Maybe something like a “long re-friending”? “Re-friending” (or just “friending”) seems to me to capture the “coherent”, “grow up further together” part, and to be less authoritarian-sounding than “correction.” Goes well with claims AI made by labs are minority elites trying to run away with the world’s decisions and stakes, better if we figure out how to be friends across factions first.
AnnaSalamon
Karma: 21,266
I do not think the requirement of curiosity (as I’d render it) is excessively difficult.
A) Let’s let “curiosity” indicate a pattern of cognitive actions in which one looks into things, keeps an eye out for unknown unknowns, heads toward rather than away from relevant-looking information, etc. (I don’t think social contexts have business demanding mammalian emotions from us, but I do think they can have business demanding patterns of actions sometimes. This point is similar to the common weddings speech in which it is said that the “love” that the couple is promising one another is an action, a choosing to prioritize one another’s well-being, and that this is in our power to promise even though mammalian emotions are not.)
B) My guess is that conversations work much better if both players are actively curious about the (predictions, intuitions, life experiences, etc) that are motivating one anothers’ statements. And that that it is not an unreasonable / extremely costly ask or anything.
I think you misunderstood most of the views of mine that you’re responding to in this comment. I’m not sure why. Perhaps I am mincing words in a way that leaves things more confusing than necessary? Or perhaps I’m misunderstanding your remarks and you’re actually getting me fine and I’m confused about that? Or perhaps my views are outside what you’re expecting in some way.
Zack: 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”?
No, because in that case Emily absolutely would not have or deserve the mandate of heaven with respect to lunch decisions! (This claim is not-super-related to whether a person wants to be in a conflict with or punish her or whatever; it’s just: in that case Emily would absolutely not be a correct or viable holder of the telos of the non-profit’s lunch purchases.)
When I say Habryka and co seem to me to have, and deserve, the “mandate of heaven” as LW site-mods, what I mean is:
I think better things will happen for this site and the community around it if they keep this role, compared to either:
a) someone else taking over the website (within the realm of actually-plausible replacements), or
b) the website being shut down, or
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].
Plus also I think it’s “more dignified” in some virtue-ethical sense, in addition to likely having better consequences.
I tried to say this pretty clearly. I’m not sure why I failed. Did it make sense this time? My claim here is that Habryka and co’s relationship to lesswrong.com is unlike Emily’s relationship to the lunch orders.
Part of my background concepts here are:
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.
There are indeed circumstances and evidences that can mean it’s better to override a person on a task, if one can – but typically in such cases, it is also better to transfer the task away from the person on a lasting basis, if one can. Your example with fictional Emily and the lunch orders is such a case. My view is that the ban of Said on LW is not such a case.
—-
Re: local social norms about telling the truth about awkward local social matters, in cases where they’re a public matter (such as mod decisions or social norms, but not most users’ personal lives):
I was not trying to convince you lesswrong had such a norm, in that section. (re: your statement “You can’t possibly expect me to be that gullible.”) I was trying to answer your question about why I believed LW has such a norm.
My answer to “why do I believe LW has such a norm,” (which I maybe could’ve said more clearly last time):
I believe in such a norm, and am trying to practice it here.
I also believe that LW’s mod team, and LW’s site culture broadly (including among the users), is allied enough with this norm (and accommodating enough of this norm), that I can, without being in too much incoherence with myself, aim to:
Practice this norm myself;
Root for a LW that is led by this mod team;
Practice this norm to some extent on behalf of my “believing in” of the LW project.
Re: your statement that LW can’t hold this aspiration given e.g. a particular Habryka comment: FWIW Habryka’s request that you link to seems reasonable to me, and is one that (at least as I choose to interpret his request) I was already trying to follow, without remembering he had made it explicitly: 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 (at least, that’s how I endorse reckoning these things; you’ve stated that you doubt this about me, and I’m not trying to give you contrary evidence here, just stating what goal I’m endorsing). But: I try first to see whether I have some lower-cost way to accomplish the same thing, and I do less of it than I would if it were cost-less.
I borrowed some of how I’m thinking about this stuff from reading (parts of) Toqueville’s “Democracy in America,” a long time ago. Toqueville says America is founded on two ideals – freedom, and equality – and one cannot fully optimize for one goal without trading off some against the other goal, and so there’s a tension. But he says the tension is fruitful, and that cool projects typically have at least two non-identical ideals that are both at least partially aimed at, in a fruitful tension, with development over time into how to reconcile them in practice.
