jimmy

• jimmy 18 Mar 2026 15:26 UTC

  2 points

  0

  in reply to: Steven Byrnes’s comment on: You can’t imi­ta­tion-learn how to con­tinual-learn

  The part that’s not clear to me is that giving Grog a database of 1000 textbooks is just as good as walking him through and explaining the contents of 1000 textbooks within a long context window, for a Grog with an impractically large brain. Or rather, I know that it’s not the same, but I don’t know what the limits are or how they work. When Claude switched from reading book length pastes serially to “having them available”, the difference was very obvious. It switched from learning the material as well as a human to the sort of incompetence you’d expect from someone who has the book but hasn’t actually done the reading.

  I’m with you on “Grog would need to spend years developing a deep understanding of optics and lasers and so on”, but it’s not obvious to me that the years of going through 1000 textbooks in one impractically large context window can’t simulate the learning process itself, and through that simulate a deep understanding of optics and lasers even in theory.

  I don’t see you make an argument for why simulation of the learning process itself isn’t possible. I see you concede that “(so-called) “in-context learning” can imitate a small number of steps of actual weight updates”, but I don’t see an explanation of how tight this bound is, exactly, or how it works. Maybe I’m just being a dummy and missing it, or maybe it seems obvious to you because you’ve been thinking about this kind of thing more than I have and so some of the arguments are implicit in ways that aren’t coming across. Either way, pointing to a reason that the learning process itself isn’t something that could be simulated in an astronomically large context window would be a thing you could add which would help me understand.

• jimmy 17 Mar 2026 18:17 UTC

  2 points

  0

  on: You can’t imi­ta­tion-learn how to con­tinual-learn

  I’m nodding along on the basic claims (I think), but still trying to digest the implications. I think one of the things I’m taking away from this is that even though human architecture is different, this failure mode applies and is really common. Not sure what to make of that yet.

  (If we’re talking about what a sealed “country of human geniuses” could do over the course of, like, one minute, rather than over the course of 100 years, then, yeah sure, maybe that could be reproduced with future LLMs! See von Oswald et al. 2022 on how (so-called) “in-context learning” can imitate a small number of steps of actual weight updates.^[1])

  Am I correct in understanding you to be pointing at a practical rather than a theoretical limitation here?
  
  Is the reason that you think it could work for a minute but not 100yr a practical matter of efficiency or one that has a more fundamental limitation that you couldn’t get around with infinite context window/​training data/​etc?
  

  Will the trained imitation-learner likewise keep improving over the next 10M moves, until it’s doing things wildly better and different than anything that it saw its “teacher” deep Q network ever do? My answer is: no.

  Even with a context window that contains all 10M moves, or do you mean within reasonably limited context windows?
  
  It seems like the answer with unlimited context would depend on whether the transformer was able to model the teachers learning process itself. I don’t see any reason this shouldn’t be possible in theory, do you?

• jimmy 13 Mar 2026 23:15 UTC

  2 points

  0

  in reply to: NoSignalNoNoise’s comment on: Don’t Let LLMs Write For You

  English class always seemed to me to be about saying absolutely nothing in flowery language. Almost like a politician speech, where the point is to talk in such a way that the point is to make the words sound so nice that you don’t notice that the content is missing. Only I don’t think it’s intentional, just a byproduct on focusing on the language without having anything interesting to say.

• jimmy 12 Mar 2026 15:59 UTC

  5 points

  0

  in reply to: Mo Putera’s comment on: Not Lov­ing Lik­ing What You See

  Because she’s not a counterexample at all. She’s a demonstration that you can get what superficially appears to be passable results while giving less fucks about things. The rule are the rules, and knowing the rules helps to highlight what’s really going on because you know where to look to find the “missing” downsides.

  If I were as unmotivated as she was by the loss of her husband, there’s someone who might be dead right now. If she were as motivated as I was, her husband might be alive. The only reason “Her husband killed himself” doesn’t come off as completely damning to her way of going through life is because the narratives we tell about such things are designed to numb us of these things already. “She did what she could! It’s not her fault!”. Did she really? Do everything she could? Would he likely have killed himself if she grew up suffering more? Sure. Would it be bad to blame people too readily for their spouses’ suicides? Sure.

