I think you should count this as a failure of your week’s goal. I understand that it is a 502-word blog post and therefore wouldn’t get you kicked out of Inkhaven, but the previous day’s installment said “challenge to publish a daily essay”, and a quote collection is not an essay.
Zack_M_Davis
Contra Pace on When to Apologize
This piece is at its best when providing concrete detail about the nonobvious labor required to put on a conference: the shrub-tree breaking under the groundkeeper’s weight, the negotiations with the furniture and badge suppliers. At its worst, it devolves into empty ingroup loyalty-signaling that’s hard to excuse as an artifact of being written in a day: it’s easy to credit that rough prose would have been polished given more time to work on it, but a deficit of self-awareness seems unlikely to be remedied by mere time.
The attribution of the alleged “dirty, poor, graffitied, smelly, closed-down, and ramshackle” quality of the area in south Berkeley to “Moloch” betrays a marked indifference to both the allusion’s source material and the facts.
Regarding the former, “Meditations on Moloch” isn’t about everything bad in the world being the fault of “Moloch”. It’s specifically about coordination failures: things being contrary to the preferences of the relevant actors due to a lack of coordination away from a bad Nash equilibrium. I don’t doubt that the problems of urban decay could be analyzed in terms of coordination failures, but there’s nothing in the text that hints at any such analysis. As far as the reader can tell, the content of the allusion amounts to, “Moloch = bad stuff, Lightcone/Lighthaven = good for resisting bad stuff.” Indeed, the distinguishing feature of coordination problems is that they can’t be solved by a lone actor being unusually good! The Moloch allusion would make sense if Lightcone Infrastructure had been active in local government to find collective solutions to urban decay that couldn’t be pursued unilaterally, but no such effort is mentioned in the text or exists to my knowledge.
Regarding the latter, the evidence for urban decay in this south Berkeley neighborhood is questionable. I guess I can buy “graffitied” insofar as I do remember seeing a graffito on the wall of the building on the corner of Stuart and Telegraph. But “poor”? That neighborhood is more expensive than 97.2% of the U.S.! “Closed-down” would initially seem to be supported by, e.g., the former home of the Pacific Center for Human Growth near Telegraph and Derby appearing to be abandoned—but that’s because developers are planning to put a 50-unit appartment building there.
The post descends into self-parody in the entry about conference attendees working for frontier AI companies despite having read ingroup authors. If you understand concepts like “adversarial debate” or “being well-read”, it should not be surprising that someone might have read your group’s literature and yet disagree with you about the ethics of working in one of the most productive sectors of the economy! This is sadly consistent with my read from previous experience that the author and his employer don’t really believe in debate and don’t really believe in being well-read. Robust, informed disagreement is defection against the ingroup.
How internally coordinated is the Trump administration? The fight with DoW leadership didn’t even stop the NSA (which sits under DoW) from using Mythos, and the new export control is coming from Commerce, not War. I assume Pete Hegseth can’t just call up Howard Lutinik and say, “Hey, can you slap an export restriction on Anthropic?” Maybe it doesn’t matter that much. (A shared narrative within the administration that Anthropic is Bad might be all the coordination needed for this action, but that theory could make different predictions about future actions.)
I think “transformative AI could be slightly nice” arguments aren’t logically dependent on LLMs-as-AGI per se, even if belief in the two are correlated: [1] Christiano’s formulation (very roughly, that it’s not obvious why the evolutionary quirks leading to humans not being maximally ruthless couldn’t have ML analogues) doesn’t seem to depend on levels higher than “(D) systems centrally involving deep learning” in your plateau-ism taxonomy.
- ↩︎
Where delusional optimism would be an obvious candidate for the source of the correlation.
- ↩︎
It’s really impressive! I’m pretraining-famous enough that models as early as Opus 3 already had the basic idea of me (with hallucinated details), but I’m blown away by how much detail Fable 5 has correctly memorized. It names twenty-four of my posts by exact title, and correctly notes the year for most of them! It has a (correct) “hazy memory” of my Rust compiler contributions. It knows about my criticism of dath ilan’s secrecy obsession—I didn’t even publish that post yet! (I had merely mentioned it on Twitter.) It’s just wild.
re: credibility of claims to believe in local social norms about telling the truth about something
I believe in such a norm, and am trying to practice it here.