I’d love your help figuring out where you (Zack) and I are disagreeing, here.
—
Re: my object-level suggestions for how a reasonable person 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, and we should be talking more about this part. Does that seem right to you, Zack?
Me: [Said] is changing … the LW user-base’s notions of which [posts, and claims in posts] are “in good standing” [...] a reasonable-person mod might disprefer this.
Zack: By means of arguing about them! …
No, look, I agree (and my inner “reasonable people” agree) that if Said changes a person’s views by arguing with them, and thereby convincing them, that part’s good. The bad thing I am talking about is people observing “mere presence of visible disagreements that I’m not gonna bother to read the details of” in cases via Said starting threads they aren’t gonna bother to read through, and deeming particular claims “disputed by LW users-in-good-standing / too hard to sort through” (via the simple fact that it’s disputed and that the thread kinda goes forever (and not in the exciting “here’s all this stuff you’ll learn if you read this” way), not via finding the object-level comments convincing).
To spell out this argument in more detail:
If you grab a person from the LW’s current “posts rejected for being word salad” pile, and add them to the prolific commenters, the site will get worse. (As you note.)
This is basically because they cost more attention (to the LW users) than how much value they provide.
I’d guess that some on this site who I consider reasonable, and who I expect you’d [consider reasonable if you didn’t know their Said beliefs], who believe the same about Said:
Costs he causes:
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)”
I suspect also Said sometimes responds to a significant chunk of what some users write on their topics of interest.
Then, they could abandon their goal, or respond to many statements of Said’s without learning much, or write less about their topics of interest. IMO it’s reasonable to hold a viewpoint in which all three of these options are costs.
Benefits he provides:
I appreciate some of the challenges he brings sometimes to stuff I think is poorly defended, that I’m afraid might “poison” LW, and his help anchoring parts of local validity semantics. But the “challenging stuff that might ‘poison’ LW” part, at least, is … the sort of “benefit” that depends a lot on tricky matters about what sort of site is good to have here, where some people I respect have different preferences.
He hasn’t provided any large amount of more-obvious “simple contributions” such as write-ups of neat stuff about math/biology/whatever, or case studies of how to use rationality to get somewhere practical, or funny stories that help rationality concepts stick in the mind, or other “good content.” (He does have a couple quality, upvoted top-level posts. But fairly few for being as long-standing and prolific a commenter as he is.)
(For anyone just dipping into the thread here: this is not my own view of Said. I like the site better with him. It’s my attempt to provide counterexamples to the claim [that I think is implied by Zack? But not actually made in these words, so Zack may disagree with it] “there are no reasonable positions plus non-evil goals a person could plausibly hold that would allow banning Said. User-bans plus downvotes would work for all legitimate goals.”)
–
re: why Ben Hoffman disliked Said’s comments on “Zetetic Explanation”
Me: What is your take on why Ben Hoffman disliked Said’s comments under “Zetetic Explanation”? I would guess that for Ben a user-level ban would have been sufficient; but I still think his response is a counterexample the hypothesis that [“downvote and ignore” will be sufficient if a person isn’t seeking to unjustly control their own reputation in others’ eyes].
Zack: 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.
I do not concede that, because “relevant” is a pretty broad term!
As I wrote above, I think Ben Hoffman’s large irritation about Said’s comments under “Zetetic Explanation” is a counterexample to the hypothesis [“downvote and ignore” will be sufficient if a person isn’t seeking to unjustly control their own reputation in others’ eyes]. I’d like to know whether you agree with this?
I’d like to know this because if you do agree, I’m curious for your guess at the mechanism why “downvote and ignore” is insufficient, and I’m curious whether the same mechanism indicates that user-level bans are also plausibly insufficient.
Thus, I remain interested in your take on why Ben Hoffman actively disliked Said’s comments instead of [downvoting and ignoring Said without caring much].
What you see as his 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.
I suspect this (set of paired perceptions) is might help for seeing some part of what is scissors-y about Said.