  But what probability of saving your spouses life does it take to justify enduring suffering for a couple years as it tears you apart and allows you to rebuild into something that might have made the difference? Can you actually justify that the chance of making the difference is lower than this number? For her too? Confidently?

  These are hard questions to answer affirmatively, and you don’t get the answer right except by luck if you’re running on defaults. And coming up with convincing answers to this does chip away at the suffering, to the extent that we actually find them convincing. The woman I know who lost her husband to suicide would probably say something similar to Jo, but I know it hurt. I know she cared. The reason she was able to get back on her feet relatively quickly and successfully is that she actively threaded that needle between the courage to change what she could and the serenity to accept what she can’t.

  Jo’s story is of finding motivation to stop touching hot stoves somehow, leaving the “somehow” unspecified. And of failing big where it matters most. In a way that we give her a pass for because the rest of us are prone to fail there too. Potentially even more often, given how common and debilitating saturation can be around serious issues like that. Her story demonstrates that it’s possible to get by alright if we turn down our sensitivities a good bit, and blame luck for the bad things like her husband’s suicide more than we credit it with the good things like “not dying on the stove”.

  It doesn’t demonstrate that it’s better, either in expectation or in her exact case. Or that the rules somehow don’t apply to her. Just that we generally have so little clue of what we’re doing that it’s easy to spin that way.

• jimmy 12 Mar 2026 3:02 UTC

  2 points

  0

  in reply to: JustisMills’s comment on: Don’t Let LLMs Write For You

  What’s your explanation for how LLMs can create non messy writing without conveying a clear point, and in your view do humans ever do it? I’m finding myself simultaneously agreeing with your first claim and also thinking “but English class”, and I’m not sure how to square these conflicting ideas.

• jimmy 11 Mar 2026 17:11 UTC

  3 points

  1

  in reply to: Mo Putera’s comment on: Not Lov­ing Lik­ing What You See

  If you step over enough disfigured corpses and look only at seemingly normal people, eventually you’ll find one who through some combination of luck and incomplete impairment seems to have made it work.
  
  Some of this applies to me. The only time I’ve ever asked a doctor for “pain drugs” was to see if they could give me something to make injuries more painful so I was less likely to ignore the signs. The other day I noticed that I had cut myself because I saw bleeding, and I have no idea how or when it happened. If I play up this side of things, I could make a case for suffering being unnecessary and make some valid points.
  
  At the same time, when I’ve faced similar loss/​threat of loss, it hurt. And that hurt motivated quite significant changes to who I am as a person that definitely would not have happened if I were numb to it. On a brief overview like that, the difference isn’t gonna show up. Sure, my wife hasn’t killed herself, but that’s normal and I don’t have any reason to believe she would have otherwise. But if there were a missing corpse walking around living somewhere… who’d know?
  
  There has to be some ability to detect harm, and motivation to correct it, or it won’t happen. At the same time, sensitive systems saturate easily, and saturated controllers don’t control well. Turning down the gain can be useful if you have reason to believe saturation is worse than insensitivity in the context in question, but avoiding one failure mode invites the other. It’s a discrimination problem, and not a trivial one. Ideally you train capacity and have both.
  
  
  

• jimmy 10 Mar 2026 21:34 UTC

  6 points

  2

  in reply to: cousin_it’s comment on: Not Lov­ing Lik­ing What You See

  We might not procreate many times per day, but much of what we do is in service of this goal. It might be many times per day that you meet a potential spouse. Or a person who might introduce you to their similarly-attractive friends, who could end up being your spouse. Or a person you might be seen with, and therefore judged by, when a potential spouse sees you and has to infer what hidden attractiveness you may posses. And so many times per day you have to decide whether to take step towards or away from these potential futures and whatever procreation they may or may not entail.
  
  That’s not to say “it’s good that ugly people get bullied”, because there’s quite a lot of room between that and “it’s so bad, that it’s better to fully numb ourselves to appreciation of beauty”. Pain killers have their uses in specific cases, but congenital insensitivity to pain is no bueno. You don’t want to throw the non-ugly babies out with the bathwater.