I don’t believe you. I think you have a self-deceptive belief in believing-in such a norm, but your behavior is not consistent with sincere belief in the stated norm.
I’ve been acting in this conversation as though there’s a cost in person X’s attention to saying loudly “person X did bad thing Y,” and also as though there’s a cost to making it such that moderators expect huge amounts of such attention-costs if they take any moderator action. [...] It seems worth-it to me to cause those costs sometimes [...] I do less of it than I would if it were cost-less.
I think you would see the problem here if we were talking about any other subject. Forget about awkward local social matters. Think about a sequence of random experiments with two possible outcomes—coinflips.
What would it mean for someone to believe in a norm about telling the truth about the sequence of coinflips?
I think it would mean: carefully recording the results of any coinflips they see, no matter whether they were Heads or Tails, and socially punishing people who lie about coinflip results or who selectively report conflip results without being clear about the selection criteria.
The social punishment is necessary because if there are no consequences for breaking a norm, then there is no norm, so people who aren’t willing to punish norm violations don’t believe in the norm. The clause about selective reporting is necessary because we don’t want people to be able to disproportionately ignore Tails outcomes in order to make the coins look Heads-biased and claim that they weren’t “lying” because all of the flips they reported actually happened.
Suppose the coin room supervisor Olivia says, “I would appreciate some courtesy to keep reporting of coinflips focused on Heads outcomes, because recording Tails outcomes increases the costs of coin room supervision.” Ian thinks this is a reasonable request and says he was already trying to follow it, without remembering Olivia had made it explicitly: it seems worth-it to him to incur the costs of reporting Tails outcomes sometimes, but he does less of it than he would if it were costless.
In this scenario, I think it’s clear that Ian does not believe in a strong local norm about telling the truth about the sequence of coinflips. He just doesn’t! Someone who wanted to know the truth about the sequence of coinflips would want to know about the Tails results, and they won’t get that by taking Ian’s reports at face value. If Ian claims to believe in a norm about telling the truth about the sequence of coinflips, he is lying or self-deceiving and it makes sense to tell him, “You can’t possibly expect me to be that gullible.” Right?
re: scope of purported mandates
Many tasks work better when a single person or team is in charge of them for a decent chunk of time, and isn’t “micromanaged,” and acts on their all things considered best guess about what’s good (rather than being tasked with doing what their manager would want, say). I believe “make lesswrong.com good” is such a task.
I buy the anti-micromanagement argument for tasks like “what database schema should the website use” or “how should the hotel be decorated.” You don’t want arbitrary other people in the vicinity of the organization to be able to second-guess the subject-matter expert making the decision when there’s no particular reason why the subject-matter expert might be serving their own goals contrary to those of the organization, and there’s no particular reason to think that the arbitrary other people would make a better decision.
I don’t buy the anti-micromanagement argument for deciding to purge a long-standing community member from a space that’s ostensibly being managed for the community’s benefit, especially a public forum. (Hotels are supposed to be a coherent service and it makes sense for them to have a CEO who hires and fires employees to make the hotel good. The clash of ideas in public isn’t supposed to be coherent and there isn’t supposed to be a CEO.)
Unlike the case of the hotel decor (where we expect the interior designer’s taste to be naturally aligned with what’s good for the hotel), it’s easy to imagine how the power to purge anyone who questions you might be misused. (I have some relevant illustrative evidence that I don’t think fits in this public comment; I’ll send you an email.)
c) the website continuing, with Habryka and co still technically in this role, but with the user-base mostly thinking of them as random forces rather than as [stewards of a cool project that it’s worth them lending some believing-in to].
Why isn’t this just the obvious common-sense default? I care about the practice of human rationality, but it’s not a religion with authority figures!
It makes sense to have a website for students of human rationality to talk to each other. It makes sense for the people who take donations to run the website to use their best judgement on what database schema to use, and to take care of censoring the slushpile of LLM slop that no credible rationalist would miss.
