seems unlikely to me that just saying "we are going to carry out the following covered procurement actions" counts as actually carrying those actions out, although who knows. Then, separately, there's the six month phase-out period. My understanding is that DoW is still using Anthropic's models as of now (and will continue to do so while beginning to phase them out and replace them over time), so Anthropic does still have active DoW contracts and an ongoing contractual relationship with DoW. I think that practically speaking, the 4713 designation is likely to exclude Anthropic from most *new* DoW covered procurement contracts while the 4713 designation is active (again, some uncertainty here). But it doesn't exclude them from contracts with other agencies, and it doesn't exclude them from non-covered-procurement contracts with DoW either. So that's what I meant by that tweet. I should clarify that I'm not commenting on how any particular prediction market outcome should be adjudicated. I haven't read the resolution criteria or looked at the market at all. I'm just answering the question that was asked to me.

Can a lawyer please answer my Anthropic question. Is Anthropic being banned from US Government contract under the mechanisms of §3252 or §4713? Not because of Hegseth or the President's directive or political chilling effects, directly via those statutes? cc @ryanbeck111

@CharlieBull0ck @NathanpmYoung Why do you say they're not prohibited from contracting with "~any" agencies? Hegseth's notice of supply chain risk designation explicitly says it prohibits all DoW procurements.

@NathanpmYoung @ryanbeck111 Short answer is “no, Anthropic is not currently banned from contracting with the U.S. government by either of those statutes.” The 4713 supply chain risk designation is still in effect, but doesn’t prevent Anthropic from contracting with USG or ~any agencies.

@deanwball In pop parenting convention these are "developmental leaps", which the formalized version of I guess isn't supported by evidence. But yeah 100% saw the same with my kids. "Why is she so fussy lately?" followed by back to normal + sudden new skill. fatherly.com/parenting/the-…

I am convinced that babies undergo something that resembles grokking in neural networks whereby, almost suddenly, they gain mastery over some movement or action that they once could only vaguely perform. My wife calls it a “software update.”

@finnhambly The headline questions are the employment ones, overall and by occupation, but we see these as painting a picture about how labor might change over the next decade. They'll be shown on a custom hub page by topic along with brief summaries of what's driving, for public benefit.

@ryanbeck111 Hi Ryan, sorry I missed this! Do you see them more helpful as a collective then? Or are there some standout questions that are obviously helpful/informative? I'd like to get involved but find it hard to get motivated when the value for the recipient is unclear

New Metaculus tournament, but there are 93 questions to fill out I doubt the funders care about all 93 questions, but need them so that participants don't win due to overconfidence and luck And if you're a specialist on a couple of these Qs, there's ~no point participating

@waim What do you put in there, is it just like a list of pre-approved commands?

@ryanbeck111 I asked it to write to cmd.sh and run that for all bash commands and that’s been working of for me

It would be reckless to dangerously skip permissions in Claude Code, that's why I make it ask me first and then rapidly click "always allow" on everything without reading it first.

How will AI reshape the labor market? We just launched the Labor Automation Tournament to forecast how automation will affect jobs, wages, and the workforce through 2035, with $35,000 in prizes for predictions and analysis. More info below!

By overwhelming margins, top economists agree the Jones Act's economic costs outweigh any possible national security benefits (and at least one "disagree" clearly misunderstood the question): kentclarkcenter.org/surveys/the-jo…

While I generally agree with @mattyglesias' Slow Boring piece today about pundits embracing a self-awareness of being out-of-touch from 'typical' Americans, I want to challenge this particular assertion -- it's not quite accurate.

This is so cool, hearing straight from the guy behind some of the most famous CGI shots in history how they approached it and pushed the technology forward from movie to movie. youtu.be/-wpXOIPl0wo?si…

@JgaltTweets Not sure, he does account for interest rates but I wonder if @besttrousers has assessed housing prices as a cause.

@ryanbeck111 Housing costs not mentioned? I would've thought that's the main thing

Good stuff in here

And a potential alternate explanation: x.com/i/status/20339…

Why hasn't Newcomb's problem been adjusted for inflation. 1969 was a long time ago. 9x your numbers people.

@PlastiqSoldier @jakehalloran1 I'm curious why you think that's the case? My understanding is Claude hasn't beat Pokemon yet, which is fully static, there are no real-time events that require precise timing, while Montezuma's Revenge requires timing so I would guess it's even more challenging.

@jakehalloran1 Beating this game is part of Metaculus' definition of weak AGI. It really seems like no one has bothered to do it, but easily could.

Can current Claude beat the classic Atari game "Montezuma's Revenge"?

@Noahpinion Our criteria for the question has been essentially the same since it was launched in early 2020 x.com/StefanFSchuber…

This is because we keep changing our definition of "AGI". The truth is that the AI we have right now is already AGI.

@AlwaysUhhJustin @Noahpinion @Polymarket @metaculus @Kalshi The preview is cached, it's at 97%

@ryanbeck111 @Noahpinion @Polymarket @metaculus @Kalshi Different from regime change. And should be 99% now. He's dead.

Holy shit

@AlwaysUhhJustin @Noahpinion @Polymarket @metaculus @Kalshi Note that this one requires total regime change and eliminating Islamic Republic from the constitution. metaculus.com/questions/3877…

@Noahpinion Yeah +5-10 top leaders and 40 total. I'm trying to find a prediction market for regime change. Can't find one yet. @Polymarket @metaculus @Kalshi please post!