• jimmy 6 Mar 2026 19:01 UTC

  3 points

  0

  in reply to: Taylor G. Lunt’s comment on: An­drew_Critch’s Shortform

  The more interesting argument for that norm is that it makes people accountable for their downvotes and therefore less likely to give dishonestly motivated downvotes.
  
  There’s still no obligation to upvote anything, so if it’s plainly visible that a post is bad and no one cares to explain why it’ll just sit at 0. Downvotes become important when some people (incorrectly) think a post is good, because then it will accrue a positive score if uncorrected. But in that case the downvoter thinks they understand something the upvoters don’t, so maybe they should be explain.
  
  The problem with downvotes without accountability is that if I post something about how people named Taylor are statistically likely to be <bad thing>, it might be true and well supported and important and empathetic… and still you could still just downvote it to censor what you don’t like while most people don’t care enough either way to vote. So we get good posts systematically suppressed in ways that wouldn’t happen if you had to comment “Downvoted because my name is Taylor”, whenever a minority is hostile to a particular truth.
  
  Downvote explanations could be hidden by default so it wouldn’t spill over, but I find myself frequently expanding “comment scored below threshold” to see what it is the community in question really doesn’t want people to think. These comments are rarely boringly bad.

• jimmy 23 Feb 2026 15:43 UTC

  4 points

  0

  in reply to: Valentine’s comment on: Ir­ra­tional­ity is So­cially Strategic

  I’m not sure what I said that you’re interpreting as my having said “probably important”.

  Hah
  
  I got it from here:

  I agree that there’s a way for him to rearrange his relationship to the situation such that truth-seeking isn’t bad anymore. Strong agree. I think developing that type of rearrangement is probably an important part of a mature art of rationality.

• jimmy 20 Feb 2026 18:35 UTC

  8 points

  1

  in reply to: J Bostock’s comment on: Ir­ra­tional­ity is So­cially Strategic

  
  It was her feet and stress induced, but yeah, reduced circulation. I literally mean “Just don’t do that”, as in “Send more blood to your [extremity]” which is how I actually worded it when my wife noticed reduced circulation in her hands due to cold air despite a warm core.

  I have a long sequence about how to resolve “psychological issues” with rationality, and give examples of this friend learning to not swell her injuries because I told her it’s something you can just decide and of helping my wife constrict her blood vessels with a bit more guidance when that was desirable for stopping bleeding. Hopefully the explanation on the latter will help it feel a little less weird, but I’m not sure how much sense it’ll make as a stand alone post without going through the build up.

  The problem really is as simple as “she wasn’t trying”, but the hard part is not getting stopped from trying by the belief that you can’t do it (e.g. if I told you to “start levitating” it’d feel impossible because it is and you wouldn’t know where to begin, and this makes the process of actually trying difficult. It’s the same issue even when you’re wrong about what you can do). So like, the question becomes “What does it take to change this belief” and the answer can be as simple as “just tell them” which is why it worked in these cases, but there’s a lot going on under the hood such that when I say weird shit like that she actually believes me.

  And even then there’s some residual weirdness, like the fact that she thought she had tried to dilate her blood vessels before, but despite succeeding on the swelling on her own it wasn’t until I actually walked into the room that she was able to expect her blood vessels to dilate… which turns out to be the same action as intending. So I’m still working on getting things to generalize as far as they “should”.
  

• jimmy 20 Feb 2026 18:24 UTC

  6 points

  2

  in reply to: Valentine’s comment on: Ir­ra­tional­ity is So­cially Strategic

  
  I’m trying to tease out where we differ, exactly.

  I think it’s here:

  Which means that if having a distorted model helps solve the embedded problem, you’ll probably have a distorted model of what the problem is and/​or how you’re solving it.

  I think there’s a tricky distinction between “model” and “meta model” being lost here. The model is distorted relative to what and according to what model?

  Control systems have to be a model of the system it’s controlling because of the good regulator theorem. If we take a maximally simple control system like a bang bang controller, the structure of the control system has implications about the structure of the plant being controlled. So for example, a bang bang controller is not going to work as a servo position control. It could work for the velocity control loop, or a thermostat, but if you try to use it to control position directly it will oscillate.