But when the people who run the website start deciding to use their ownership of the infrastructure to prevent my friends and collaborators from commenting on my blog posts in the same place where everyone else does, of course I’m going to regard that as random forces getting in the way of my interests, rather than something I should defer to! If I can’t find a better forum, I’m still willing to chip in to pay the server costs for the service that I use (much as I pay for Twitter), but no, of course I’m not going to lend any more believing-in than that to these people!
re: object-level suggestions for how a reasonable person allegedly might find user-level bans insufficient
I think the question “could a reasonable person find user-level bans insufficient for non-evil reasons?” is fairly central to our dispute [...] Does that seem right to you, Zack?
Yes.
I suspect a sizeable chunk of users have a goal like “don’t incur needless reputational damage for via failing to respond to confusing-to-others claims that I made errors I didn’t make (especially if the claim is loud, reads as confident and Sequences-fluent, calls me out by name, etc)”
As an author, I deny that that’s a legitimate goal. I write things on the internet. Sometimes other people criticize my writing. I can’t prevent third parties from making Bayesian updates about me based on my response or lack of response to the criticism. An example of such a Bayesian update is, if I don’t reply, maybe some the third parties think, “Gee, Zack is probably really busy.” Another possible example of such an update might be, if I don’t reply, maybe some of the third parties think, “Gee, Zack probably doesn’t have an answer to this one.” Which one pertains depends on many details to be resolved in each of the third parties’ individual judgements.
You seem to be telling me a “reasonable person” thinks I should be able to censor my critics in order to prevent third parties from thinking, “Zack probably doesn’t have an answer to this one.” As an author, I categorically deny that this is reasonable. I think that if reality puts me in a situation where someone is inclined to think “Gee, Zack probably doesn’t have an answer to this one”, the honorable and sane responses available to me are (a) let them think that about me, or (b) answer it.
It’s true that (a) is a “cost” to my social goal of having everyone think well of me, but as a citizen in a free Society, I understand that the rest of Society does not have to reorganize itself to maximize my personal social goals. To think otherwise would be totalitarian and childish.
re: comments on “Zetetic Explanation” and mechanisms for downvoting and user-bans not being sufficient
I’d like to note that it would appear hypocritical for Hoffman to object to Achmiz’s infamous “Hmm” comment on “Zetetic Explanation” because, as I point out in footnote 10, Hoffman has used substantively the same rhetorical device in correspondence with me.
One of the things Hoffman seemed particularly annoyed by was Achmiz’s failure to provide feedback on Ben Pace’s Ideological Turing Test attempt. Seems fine as grounds for a user ban, I guess: if you want a cultivated space where people know how to play the “ITT” and “interpretive labor” games, then I suppose it follows that you’d want to exclude people like Achmiz who are skeptical of the value of interpretive labor.
I’m curious for your guess at the mechanism why “downvote and ignore” is insufficient
Seems straightforward: downvote and ignore requires someone to do the downvoting, and it might generate discussion that the author doesn’t want on the page at all.
and I’m curious whether the same mechanism indicates that user-level bans are also plausibly insufficient.
The issue is property rights. Authors with their own websites have control over what appears on their sites, so it makes sense that LessWrong 2.0 wouldn’t want to offer much less control, so as not to disincentivize people from cross-posting to LessWrong 2.0.
But control over your own posts is all you’d ever get on your own site. You don’t get to censor other people’s website, and there’s no reason you should get to censor other people’s Less Wrong posts.
It’s an unreasonable ask because it differentially advantages phonies over people who care about whether what they say is true or not. People who can support their claims, do so. They don’t need to whine that their interlocutors aren’t being curious enough.
Goofusia says, “If
is complex-differentiable at a point , then is complex-differentiable in an open neighborhood of .”Gallantina says, “False! Consider
, which is differentiable only at the origin. Reply!”Goofusia says, “No, I won’t reply, because you consistently create the kind of conversations I don’t want to have on my posts. I would invite your criticism and critique, but you deliver it in a way that is difficult to engage with because of your refusal to be actively curious about the predictions, intuitions, life experiences, &c. that are motivating my statements. That is not an unreasonable or extremely costly ask or anything.”
As Gallantina, what do you say here?