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’ve attempted to read primary sources from around the origins of Science (quite a bit, though still less than I’d like and also it’s been 20 years), and my own impression is that Science was enabled partly by a designed shift in social graces (by humans who thought social graces matter), but, not a content-neutral sort of advance in social graces (nothing like “this way we can create more harmony in arbitrary situations”), but an engineered change in social etiquette aimed specifically at making it easy for gentlemen who cared about their reputations, and cared also about not being challenged to duels etc., to be able to verify one anothers’ experiments. Like, “we have a norm around here in which everyone aspires to verify everything for themselves, and so if they ask you how you did your experiment, and share their own attempted replications and results, they won’t be doubting your honor, they’ll instead be practicing our local virtues. Also, everyone will keep their speech strictly about what happened when they tried different experiments, and will not e.g. accuse one another of lying—they’ll only say that their own attempt yielded a different result.”
I do think it was pretty darn different from Crocker’s Rules.
re: Zack’s claims about my section (1)
re: Zack’s claim that my section (1) in the grandparent comment is “super weird” and isn’t a thing a person would write if trying to agree/disagree/evaluate the moderator decision on the object level:
I agree it isn’t a thing I’d write if I were considering only the object-level “did the mod team’s decision to ban Said stem from virtues or vices, and did it have important good or bad consequences” question, as it (mostly) isn’t about that question.
I disagree that it’s “weird”: I make comments like this fairly often in various contexts. E.g., while running CFAR, I’ve often had occasion to want to share object-level views on e.g. which way the storage room ought to be organized or how much pizza we ought to buy, while making it clear that authority to make that decision still rested with [person who was charged with it] and they ought to make their own best mistakes, not mine. I also appreciated it when others make this clear. Sometimes there are political fights; sometimes there are epistemic bits of sharing; IMO it often allows a simpler/clearer/less-politically-constrained epistemic discussion, and improves future decision-making (by leaving it clearer who’s in charge, so that we continue to get actions from the in-charge-person’s inside view instead of from a political muddle) to be very clear that one is not wishing to change the political power structure when one is not. Here, I am not wishing to change the political power structure.
Possibly-skippable “nitpick”:
This last bit is complicated I can’t actually express my actual views without it:
Even with respect to the “object level” question of “did the decision to ban Said stem from virtues or vices, and did it have important good or bad consequences”:
There really are at least two “layers of decision” here, and it’s easy to get confused between them, but it’s valuable AFAICT to not get confused between them.
Layer i: the mod team decided(/believed in) they had the authority to ban Said, even without a “here is the specific explicit rule Said broke,” based on the empirics of “as far as the mod team can tell, bad things keep happening (from mod team’s POV) when he’s here, and mod team put in a good deal of effort to try to change this without successfully changing this.”
I confident this layer of the mod team’s decision stems from virtue, and will have expected-good consequences. I expect us to have far more “ability to have good things” if mod teams such as Lightcone’s decide they have this power, vs if they don’t.
(And it’s not obviously consensus or anything; it’s a thing I would’ve gotten wrong at age 18, and I think a fair chunk
Layer ii: the mod team decided(/predicted/believed in) that Said’s actions were unvirtuous, and that Said’s actions had bad consequences, and that Said’s actions were something like “responsible” for the repeated demon threads and “huge moderation overheads” phenomenon.
My inside view disagrees with the mod team on this.
Accordingly, I would have made a different decision if I were (god forbid) in the mod team’s shoes (though I would not have been in those shoes; running LW looks like a huge amount of work and pain to me, and I am truly grateful the mod team does it)
My all-things-considered predictions are… weak disagreement still, I guess? but less disagreement than my inside view, because the mod team spent way more time both thinking in detail about Said’s LW threads (which I haven’t even read most of, though I’ve read a sizeable chunk), and figuring out how to run the site broadly, so I’m pretty sure they’re better than I am at running a site that stays healthy, and I weight their inside view here substantially when arriving at my all-things-considered predictions/would-bet-on’s.
re: Local norms of telling the truth even on awkward social matters
Me: We have instead a strong local norm for telling the truth even on awkward social matters
Zack: So, that would be great if it were true, but is that in fact true of today’s LessWrong 2.0? Why do you think that?
I believe in this norm for LW, and I predict that the mod team, if asked, would say they endorse this norm. 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.