  If the controller works then it’s fit to reality. The explanation can be wrong in all sorts of ways but the controller itself has to conform to reality or it won’t work.

  Bringing this back to the guy trying to attract women, he can say whatever he wants about “She’s gonna like me because I was born on a prime day which is interesting, and chicks dig interesting guys”, and if he uses that to justify the decision to stop communicating to women “Don’t like me” then that will help him out… but the “distortion” is in his meta model, according to my meta model, and his object level model is actually more fit to reality or it wouldn’t work.

  So when he says “I gotta believe that she’ll like me, so she’ll like me. Truth tracking isn’t necessarily optimal when you’re embedded in the system!”, I see that as another distorted model. He’s trying to twist his meta meta model in such a way that it doesn’t untwist his meta model in such a way that twists his object level model out of congruence with reality. Instead of just untwisting the relationship between his meta model and object level model, and playing it straight which is gonna work better.

  Scratching the next couple paragraphs that came after this, I think I see now that you were using the word “embedded” a bit differently than I interpreted it. Hm. Now I’m less sure I disagree about anything. Because you agree that it’s possible to do truth tracking so that it works, even though “probably important” is a much weaker endorsement of this path than I’d have given. And I strong agree that occlumency is an incredibly important rationality skill (at least, in existing suboptimal cultures).

  Maybe I’d just have phrased it “Irrationality is socially strategic (if you’re irrational)”

  Okay, maybe this is it: I think occlumency is important because figuring out how to get your meta models sufficiently clean is hard. Like you’re not gonna manage delivering harsh but accurate criticism in a way that people receive as loving and helpful the first time you notice that accurate criticism would be harsh, because you don’t know how to be loving and helpful yet. You don’t yet see your errors for what they are, and so you’re necessarily going to have all sorts of distortions in your meta models to where naive implementations are going to predictably fuck up your object level models if you try to play it the first way that comes to mind. So like, “Back the fuck off. I don’t know what I think yet, and I need to be able to explore thoughts that won’t hold up” is critical. In the same way that training wheels make it hard to learn how to ride a bike. Sure, they keep you upright which is where you generally want to be, but it’s only by leaning that you can learn how to regulate your lean, and that’s a crucial part of balancing for yourself let alone turning.

  So we end up with a bell curve meme where the low skill end says “Have you tried not having the problem, you dummy?”, the center of the curve says “Nooo! Saying this is going to cause all sorts of problems because this and that!”, and the high skill end comes back to “Have you tried not having the problem, you dummy?”—which actually works after having untangled all the distortions that lead to that coming off as unloving and un-reality-tracking the first time it comes to mind.

• jimmy 19 Feb 2026 21:57 UTC

  3 points

  0

  on: Emo­tional Disper­sion and Patience

  At every moment that you have to put up with fools, you have the option to choose to be patient, to not interrupt their ramblings. You can feign losing your temper with them if it will result in a better outcome, but no matter what, it is always better to be capable of continuing to put up with them. And when you reframe patience as a choice, that becomes possible.
  [...]
  Either Ego Depletion doesn’t exist, I don’t experience it, or I haven’t been in annoying enough situations to experience it. But I have never felt that I couldn’t possibly continue to choose patience, despite having bad executive function (lol)

  I’ve experienced something close, which was really weird because I normally experience things the way you describe. I’m not ashamed to yell at someone if it’s the right thing to do, so I don’t have any motivation to pretend things “aren’t my choice”. As a result things almost always feel like a choice, it’s almost always easy to choose to not take anger out on people, and the times when I choose anger it actually works because there were reasons to choose it.
  
  The time where that broke down wasn’t an exceptionally annoying situation, but it was after a month of being overloaded with stakes high enough that I couldn’t afford to decline to take on the stress. The way it felt is like each time an imperfect decision is made there’s a rounding error to deal with, and work to do maintain connection between my reflective thinking and my object level thinking. And then after enough overload I just couldn’t repair the damage as fast as it had been accumulating, so I was starting to become an outsider to my own behavior.
  
  Not like “Leave me alone, I can’t help it! Not my responsibility”, but like “That idiot is about to make a big mistake. I’d stop him if I could, but he’s not gonna listen to me”—only I was talking about myself.
  