I think your “go down this rabbit hole where you’re not actually trying to talk about” story could potentially fit the improv game framework, actually? (Briefly: improv performers and people who don’t like adversarial debate try to “Yes, and” each other’s bids to steer the scene, rather than “blocking”, denying the other’s bid.)
Say: I write about my research program which you think is fundamentally flawed. I’m interested in hearing about in-paradigm improvements on my ideas, but not interested in hearing that the whole research program is shot. You ask questions that you think I should be able to answer if my research program were good. I have trouble answering the questions, and get frustrated that you weren’t talking about what I wanted you to talk about. (I wanted you to build on my ideas, not try to destroy them! Whether the research program is good is a separate topic, not what I wanted to talk about!)
Maybe there’s a case that you should have more clearly flagged that you were commenting in an “adversarial” role rather than a “cooperative” one with respect to my research program, so that I could have made the decision to ignore you earlier. But maybe I should stop trying to be so controlling of how other people interact with my ideas? (Maybe you could have been won over on the merits of the program if I had successfully answered your questions, and you couldn’t know until you asked.)
talk of “hurt feelings” in this whole context is confused about some of the cases, because the feelings aren’t “you said mean things about me” but more like “I turned out to be frustrated with this thread, and even if I can’t explain exactly what’s happening I still think it was kinda adversarial and a regretted use of my time, which is an annoying situation and I just want to avoid this.”
I think it’s pretty gullible to treat those different self-reports as representing relevantly different things in the territory. My theory is that the “kinda adversarial” thing that’s happening that the person “can’t explain exactly” is: disagreement unadulterated by social improv games—disagreement with someone who refuses to soften, hedge, or obfuscate their vision of reality in order to maintain a cooperative vibe of mutual respect. It’s annoying and people want to avoid it because it’s less socially rewarding than the improv game.
re: alleged local norms of telling the truth even on awkward social matters
I predict that the mod team, if asked, would say they endorse this norm [for telling the truth even on awkward social matters]. That is, I predict that the mod team, if asked, would say something like “yes, Anna (or Bob or whoever), we do not have a request that you avoid public speech that might undermine our narratives about it being good to have banned Said or whatever; we rather have a view that it is prosocial for you to attempt to promote truth and clarity as you see it, including here.” We can check this by asking the mod team if you want.
You can’t possibly expect me to be that gullible. You wouldn’t accept claims about what “strong local norms” hold at OpenAI by asking Sam Altman.
Sure, if I accepted that as the adjudication criterion for the claim and we then asked the mods whether they affirm that we have a strong local norm for telling the truth even on awkward social matters, there’s a pretty good chance they’d say Yes.
But if so, they would be wrong. As I pointed out in §VI.3, Oliver Habryka is already on the record saying that he “would appreciate some courtesy to keep discussion to the principles and decision-level instead of critiques of my personal behavior, as indeed much of the cost of moderation is measured in having any moderation-adjacent action be torn apart and be requested to be justified or defended.”
Asking people to withhold “critiques of [a moderator’s] personal behavior” in their capacity as a moderator is not compatible with a “strong local norm for telling the truth even on awkward social matters”! It just isn’t! You can’t wiggle out of this one by claiming that it’s covered by “LW does not fully hit this aspirational norm.” The request is a denial of the aspiration!
(For completeness, I should note that a footnote disclaims, “though of course in as much as something seems egregious, you and others should feel free to call it out”. I don’t think this changes anything. The only reason to make the request but provide an escape hatch for “egregious” bad behavior is to give a free pass for less-than-”egregious” bad behavior.)
re: separating object-level from authority claims
I don’t think a manager telling an employee “My opinion is X, but I want to emphasize that it’s your call” is analogous to the present situation.
Suppose someone wrote an 18,000 word post with careful quotes and citations accusing CfAR employee Emily of abusing her pizza-purchase responsibilities for personal gain against the organization’s mission: Emily deliberately purchased too much of her own favorite kind of pizza, knowing that the workshop attendees wouldn’t eat that much, so that Emily could keep the leftovers. Would you begin your response with “I believe Emily has, and deserves, the mandate of heaven and I support her authority to decide what pizzas to buy”?