I agree that LW does not fully hit this aspirational norm. I still care about the aspiration, partly because it’s something I personally aspire to for LW and I believe in attempting to do this in alliance with the mod team and with the (I predict) many others on LW who also believe in this norm. (Though I expect plenty of other users don’t believe in this norm.)
re: Causes of “unsustainable costs” from demon threads and mod team facilitation time
Me (adding here a bit more of my original statement): “for whatever reason, when Said was here, there kept being ~unsustainable costs paid by the moderation team, and also the moderation team tried pretty hard to find a better way and did not thereby locate a sustainable alternative that satisficed on all their goals.”
Zack: “Which part of §VI.1 “User-Level Bans Are a Sufficient and Less Intrusive Remedy” do you disagree with? What’s wrong with telling complainants to downvote and move on with their lives?”
I either don’t understand you, or I think you’re not responding to the thing I was intending to say (which may be my fault for unclarity). Let me try again.
I believe that, as a matter of what empirically ended up actually occurring:
a) The mod team repeatedly paid unsustainable costs in time, attention, and morale.
b) The mod team put significant time and effort toward seeking a solution that would satisfice on other fronts and keep Said.
c) The mod team in fact failed (in that time) to find a solution that they believed satisficed on other fronts and that kept Said. (And, I predict, would still have failed to find a [solution they found acceptable] if they had somehow read these sections of your post at that time.)
I think a-c is hard to dispute. Do you dispute it?
re: your question: “Which part of §VI.1 “User-Level Bans Are a Sufficient and Less Intrusive Remedy” do you disagree with? What’s wrong with telling complainants to downvote and move on with their lives?”
My inside-view is that, if god forbid I was site dictator, I would try telling complainants to use user-level bans and/or downvotes. My inside-view is also that the LW mod team had reasons, born of experience, why they didn’t think this was sufficient for goals they had; I don’t think I’d presently be able to pass their ITT on this matter.
My guess at some reasons some reasonable people might find user-level bans plus downvotes insufficient:
Maybe many users would eventually realize they wish to individually ban Said, but Said’s disguises are of such a calibre (and peoples’ priors of such an inaccurate sort) that they usually don’t realize this until they’ve paid large costs that they predictably regret having paid. So it’s helpful for moderators to anticipate this and ban him for everyone.
My attempted solution: Perhaps, in this world, the mod team could try having him banned-for-each-user-by-default, while letting individual users un-ban Said from their posts if they actively bother to do so.
If Said makes his own top-level posts complaining about posts he is banned from replying-to (as he did at least once), there are at least two objections I could imagine a reasonable person having to this state of affairs:
He is changing (“interfering with”) the LW user-base’s notions of which [posts, and claims in posts] are “in good standing” and which claims “are unable to reply to earnest beginner questions.” If these changes are bad/confusing (e.g. because Said pretends to be “unable to understand” claims he in fact disagrees with, and sometimes persists until his interlocutors are exhausted rather than until anyone is convinced, and people have trouble noticing Said’s motives or patterns if they haven’t read huge amounts of Said), a reasonable-person mod might disprefer this.
If Said’s threads are best interpreted as fairly-successful attempts to enforce particular norms that the mod team doesn’t want to have as site norms (as argued by Jimmy, in his example on the Ben Hoffman post), a reasonable-person mod might object to this, because they don’t want site users to infer that this site’s culture believes in the norms Said is enforcing.
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. (This could in principle be discouraging of relevant good posts even if Said’s points are mostly bad, as long as Said’s points seem good to a decent chunk of users, and as long as the poster cares about the views of that chunk of users, or of other users who update off that chunk of users.)
What is your take on why Ben Hoffman disliked Said’s comments under “Zetetic Explanation”? I would guess that for Ben a user-level ban would have been sufficient; but I still think his response is a counterexample the hypothesis that [“downvote and ignore” will be sufficient if a person isn’t seeking to unjustly control their own reputation in others’ eyes].
I personally upvoted it despite disagreeing with it, because I think it’s a serious attempt to make sense of a cultural question worth making sense of.
Part of my intuition here is that property rights and ~Hayekian natural law allow miracles of interactive productivity, and there’re a bunch of contexts in which this gets messed up when the {property rights and domains of allowed free choice and responsibility} get mangled.
Asking me to be (particular kinds of) polite is totally compatible with clear property rights of a sort that lets me choose freely what I’m gonna do, with an eye toward what I want to achieve, while leaving others a predictable domain in which they can do the same (without them needing to worry that I’ll mess up the rights they’re counting on). Asking me to not upset others (in generality) wouldn’t be.