  Welcome to LW :)

• jimmy 18 Feb 2026 17:57 UTC

  22 points

  0

  on: Ir­ra­tional­ity is So­cially Strategic

  A really powerful rationality scene would, I think, systematically cause its members to dissolve their stubborn problems simply by being in the scene for a while. The dissolution would naturally happen, the way that absorbing rationalist terms naturally happens today.

  In my dream fantasy, just hanging out in such a space would often be way more effective than therapy for actually solving one’s problems.

  Like keeping things dry and oxygenated instead of using antibiotics to stop anaerobic rot.

  I’ve seen this work. One time I tried to help an online friend with her fear of needles, and it didn’t really work. Then she flew out and visited for a week, and the issue resolved itself without ever addressing it further. My best attempt to track back what happened is we did scarier stuff for fun (like jump rock) and she got to experience trusting someone else more than her own fears.

  Recently, a different friend told me her Raynaud’s issue dissolved as soon as she saw me enter the room, because she was gonna ask for help and then realized that she didn’t actually need help because I was just gonna say “So stop?” and that she would. The active ingredient here being that I actually knew she could, and beliefs are contagious when well grounded and embodied.

  I agree that “create a space where nonsense falls apart” is the way forward, and it’s something I’ve been thinking about myself recently. Like, how do we scale this?

  Like if a man approaches a woman he’s interested in, it can be helpful for him to bias toward assuming she’s also probably into him. His bias can sometimes be kind of a self-fulfilling prophecy. Truth-seeking is actually a worse strategy for getting the result he actually cares about. That fact wouldn’t be true if his epistemic state weren’t entangled with how she receives his approach. But it is.

  I get that your point here is that the entangling distorts things, but I worry it comes off like the problem is inherent in him being part of the system. “Truth seeking is actually worse” is a local property of his broken truth seeking strategy, not one inherent to being in the system.

  The guys I’ve seen be the most successful with women don’t project “She’s gonna like me!” vibes at all, they’re just not worried that she won’t. And yes, “Not worried that she won’t” does convey some information about his likability, but it is completely compatible with “She really might not” and often the engagement with the latter possibility is active ingredient. The problem isn’t with his embeddedness in the “him+her” or “him+her+others” system, it’s that he’s embedded in his own mind and hasn’t yet figured out what to do with “she might not like me”.
  

• jimmy 18 Feb 2026 8:20 UTC

  4 points

  0

  in reply to: Zack_M_Davis’s comment on: Hazards of Selec­tion Effects on Ap­proved Information

  Sometimes there’s a recursive problem that I haven’t figured out how to deal with, when someone is demanding narcissistic ego supply as a precondition for talking, and I don’t see any way to comply while still making progress in the conversation, because the specific thing I want to talk about is how it’s bad to demand narcissistic ego supply as a precondition for talking.

  Haha, yeah. It’s a tricky one for sure. And really tedious. And believe me, I feel ya on this. “Demanding narcissistic ego supply” makes it sound really pathological, which it is, but it’s also something that’s essentially ubiquitous. Especially when you get to the truths that matter most. Figuring out how to deal with this from the inside and out is something I think is of utmost importance for rational thinking both on the individual and communal level.

  I spent a while talking to a woman with pretty serious BPD as an exercise in figuring out how to talk to people who are difficult in this way. It was difficult, and required being very meticulous with my wording, but eventually I did figure it out. Also, eventually the effort paid off in clearing some room for less sensitive interactions. By the time I felt I had learned what I needed to learn and we drifted apart, I was able to laugh at her for stuff that would make most people freak out, and she’d laugh with me because she realized she had been a bit silly.

  This isn’t to say “you should do this” because again, tedious and all that, but it’s nice knowing that there is a way to go about it for when it’s worth the effort.

  My strategy has mostly been to address it an a meta-discourse post like this one, or in my memoir sequence. The reason to stick to the object-level in the moment is because I anticipate that addressing it head-on would just immediately deadlock. There’s nowhere to recover from “You’re being unfair”, “No, you’re being unfair”.