I still think that would be weird! If I were President of CfAR in that scenario, I would not say that, even if I liked and trusted Emily and had no intention of firing or punishing her under any circumstances. (Quietly overriding her on the pizza-ordering task need not be a punishment, if Emily’s pay and status were to remain the same.) I would say something more like, “I trust Emily, and that doesn’t sound like something she would do.” I would then say either, “I expect this to be false upon investigation,” or “This is so implausible that I’m not even going to bother to investigate it.” I would not affirm Emily’s authority independently of the accusation being true!
If my support for Emily were unconditional, it would be dishonest to claim that norms questions about the use of CfAR’s food budget are important. Yes requires the possibility of No: if there are no consequences for breaking norms, then there are no norms. The honest thing to say in that situation would be, “Why are you even telling me this? I don’t care what would be good for the workshop attendees or the organization’s mission; what matters is that Emily gets the pizza she wants.”
re: causes of “unsustainable costs” from demon threads and mod team facilitation time
I think a-c is hard to dispute. Do you dispute it?
If I’m supposed to accept “reasons, born of experience, why they didn’t think this was sufficient for goals they had”, then it’s vacuous! I’m not going to deny the the tautology, “If the mods are always right, then the mods are always right.”
If I’m allowed to doubt such reasons, I deny (b). As I suggested in July 2025, I think an obvious thing to try, that was not tried, would be to actively raise awareness and encourage use of the per-author user ban functionality. We have evidence that there was low-hanging awareness fruit in the form of, e.g., Romeostevensit’s comment on this post that he “didn’t know that was a feature”, and the Surprise react on my pointing out the existence of the feature in the first paragraph of §IV.1 by abstractapplic (who “voice[d] strong approval of the meta-level approaches on display” in the original ban annoucement).
He is changing (“interfering with”) the LW user-base’s notions of which [posts, and claims in posts] are “in good standing” [...] a reasonable-person mod might disprefer this.
By means of arguing about them! Changing the userbase’s notions of which posts are in good standing by means of arguing with them is what intellectual discourse is all about! Your “reasonable person” who disprefers this is not a “mod”; they’re a wannabe religious authority.
Users might feel some (need/desire/obligation) to respond to top-level posts about their posts, and wish not to have to engage with top-level posts about their posts by Said, and so be discouraged from posting to LW
So it’s not enough to let people censor criticism from the comment sections on own posts (which I’m supporting as a pragmatic solution to help people share the website), it’s not allowed to appear anywhere else on the website, either? Again, this amounts to a religious authority declaring Achmiz a heretic whom the faithful should ostracize. This is intellectually indefensible.
I would guess that for Ben a user-level ban would have been sufficient
So you concede that this is not relevant to my case that a site-wide ban was unjustified given the existence of user bans as a sufficient and less intrusive remedy as articulated in §IV.1.
doesn’t argue in good faith or seem to know what that means
As I’ve pointed out before, your definition of “good faith” is nonstandard (albeit not entirely unmotivated).
virtue of asking questions to get you to say exactly what you mean, I see as the vice of refusing to meaningfully engage with the author to try to understand them
As an author, I expect to be able to answer questions aimed to get me to say exactly what I mean! Don’t you?
I don’t always get around to answering questions, but I would never, ever disparage someone asking real questions about my work as engaging in an “the vice of refusing to meaningfully engage” with me. That just seems ungrateful: in a world of abundant information, the fact that someone is paying attention to my work at all is a precious gift; I’m not in a position to demand that they do it on my terms.
Thanks for commenting! You raise some important points here that I should address.
And also Zack and Said rail against this
I think “rail against” is arguably true of Achmiz but not true of me. (I consider him a friend and I’ve put a lot of effort into defending his interest in using this website, but we’re different people who disagree on many things.)
To clarify my views: while I agree with Achmiz’s argument that the ban feature enables authors to impose a tax on criticism, which is epistemically distortionary, I support the existence of the feature because it facilitates people with different preferred discussion norms being able to share the website. I sincerely believe it should have been possible to share the website. (See footnote 5 and §IV.1 in the post.)
this is strong evidence that the person using it is wrong and fears them/their arguments/isn’t adequate to the task. Other hypotheses (such as “engaging with you is powerfully net negative and often produces no value at high cost”) seem not to be considered at all.
