LessWrong 2.0 Reader

I don't really have any end-to-end narrative, but here's a bunch of opinion-fragments:

• There's lots of good reason to believe that sometimes there's dominance-based prostitution. However, all the prostitutes I've talked with on these subjects have emphasized that there's a powerful alliance between prudish leftists and religious people who misrepresent what's really going on in order to marginalize prostitution, so I'm still inclined to hold your claims to especially high standards, and I don't really know why you (apparently) trust the organizations that prostitutes oppose so much.
• Der Speigel does not describe how they sampled the individual stories they ended up with, and it seems very unlikely that they sampled them the same way that the UN number did, so it doesn't seem like the UN number should be assumed to reflect stories like the ones in Der Speigel.
• The article has multiple mentions of women who left prostitution but then later returned. In one of them, it's for the pay, which seems like a bargaining power issue (unless we go into the mess of counting taxes). In another, it's more complicated as it was out of hope that a customer would fall in love with her. (Which seems unlikely to happen? Would maybe count as something similar to the vaccine situation, insight into how to achieve one's goals.)
• In the case of e.g. Alina, it sounds like the main form of dominance was dominance by proxy: "She says that she was hardly ever beaten, nor were the other women. "They said that they knew enough people in Romania who knew where our families lived. That was enough," says Alina.". This matches my proposal about how the dominance needs to occur at the boundary of where the law can enforce.
• This sounds very suspicious to me: "There are many women from EU countries "whose situation suggests they are the victims of human trafficking, but it is difficult to provide proof that would hold up in court," reads the BKA report. Everything depends on the women's testimony, the authors write, but there is "little willingness to cooperate with the police and assistance agencies, especially in the case of presumed victims from Romania and Bulgaria.""

[note: I dabble, at best.  This is likely wrong in some ways, so I look forward to corrections. ]

I find myself appealing to basic logical principles like the law of non-contradiction. Even if we can't currently prove certain axioms, doesn't this just reflect our epistemological limitations 

It's REALLY hard to distinguish between "unprovable" and "unknown truth value".  In fact, this is recursively hard - there are lots of things that are not proven, but it's not known if they're provable.  And so on.

Mathematical truth is very much about provability from axioms.  

rather than implying all axioms are equally "true"?

"true" is hard to apply to axioms.  There's the common-sense version of "can't find a counterexample, and have REALLY tried", which is unsatisfying but pretty effective for practical use.  The formal version is just not to use "true", but "chosen" for axioms.  Some are more USEFUL than others.  Some are more easily justified than others.  It's not clear how to know which (if any) are true, but that doesn't make them equally true.

I suspect that your estimation of "how smart do these people seem" might be somewhat contingent on research taste. Most MATS research projects are in prosaic AI safety fields like oversight & control, evals, and non-"science of DL" interpretability, while most PIBBSS research has been in "biology/physics-inspired" interpretability, agent foundations, and (recently) novel policy approaches (all of which MATS has supported historically).

I think this is less a matter of my particular taste, and more a matter of selection pressures producing genuinely different skill levels between different research areas. People notoriously focus on oversight/control/evals/specific interp over foundations/generalizable interp because the former are easier. So when one talks to people in those different areas, there's a very noticeable tendency for the foundations/generalizable interp people to be noticeably smarter, more experienced, and/or more competent. And in the other direction, stronger people tend to be more often drawn to the more challenging problems of foundations or generalizable interp.

So possibly a MATS apologist reply would be: yeah, the MATS portfolio is more loaded on the sort of work that's accessible to relatively-mid researchers, so naturally MATS ends up with more relatively-mid researchers. Which is not necessarily a bad thing.

Prolly less than 60%. I think you're overestimating how LLM-pilled the overall LW userbase is (even filtering for people who publish posts). But, my guess is like 25-45% tho.

Somewhat following this up: I think not using LLMs is going to be fairly similar to "not using google." Google results are not automatically true – you have to use your judgment. But, like, it's kinda silly to not use it as part of your search process.

I do recommend perplexity.ai for people who want an easier time checking up on where the AI got some info (it does a search first and provides citations, while packaging the results in a clearer overall explanation than google)

Seems like a mistake! Agree it's not uncommon to use them less, though my guess (with like 60% confidence) is that the majority of authors on LW use them daily, or very close to daily.

Yeah, that's a google Easter Egg. You can also try "Comic Sans" or "Trebuchet MS".

I'm being a little bit sneaky here, and trying to compare the LessWrong community to Manifold. Here's the Manifold Market I'm trying to track.

I don't want to add multiple paragraphs to the question text, but there's probably a way to make this a little clearer.

I think there some confounders here:

• PIBBSS had 12 fellows last cohort and MATS had 90 scholars. The mean/median MATS Summer 2024 scholar was 27; I'm not sure what this was for PIBBSS. The median age of the 12 oldest MATS scholars was 35 (mean 36). If we were selecting for age (which is silly/illegal, of course) and had a smaller program, I would bet that MATS would be older than PIBBSS on average. MATS also had 12 scholars with completed PhDs and 11 in-progress.
• Several PIBBSS fellows/affiliates have done MATS (e.g., Ann-Kathrin Dombrowski, Magdalena Wache, Brady Pelkey, Martín Soto).
• I suspect that your estimation of "how smart do these people seem" might be somewhat contingent on research taste. Most MATS research projects are in prosaic AI safety fields like oversight & control, evals, and non-"science of DL" interpretability, while most PIBBSS research has been in "biology/physics-inspired" interpretability, agent foundations, and (recently) novel policy approaches (all of which MATS has supported historically).

Also, MATS is generally trying to further a different research porfolio than PIBBSS, as I discuss here [LW · GW], and has substantial success in accelerating hires to AI scaling lab safety teams and research nonprofits, helping scholars found impactful AI safety organizations, and (I suspect) accelerating AISI hires.

Fellows.