  Yeah, I agree that posts like these are valuable. And also that addressing it head on is… problematic, at the moment. I think there are times when it can be worthwhile, but the battles have to be both picked and fought very carefully.
  

• jimmy 16 Feb 2026 20:16 UTC

  4 points

  0

  in reply to: Zack_M_Davis’s comment on: Hazards of Selec­tion Effects on Ap­proved Information

  Examples? Who is exploiting the other side?

  Answering this publicly would explode into a shitstorm of drama in the best case, and have more silent failure modes in the more likely cases, so I’m gonna decline to do that :)

  If you want to get my perspective of who is actually erring in this way, I’m happy to answer it in PM but I’ll still want to caveat it a bit to make sure its framed correctly, since I anticipate some potential difficulties.

  Here, I just want to highlight the abstraction because then people can see who fits the pattern in their perspective rather than accepting or arguing for some collective idea of who ought to be treated as if they do.

  No, it is not effortless! For example, I was late to respond to your top-level comment because I was too shy to check the comments for 36 hours after posting.

  Haha, okay. That’s what I was expecting. The effort shows, FWIW.

  Yes, and in particular the relevant standard isn’t about not having emotions; it’s about not making my emotions other people’s problem.

  Yep. And that’s a norm I wish we had here. It’s clearly correct IMO, and also clearly not enforced. I will strong upvote anything that makes this case until its beating a dead horse into a paste from which no zombie horses can arise.

  But that’s my problem, not a lever to control what other people are allowed to say to me or about me. Emotions are information. If people say things to me or about me that make me feel bad, then maybe I should feel bad.

  I’m mostly with you here.

  Where I get off is like… well, if you experience my comments as hard to face, then that becomes my problem too. Because I want to have this discussion with you, and so if you experience engagement as too emotionally challenging then I’m not going to get the engagement that I want. So if you’re having bad feels when reading my comments, I would definitely appreciate you letting me know so I can figure out what I should be doing differently. It’s not “controlling” my behavior in the problematic sense, but it is helping me close the loop so that the emotional information conveyed stays accurate.

  And this isn’t just a private exchange between the two of us. Less Wrong as a whole is able to read and might get something out of this exchange, so if you bow out for reasons that ultimately don’t track to “He had nothing to say, and nothing to hear”, then Less Wrong as a whole loses out.

  That means if I were to start my comments off with “Zach, you idiot, you’re wrong here because...”, that would probably be less pleasant for you, you’d probably engage less, and we’d all get less out of the interaction. And that’s bad, so we want norms that result in discouragement of that kind of comment. Even if I omitted the explicit insult and were to just talk to you the way one talks to idiots, it would just be “tone” but it would still be bad, for the same reasons.

  Which is tricky, because the proper response to “Zach, you idiot” is “No, hypothetical-Jimmy, you are being an idiot for thinking that kind of attitude is appropriate towards Zach here. It’s appropriate towards you here”. So you can’t tone police by saying “uncomfortable tones are disallowed”, just inaccurate tones, and wtf is an “inaccurate tone”? I mean, there’s an answer, but not in the communities common knowledge, you know? The whole thing gets messy and the answer does keep coming back to your “free marketplace” ideal, but it highlights a few hidden points that really need to be a lot more free than they are for the market to work.

  If people say things to me or about me that make me feel bad, then maybe I should feel bad.

  Yeah, I mean, maybe. Or maybe not though, right? You don’t take yourself to be above having wrong feelings, do you? :p

  The bad feels are yours and therefore your problem, yet. But who are they about?

  If they’re saying “Oh no Zach, your post was bad!” then sure, I guess you gotta figure out whether that’s true or whatever. If they’re saying “This meany jerk man isn’t being fair to me! Their criticisms are dishonest!” then like… maybe you’re right?

  Maybe you aren’t, of course. Maybe your post is just bad and you’re avoiding the humility to notice, so we have to account for that possibility too, but it’s also possible that you’re just right.

  To the extent that you’re right, there’s room to hold people accountable for being dishonest and unfair. Your strategy of “Take full responsibility, respond on the object level, let the truth shine through” is a great start because your arguments for it all hold up.