I wouldn’t call it strong evidence, but it’s reasonable for third parties to be suspicious. The problem is that “This commenter is net-negative” is something someone could easily say if they were wrong and inadequate to the task, so an individual author saying it doesn’t produce a large likelihood ratio without additional evidence for why they in particular should be trusted. (I think we’d probably disagree about what would constitute such additional evidence.)
It’s rather disingenuous of Zack to [...] blink innocently and say “well you can just [costlessly] ban people one-on-one? [Please ignore how we will try very hard to make this very costly.]”
It would be bad for me to be disingenuous, so let me try to clarify. I think there’s an inherent conflict over setting the zero point when it comes to assessing social “costs.”
As I said above, I’m in favor of the ban feature existing, because that helps different people share the website, and I strongly believe in sharing the website. It seems to me that the natural “zero point” is that everyone is free to use the ban feature to control their own posts, and everyone is free to criticize use of the ban feature on their own posts. That’s how it works on most websites. (You can block someone on Twitter, but you can’t stop someone from Tweeting that you shouldn’t have blocked them.)
But the zero point that seems natural to me isn’t the only possible one. You could imagine trying to enforce that banning is “zero social cost”: not only does the ban feature exist, but banned users also aren’t allowed to talk on their own shortforms &c. about being banned. (Or possibly even on other websites? If someone Tweets about being banned, the Less Wrong mods could theoretically consider that actionable.)
I’m not in favor of the zero-social-cost version (especially not the off-site version), because the epistemic distortion of that seems much worse than the mere existence of the ban feature: you’d be preventing criticism from appearing on the website (or the world, in the off-site version) at all, rather than just on some pages.
Are you advocating for zero-social-cost version? (I don’t want to put words in your mouth.) If so, that’s our disagreement. If not, I’m not really sure what to make of your claim that I’m being disingenuous. Maybe you thought I was being deceptive by implicitly claiming that the ban feature is or should be socially costless (that no one should think less of you for using it), and it suffices that I’m clarifying here that I don’t think that?
also mostly pointless
I believe in seeking death with dignity.
Honoring the spirit of Said, what evidence do you have for this claim?
Um, subjectively he just seems socially sharper in private emails than public comments. (Honoring the spirit of Said, I should note that my subjective perception of private correspondence is worthless as public evidence, but that’s my answer.) This is the part where his persecutors say, “A-ha, so you admit we were right about the weaponized obtuseness!” And that’s the part where I say, “Okay, but ‘weaponized obtuseness’ is just your pathologization of ordinary politeness, and you people obviously don’t want to hear the rude version.”
Thanks! I like the thing ChatGPT says about, “Zack has already accepted some censorship for SNR, so the live question is threshold-setting.” I tend to want to argue about principles, but maybe saying something about empirical threshold-setting would make some progress. We agree that, in principle, there is such a thing as bad or marginal criticism. (You could totally prompt an LLM to come up with the most trivial and unimportant nitpicks to post en masse, and that would definitely merit a ban.)
But in this case, look, as I mentioned in §II.3, Said has been compiling a collection of his best comments. I think that if you read it, it’s just very obviously good stuff by the usual standards that rationalists use to evaluate our stuff. I think as a 9x Curated, 4x Best of Less Wrong, 3x Less Online invited author, my judgement of that means something, and former MIRI employee Jessica Taylor agrees (“Said is far above average at” “specific criticisms, including philosophical criticisms”). And you’re telling me with a straight face that this guy is a broken stair, below the empirical threshold? On the intellectual merits? I just—really have trouble taking that seriously. As I argue in §IV.3, I think this is actually about covering for ingroup members’ feelings and reputations and not about the intellectual merits.
