astrolabia
5mo ago
My favorite example of this is complaints about the "cotton ceiling"[https://old.reddit.com/r/asktransgender/comments/6a3e3a/who_else_here_is_put_off_by_the_idea_of_the/]("cotton ceiling"), which to me paints a hilarious and sad picture of aspergers guys becoming trans as a gambit of rules-lawyering lesbians into sleeping with them.
astrolabia
5mo ago
Language evolves. Definitional battles are not worth it.
The woke definition has big upsides for trans people for little costs
It sounds like you think this definitional battle is, in fact, worth it?
I'm encouraged that you acknowledged that there are costs - can you elaborate? I think Zac would claim that one serious downstream cost is autogynophiles being encouraged to castrate themselves. To me that is the main problem - confused and unhappy people being encouraged to mess up their bodies unrecoverably. I think that frank acknowledgement of the senses in which, due to the limitations of medical technology, trans people aren't actually their desired gender, would lessen this problem. So I do think that this is a definitional battle worth fighting (as do pro-trans advocates).
astrolabia
5mo ago
I agree, but I was trying to be maximally charitable in case all that Folamh3 knew about Hanson was that he was arguing against rape being worse than something else. That's why I was asking if he read the post.
astrolabia
5mo ago
Okay, but the framing as "sus" makes it sound like a hidden, rare opinion. What do you think of his claims?
astrolabia
5mo ago
I was thinking of https://slatestarcodex.com/2014/11/21/the-categories-were-made-for-man-not-man-for-the-categories/
but that's fair, it might just have been motivated cognition. But given that Scott has independently reached unpopular contrarian opinions on his own so many times, and doesn't address the downsides of gender-defined-by-fiat head on, it's almost the same phenomenon as dishonesty imo.
astrolabia
5mo ago
I too would love it if rationalists were forced to bite the bullet and say something like "yes, racism (in some senses) is rational". However, I'd say that most of them are simply deliberately silent on these issues because they know that dissenting would wipe out their credibility and force them to become a full-time advocate on an issue that they don't particularly care about. For example, James Damore.
I too find it incredibly sad when the ones that do write about sensitive topics toe the line dishonestly, e.g. like Nick Bostrom did on race in his apology, and Eliezer and Scott Siskind on trans issues. I commend Zack M. Davis for calling them out on this and being brutally honest, but he has a horse in this race.
Also, what did Razib Khan and Stephen Hsu do wrong? They put their jobs on the line to talk about the truth. They didn't go so far as to explicitly say that racism (in some senses) is fine, but they pull their punches less than anyone who hasn't been banned entirely.
astrolabia
5mo ago
Did you read the linked post? He's making the claim that it's a bigger harm in terms of genetic interests.
Can you expand on what you mean by "majorly sus"? Is the idea that the fact that he'd raise a hypothesis that could also be used to argue for taking crimes against women relatively less seriously, that's evidence that he's misoginystic?
I suppose that's a reasonable inference, but I also think he does raise a good question and point to a genuine mystery. More generally, if academics can't raise wrong-sounding ideas without being cancelled, then there's not much point in having them or listening to them. So I guess I implore you to ask if there is any venue or method by which someone could discuss disgusting-sounding ideas that would lead you to actually try evaluating their claims.
astrolabia
5mo ago
I agree the benchmark is a real contribution. My claim is that there are similarly novel-reasoning-heavy benchmarks that also look more like useful tasks. such as https://klu.ai/glossary/gpqa-eval
astrolabia
5mo ago
I agree with this take. One interesting open question is: what fraction of the wealth created will the big labs be able to capture? Right now the differences between the big LLMs are relatively small, so it seems plausible to me that the LLM provider becomes a low-margin coke vs pepsi kind of market. However there could easily be some winner-takes-all dynamics or services bundling dynamics that might favor e.g. Google since they might be able to integrate across services better. Am curious for your thoughts.
astrolabia
5mo ago
Francois Chollet is a disingenuous midwit who makes himself sound smart by arguing against strawmen. He reminds me of Stephen Jay Gould. In general his claims are a mix of: things that everyone already agrees with (like that LLMs are currently missing some sorts of long-chain reasoning abilities) or things that are a mix of simply wrong and meaningless word games (like that LLMs are merely memorizing).