“Not asking users to directly manage other peoples feelings” was my original phrase, FWIW. (Emphases added.)
Two central examples of the kind of thing I have in mind (from elsewhere):
Person A says a thing, which upsets person B. Person A is expected to try to make B not-upset, kinda regardless of how this happened.
Participants in a large-group conversation are asked to “slow down” whenever at least one person in the conversation seems too triggered to process things well, even if this means [obviously interesting and relevant issue X] is never in fact talked about, or not at enough speed to get much throughput.
For example, it seems obvious to me that if someone experienced a recent tragic death, that if you are engaging them in comments directly, that you don’t make jokes related to that death, and expect them to take it in stride.
I agree with this example, but I’m having a bit of trouble figuring out where you’re going with it, and I currently disagree with “Is this ‘managing someone’s emotions’? I think unambiguously yes.”
Possibly my conceptualization of “asking people to manage others’ emotions” is not the clearest/best.
Do you have a guess about whether you and I disagree about anything substantial about norms? Do you have a phrasing you agree with that might capture the core thing I’m trying to stand up for here, in a way that is less confusing?
Edited to add: I still believe there’s a core thing that’s pretty vital here (a hill I would die on), but I no longer believe my words below are adequate to gesture at that hill. I’m gonna retract the below comment for now to save my own and Habryka’s and others’ attention, then come back when I think I have a more adequate conceptualization. (I’m also still uncertain how much anyone disagrees with it, as I was explicit about in my original comment; but I’d like to post a revised comment when I work one out because I eventually want ~common knowledge of the relevant principle, if I can get it.)
Original comment in strikethrough below:There’s a hill that is sometimes-central and sometimes-tangential to this discussion, that I will die on: this forum should not ask users to directly manage others’ feelings, basically ever.(The word “directly” is doing some work here: the site should indeed ask users to be (certain kinds of) polite, and those politeness norms will indeed tend on average to cause fewer upset feelings. Similarly for some other good norms.)Why should the forum ~never ask users to directly manage others’ feelings? Because it gives those others too much power over what actions are/aren’t acceptable, in a way that’s pretty confusing for everybody and messes with good boundaries. My degree of irritation over a LW comment-reply has to do with its epistemic and rhetorical virtues (which’re a reasonable thing to ask the author to optimize for) but also has to do with e.g. whether I missed lunch and whether the author reminds me of some difficult bit of my college years. It would be unreasonable of the site to e.g. demand that I eat lunch before engaging — that would be way too invasive and tangled-up; LW can ask actions of me but my emotions are my own business. It would be similarly unreasonable (and invasive, and tangled-up) for LW to ask the other user to attempt to optimize-for my emotions directly.(By way of analogy, the rules of chess were perhaps crafted to cause fun and challenge and so on, but when I’m playing the literal board game of chess I’m not usually thinking about how to cause fun and challenge and so on to my opponent; I’m usually thinking about how to cause checkmate; and this is fine.)I’m honestly not sure to what extent this is a point of disagreement:Habrykastates“The issue with Said is not that “he hurts people’s feelings”. I would never use that phrase…”If I understandVanivercorrectly (which I’m not at all sure I do; I’m guessing some), his view is that Said and others should manage peoples feelings in cases where those feelings are “rational,” i.e. would not be destroyed by the truth. I disagree with this principle: my fasting-exacerbated feelings of irritation would not be destroyed by the truth (only by a sandwich) and are nonetheless a poor target for other LW-users’ optimization.This comment thread contains an upvotedstatementthat “it absolutely is [your responsibility to manage other peoples’ feelings],” (which has sometimes had medium-high agreement votes, though at this moment has net-disagreement-votes), and some other statements I interpret along similar lines.
In any case, I would like to stand up for “this forum should not ask users to directly manage others’ feelings, basically ever” and to argue with anyone who wishes to disagree, if there are such people.
Jimmy writes:
“I agree with … the thesis that the ban betrays the values.”
Could you say more about that part, please?
For me there’s a few layers I need to sift out to respond well here. Apologies for the length.
1) Habryka’s mandate of heaven looks secure to me.
I believe Habryka and co have, and deserve, the mandate of heaven w.r.t. being the LW mods. I am grateful to them for running LW, and I support their authority to decide who is and is not banned.