  But there’s an implicit “I shouldn’t address it head on” that I don’t think actually holds—or at least, doesn’t hold in a community with healthy norms.

  The community with healthy norms isn’t gonna say “Oh no, conflict is uncomfortable so we’re going to stick to the object level and pretend that it doesn’t exist and doesn’t need to be dealt with”. The community with healthy norms is going to say “I notice that you feel unfairly treated because the other side isn’t doing their duty to engage honestly. Let’s find out whether this is true”—and then you either get an additional dose of bad feels to sit with when it turns out “Nah, your post was just bad, the criticism was valid, and you were just doing arrogance” or you get off scot free, your perspective is vindicated, and the person doing dishonesty has to sit with the bad feels of “You were bad, don’t do this again”. Ideally both sides trust the resulting judgement to be fair and either happily submit to the process to correct the one that’s wrong even if it’s them, or they know that they aren’t gonna like the outcome and leave the community alone.

  I know there are plenty of people who interpret the unpleasant comments on their posts as problems with the commenters, and I know they’re not entirely wrong, but do you share this feeling at times? Not “In the end, do you decide that they were bad” but is part of the emotional difficulty that it feels like commenters aren’t living up to the standards you hold for yourself, which would be really nice to see them held to—either by themselves or by the community?

  I don’t have a good idea of how it feels to you, but I have a good idea of how I would feel in your shoes.
  

• jimmy 16 Feb 2026 1:10 UTC

  5 points

  0

  in reply to: Zack_M_Davis’s comment on: Hazards of Selec­tion Effects on Ap­proved Information

  I’m roughly equally worried about both. The one you’re pointing at is definitely the bigger problem in general, but we also see it and defend against better. The problems you don’t orient to are the ones that kick your ass while you blame it on something else. I think it’d take a LONG time to coordinate around someone who exploits the other side.

  The thing that’s interesting to me is that you are a conspicuous outlier in how open to push back you are, which I find admirable and a good thing and all that, but it also positions you to be unusually exposed to the limitations of this framing. Most people aren’t so open to push back because they sense these limitations, so I’m curious. Do you experience it as genuinely effortless regardless of the type of criticism you receive? Or do you feel like you are doing something that requires actively holding yourself to high standards because the standards are important?

  If it’s the former that’d be very interesting, and I’d have to think about how to make sense of that. If the latter, I’m pointing at the thing that makes it not effortless. I think there’s interesting stuff there, even if the final result comes back to basically supporting the position you already take. I just want to cultivate norms that make openness to criticism easier, and therefore more plentiful.

• jimmy 15 Feb 2026 8:47 UTC

  2 points

  0

  in reply to: Zack_M_Davis’s comment on: Hazards of Selec­tion Effects on Ap­proved Information

  I agree that’s the hope. And in a well functioning community of peers that can work. Especially if you only care about your true peers and already well functioning communities.

  In practice, how well do you see it working? Does LW live up to this standard as well as you might hope, or do you notice bits here and there where the market for ideas is distorted by other considerations?

  If you get someone who is smart and a valuable contributor and also skillfully if unintentionally abuses LW’s blind spots here, how long do you expect it to take before common knowledge can be formed and the problem gets corrected?

  It occurs to me (at age 38 with no dayjob, no girlfriend, and no children) that efficient markets in life activities may already sufficiently mitigate this on its own: having more time to burn arguing on the internet than anyone else is its own punishment.

  Hah.

  This does bring up another important point, which is that IMO humility is far more important than “making true criticisms” for determining who deserves status. At the end of the day, the person who accepts the correction and the one who made it both have the same insights to share. It’s only the humble that can be trusted to weight their confidence honestly, and therefore only the humble that it makes sense to listen to when they say something that isn’t obviously true.

• jimmy 14 Feb 2026 5:45 UTC

  14 points

  7

  on: Hazards of Selec­tion Effects on Ap­proved Information

  I mostly agree, and I’m glad to see the point written up here. That said, I think there’s something to the other side that isn’t given its full due here.

  (If you independently got everything exactly right the first time, then there would be nothing for critics to do; it’s just that that seems pretty unlikely if you’re talking about anything remotely complicated. It would be hard to believe that such an unlikely-seeming thing had really happened without the toughest critics getting the chance to do their worst.)