To his credit, he released a new benchmark, but imo it's more poorly-motivated than lots of other benchmarks measuring performance on similarly reasoning-y but real tasks.
astrolabia
5mo ago
I agree, but I'm saying that it also managed to not even help ADOS. And that a future "Human Lives Matter" movement would be similarly ineffectual at its ostensible aims. And that our future votes would similarly buy us a wretched welfare bureaucracy at best.
astrolabia
5mo ago
Good point. I agree they have de jure privileged status, and that this leads to some substantial concrete benefits. But I'd counter that, to the extent that competition operates, it naturally disenfranchises them from almost all angles and makes their own self-advocacy ineffective. E.g. affirmative action spots mostly go to non-ADOS, BLM was a corrupt waste of political energy (compared to e.g. the at least-somewhat effective ADL), high-income people move away leaving places like Detroit.
I'm saying that a human worker advocacy org, in a world of more sophisticated machine actors, would similarly end up being a useless skinsuit pretending to advocate on behalf of humans, but wouldn't be able to avoid the important sources of influence all naturally routing around human agency. E.g. if you were a rich human, would you use an inferior (by supposition) human-run investment firm to manage your assets, or a machine-based one?
So my ultimate claim is that we have to choose between freedom and competition versus the continued relevance of humans. I really don't want to have to choose between these things, but I think we do.
astrolabia
5mo ago
I agree, I think "a useless mouth and his clout are soon parted". To the extent that competition still operates at any scale, I worry that anti-automators will be eventually marginalized, no matter how many kids we have.
astrolabia
5mo ago
I know what you're talking about, but I'm confused about Norman vs Nordic phenotypes. Aren't the Normans also Nordic? I'm also not clear what it means when I see very thin faces with fine features, which showed up a lot in old royal portraits.
astrolabia
7mo ago
One of the scariest things from my point of view is watching some Jewish progressives I know choosing, after a period of internal struggle, to take the side of Hamas. I could see that something had to give when they started being attacked by what they viewed as their own side. And I would have been surprised to see them abandon pretty much their whole progressive social networks and worldview under any circumstances, even to defend themselves. But it seems like many of them chose to thread the needle by simply becoming "one of the good ones".
astrolabia
7mo ago
looking for self-actualization that will never arrive
Seems like it does arrive for them, in the sense that they get to be a part of a change in zeitgeist. I imagine it feels pretty fulfilling to get in early on the next big political thing.
astrolabia
7mo ago
I don't think it's that they're stupid, it's that talking concretely about reforms invites infighting. Everyone in those orgs can agree racism is bad / the problem, but they've learned that talking about specific mitigations is a recipe for drama.
astrolabia
7mo ago
I agree it was a spineless, worst-of-both-worlds kind of apology. It was clearly disingenuous for a polymath like him to claim that he's not qualified to even evaluate the evidence about HBD. It was worse than normal straight-shooting autism.
That said, he's a great thinker when he's in his element.
astrolabia
7mo ago
Yes, and now there's an incentive to not tell your associates that you're worried about your kid, for the same reason.
astrolabia
7mo ago
knowing he was crazy
Seems like you're vindicating Maiq by holding this against them. If they had never seeked professional help, they could have argued that they didn't know he was crazy.
astrolabia
8mo ago
That makes sense, but I saw it as a teen and didn't think that the good-looking actors were any indication of satire. After all, Melrose Place, and pretty much all Hollywood movies, have ridiculously good-looking leads without being satires. And it's played straight that the bugs are, in fact, in total war with humans.
astrolabia
8mo ago
I totally agree. For example, it's bonkers (but not surprising) to me that the recent misleading residential school graves reporting (and subsequent exaggeration and demonization of the Catholic Church on social media) that led to scores of churches being burned down didn't even pattern-match as "hate speech" in the eyes of anyone in the establishment.
astrolabia
8mo ago
Just never be an undesirable in the eyes of those with power.
I spent a while asking my well-educated peers why they were supporting censorship and politicization while also claiming to loathe our local conservative government, warning that the shoe could easily be on the other foot someday. My leading hypothesis is that they (unconsciously) are following a strategy of "endorse whatever social movement is clearly ascendant" in a way that's almost completely blind to the content of the movement, even being blind to self-interest to a large extent. So they legitimately don't expect to suffer from censorship in that setting - they'll simply switch sides, and maybe push for an exception to the new ideology for themselves. It makes sense, and seems close to the optimal strategy if you can't actually control social movements.
astrolabia
8mo ago
Sure, but that would still lead to runaway censoriousness, wouldn't it?
Oh, I thought it was to avoid reminding women who are breastfeeding but identify as male that they do have breasts. Which always seemed like it must be a rare request. Like how many women feeding their baby using their breasts, who certainly can feel what's happening, still get psychic psychic relief from not acknowledging that it's a breast?
More options
Context Copy link