Thanks for commenting! It’s always a pleasure. (I really mean that. What I find most disappointing about interactions with you is that you tend to leave just when things are getting interesting from my perspective!)
confusingly incapable of understanding the plain meaning of words on a page [...] >20% of them to be able to adequately answer “why talk about height instead of something else?” in a five-sentence paragraph
Okay, so this is an interesting one! Notice that I don’t just say it’s “not clear”, I give a reason why I think strength would be a more natural choice, and then go ahead and propose a cynical explanation (marked as uncertain with, “One has to wonder”) for why I think height might have been chosen (viz., to imply that fortitude is immutable in order to absolve “emotionally short” people of responsibility for engaging in emotional blackmail tactics). That should be a tell that “not clear” was a Tenth Guideline compliance move. I don’t want to confidently claim that that was the reason height was chosen, because I never have hard evidence about other people’s true inner motivations. But there does seem to be textual support for the immutability-as-absolution hypothesis: e.g., the text rhetorically asks, “Have you considered your own role in contributing to this situation where you’re claiming victim status?”, and I’m basically intending the same question literally.
I’m open to suggestions that I shouldn’t be using “not clear” in this kind of situation (because it’s, um, not clear), but I don’t think this one was a reading comprehension failure.
In general, when I seem to be saying something bizarre, I hope readers will consider the hypothesis that it’s just a language oddity which I am totally willing to unpack on request. For example, in a different conversation, my interlocutor had just made a negative judgement about me, and I said, “If you want to tell me why, I’d be really curious to hear it!”, and they said that they thought I was misreporting, that I wasn’t actually curious. My explanation for that is that “I’m really curious to hear it” wasn’t meant as a literal report of the raw emotion of curiosity; it’s an idiomatic stock phrase meant to elicit more detailed criticism by conveying the idea that I’m not going to to punish it.
Meanwhile, plenty of other commenters regularly carry that banner.
Well, yes, of course. The reason why the soul of Less Wrong is worth fighting over is because it’s a Schelling point that attracts high-quality people who are worth paying attention to. If I think I have an “edge” along some particular dimension that’s worth trading on, that obviously doesn’t mean other people don’t have their own edges over me. (For example, I think I’m particularly skilled at noticing problems due to selection effects on approved information, but you’re probably more skilled than me at noticing problems due to assuming that one lives in the max-likelihood world.)
One can thump one’s chest and claim to be many things that one is not.
Well, yes, of course! I emphatically agree with that: that’s exactly why I think a culture of diverse criticism is so important! I obviously don’t expect people to just take my word for it that me and my allies are the ones championing actual rigor: I expect them to read what we say and read what our critics say and judge for themselves. That was Said’s case against the user ban functionality (that it made it easier for people to claim to be something they’re not, by making it harder for other people to point out that they’re not).
But it really seems relevant to me that I am the one arguing Eliezer’s position from 13 years ago.
To be clear, I did say “Founding Values” and not “Founders”. Yudkowsky from twenty-three years ago put up a page about Crocker’s rules. (To preëmpt the obvious knee-jerk objection: yes, I did read the part about “Crocker’s Rules does not mean you can insult people.”)
The reason I’m considering the 2002 page about Crocker’s rules but not the 2013 comment about hedonics to be part of the founding ideal, is because to me, the founding ideal isn’t about this Eliza Yudowski person.
It’s about the philosophical vision articulated in “The Bottom Line”, “A Rational Argument”, “What Is Evidence?”, the “Technical Explanation”, and the “Twelve Virtues”—about seeking the mathematical laws that govern how a small part of the world (a “map”) can function as a predictive model of the rest (the “territory”), and how conditional predictions can be used to steer the world.
Crocker’s rules is obviously consilient with the philosophical vision. “Anyone is allowed to call you a moron and claim to be doing you a favor,” “[w]hich, in point of fact, they would be,” says the page. (You see, because having an accurate map is instrumentally convergent for rational agents: if I were an idiot, I would want to know about it, because that might have decision-relevant consequences.)
Of course, humans can’t embody the ideal, but it’s important that there is an ideal, and claims that humans shouldn’t even try to better approximate the ideal because it’s supposedly impossible don’t comport with the philosophical vision I learned from the texts. (As it is written, “The ninth virtue is perfectionism. [...] In every art, if you do not seek perfection you will halt before taking your first steps. If perfection is impossible that is no excuse for not trying.”) If later statements by this Yudowski person contradict the philosophical ideal, I go with the ideal, not the person, because that’s what the texts taught me to do.