  Of course there’d be stuff for critics to do! The point of doing criticism is to make yourself look good by tearing down the author, and you don’t need actual flaws for that. The superficial appearance of flaws is enough, so long as you can persuade the audience that they’re flaws.

  Heck, you don’t even need that. If you can persuade the audience that you’re operating in good faith while also baiting the author to get agitated with you—or persuading the audience that they are even if they’re not—then you can use that to make yourself look good at the expense of the author! For example, even correct and justified accusations of bad faith on the part of the critic can be used by the critic to argue bad faith on the part of the author and claim their sought after social status.

  Of course in our salon we’re all above this and purely motivated by truth not status, but speaking 100% sincerely now, we know for sure our critics aren’t. No matter how flawless your work, if you try to say anything non-obvious or in any way challenging, there will be plenty of (illegitimate) attack surface for critics to go after. And if the community cannot sus out bad actors because the official policy is “If you have nothing to hide you have nothing to worry about!” “If your point is valid, there will be nothing to criticize!” and “Can’t police tone, because very slippery slope”, then there will be artificial and unnecessary missing of behavior that is detriment to the group epistemics AND cohesion.

  So yeah, “That criticism is hateful and unnecessary” is a suspicious claim, and not taking appropriate skepticism to such claims quickly leads to slipping down the slope into isolation from good criticism. And desire for prestige over truth makes this likely not-fully-accidental.

  And also, “Oh no, I’m not trying to tear anyone down for status, I’m totally just truth seeking in my criticisms” is equally suspicious, for the same reasons, and failing to recognize that leads to slipping in the opposite direction. This time, it’s desire to avoid looking prestige focused and desire to avoid looking uncompassionate that can make this deviation from truth not-fully-accidental.

  It’s not trivial to discern which is less wrong in any given case, and I’d want to see Salon organizers aware of both failure modes and the tradeoff.
  

• jimmy 11 Feb 2026 17:34 UTC

  4 points

  2

  in reply to: Gordon Seidoh Worley’s comment on: Vibestemics

  
  Oh geez, looking back at my comment I was extremely unclear. Sorry about that.

  Probably not useful as feedback, but the specific things I’m most interested in here are your conclusions. Like, “Not gonna justify this yet, but I think rationalists are susceptible to getting seduced by witches in ways that will turn their lives upside down. The abstractions that predict this are after the fold, and you gotta apply it to your own raw data”. I’m mostly curious about this because I’m trying to figure out how similar our perspectives are. The more similar our conclusions, the more it seems like “Water in the eggplant” type stuff is true and important just not for me. The more dissimilar, the more I have to wonder “Wait, what do you actually mean by that. I must be missing some patterns he’s matching while thinking I get it”.

  Separately from that, and what may or may not be useful, is that in general I find it helpful to have more concrete applications spelled out. Not raw data necessarily as the point isn’t to fuel independent abstraction, but a minimal set of simulated/​curated data to highlight the connection between your abstractions and concrete use cases. Sounds like you’re mostly on board with this though, at least in theory.

  I’m with you on the “I just wanna tell you about the cool abstractions I figured out!” thing by the way, hehe. It’s a lot easier, and more fun, and genuinely worth doing first I think… just also a lot harder to get through to people IME because grounding the abstractions is much harder than holding them in the abstract.

• jimmy 10 Feb 2026 8:15 UTC

  4 points

  0

  in reply to: Gordon Seidoh Worley’s comment on: Vibestemics

  Yeah, I’m familiar with Chapman and loosely follow him. His latest post was especially interesting to me, because it was unusually concrete about what people actually get wrong in practice. It’s one thing to say “There’s water in an eggplant” to which, “Yes, and?”, and quite another to show someone’s life falling apart because of things they’re systematically not seeing. To which it’s much more “Could that happen to me? And if not, what exactly is stopping it?”

  I’m probably not your intended audience because I’m already familiar with Chapman’s work and mostly agree with your position in the quoted statement, but that’s the kind of writing I personally would be most interested in reading from you, for whatever that’s worth. I’ll read some more of your stuff though and see what jumps out at me.