It seems to me like one of the contributors here is that they believed that social graces were important, and that managing the feelings of the participants was important.
Unfortunately, the reason I haven’t been able to engage with this intuitively-surprising-to-me claim much is because I’m not a specialist in 17th Century English history, and so I’m not in a good position to evaluate the claim without a lot of catch-up labor.
The reason it seems like an intuitively surprising claim to me is because—as a non-specialist, I thought the standard explanation for the Royal Society’s success was, um, Science? The experimental method? Right? Like, these are the guys whose motto was Nullius in verba, “Take no one’s word for it”, said to be “an expression of the determination of Fellows to withstand the domination of authority and to verify all statements by an appeal to facts determined by experiment.” I take this to mean that if someone who could contribute to your project (with money, skills, &c.) is sad that you won’t take his word for it, you have to disappoint him and find some other way to keep the project alive.
Said and Zack appear to think some things should have 0 cost, and habryka and I think those things have a cost that cannot be ignored; the question is about what price the culture should put on them.
So it’s not that the cost should be literally zero. (I agree that Crocker’s rules doesn’t mean you get to insult people; Said believes in his version of politeness.) It’s that there is a normative ideal about information processing, and you want a culture that socializes the humans into aspiring towards the normative ideal even when it’s hard (Bayesian reasoners wouldn’t hide from information, I want to be more like that), and the “emotional tallness” excuse doesn’t even pay lip service to the ideal.
as the alert reader might have anticipated immediately, it is not about putting correctness before all other concerns
You say that so casually! Okay, sure, humans can’t put correctness before all other concerns. But that’s, like, non-normative, right? I’m more optimistic about a culture that encourages striving for the ideal in a non-formalized way, than a pseudo-economic culture that talks a good game about “prices” (which, according to the microeconomic theory, should theoretically exist), but which, in practice, I think psychologically functions as an excuse to not even try. Is that a better crux?
perhaps, for example, ad revenue.
I should confess that that was not my finest work, rhetoric-wise. (I was in a rush that week to ship as many posts as I could before Habryka got the draw on me, posting on the 14th, 16th, 17th, and 20th, and quality may have suffered somewhat.) Hopefully my comments here about the importance of normative ideals in culture is clearer.
we disagree about how to accomplish the mission of an intellectual forum like Less Wrong purports to be
Right. (But, um, note that I think most apparent “disagreements” among humans are actually disguised conflicts; I don’t think I’m obligated to take your self-report literally, nor would I expect you to trust mine.)
was not tracking that how much and what Eliezer was posting was downstream of what sort of an environment LessWrong was, and that it was smoother and more rational to work inside of those dynamics
I’d expect you’d agree that focusing on Yudkowsky feeling comfortable with posting is Goodhartable. We likely disagree on how Goodharted it is in practice. (As I’ve written about at length elsewhere, I think a lot of Yudkowsky’s work since at least 2016 is just not very good for reasons that have to do with him giving up on the normative ideal.)
To be clear, I agree that people do not have infinite time. That’s why I’m grateful that the rejected-posts slushpile gets censored. The slushpile’s signal-to-noise ratio is terrible! Huge thanks to the mod team, and Pangram, for that.
But the signal-to-noise ratio of a brief critical comment is great, if it’s pointing at a real flaw. (The brevity keeps the denominator small.)
This is a contingent fact about you and how you relate to spending your time on receiving and responding to marginal (or bad) criticism
An important part of the scrupulosity of caring about whether the things I say are true is that it’s not all about me. Just because I don’t have time to respond to my critics, doesn’t mean I should have the authority to decide which of my critics are “marginal” and prevent other people from reading them.
Maybe that’s a contingent fact about me, but I think I’m correct to describe the nature of the contingent fact as me caring about whether what I say is true or false, as contrasted to people who apparently don’t. If you say things, and you not only don’t want to receive corrections, but you also don’t want other people to be able to read corrections, I think it’s a reasonable inference that you just don’t care very much about whether the things you say are true! Right? I don’t see what I’m missing here.
Reply: “Contra Pace on When to Apologize”