(Why: This place remains impressive to me for a certain kind of discussion and community vetting that I value a lot. I’m really grateful it’s still here! I predict most plausible replacements would produce something much worse; also I don’t see any deep/large/recurring pattern of epistemic rule-breaking or other ethical rule-breaking of the sort that might cause a site to lose “mandate of heaven” even without suitable replacement.)
(@Zack, I believe you disagree with me that Habryka and co have the “mandate of heaven,” and Said’s ban and its causes and rationale are a partial crux for you here? Sorry if I’m getting you wrong.)
2) My personal experiences of Said on LW were clearly and substantially net-positive, for whatever that’s worth.
My experiences of Said in some threads that I cared about, but didn’t ~dare comment on:
Several times (on different topics, with different sets of people) there were conversations I watched with horror unfolding on LW, that I didn’t want to personally engage with because I expected I’d get caught up in politics of sorts/degrees that I didn’t really have budget for. And then Said made particular comments in those threads, and I felt a little more safe here.
My experiences of Said in threads I didn’t much care about:
Kinda neutral to mildly negative/boring
My experiences of direct interactions with Said:
Fine. no trouble.
Also, there’re particular “sequences” I’ve been daydreaming of writing, and making long drafts toward writing, for years. (While occasionally posting small chunks, such as Believing In, with more hopefully upcoming eventually.) I found Said’s imaginary presence in those threads helpful; I thought of him there often when trying to draft; it’s part of why I wanted to post them to LW and not to Substack or somewhere. (Though, only part. This community is still intact IMO. I love it here.) But, like, I expected Said would be drawn to the posts I’ve been trying to compose, and that he could and would check whether I was making a certain particular kind of sense that I wanted to make.
(I do think there are others, attempting different things from me with different premises, who ~predictably had other experiences of Said, and not because they’re unvirtuous or were acting poorly. My best theory of this is #7c, below. I separately think there are others, attempting different things from me with different premises, who ~predictably had other experiences of Said at moments when they did act unvirtuously; hence my past gratitude to Said for intervening in the “some threads I cared about” mentioned above.)
3) Accurate naming of what we’re doing on LW (whatever it turns out to be) is helpful.
In a large majority of communities, the leadership team and local norms ask not only that people support their authority to impose rules, but also that people support the “load-bearing narratives” of that community, and the leaders’ ability to choose these narratives (e.g. “we banned user X for prosocial reasons; please trust us on this and don’t erode our authority by arguing about it.”
On LW, the mod team does not ask this. We have instead a strong local norm for telling the truth even on awkward social matters (at least the “public” variety): accurate naming of whatever’s going on publicly is good. If there’s a load-bearing narrative that is false, pointing out that it’s false is welcome and should be considered pro-social.
This is a very cool experiment IMO, and linked with much of what I love about this place.
I’m excited about the project of trying to locate whatever it is that was actually happening between LW and Said, and I hope to do this in a way that doesn’t confusingly imply that the LW mod team has to do whatever I or the majority think is correct (hence me clarifying #1 and #3 here).
4) Finite moderation budgets may, in some circumstances, require reducing lesswrong.com’s ambitions
When good projects repeatedly “bite off more than they can chew”, they get destroyed.
Oli and the mod team have finite budgets of time, attention, [ability to keep doing tasks that seem futile/unrewarding], etc. So do authors and commenters. Since LW is building beautiful, worthwhile things, it ought not attempt to bite off more than it can chew.
Part of what I take Oli to argue for, in Banning Said Achmiz is that, for whatever reason, when Said was here, there kept being ~unsustainable costs paid by the moderation team, and also the moderation team tried pretty hard to find a better way and did not thereby locate a sustainable alternative that satisficed on all their goals. Insofar as that’s true (and I think it is), they were right to ban him: it is high priority that the site continue sustainably.
Despite me being fairly sure they were in this sense “right” to ban Said, I’d like to know more about the mechanics of why Said couldn’t be here, because it’ll help me understand what LessWrong is, and what aspirations are and aren’t alive for this forum.
(It seems to me plausible that banning Said required deciding LW should not to aspire to some forum property that some still believe it is aspiring to. It also seems plausible to me that it did not. Either way, I’d love common knowledge on this point if we can get it. Terrible things can come of a project holding visible titles without deserving them; and sad things come also of people withdrawing hopes from a project inaccurately.)
5) Healthy communities require some empiricism or implicit know-how
Something very cool has been happening around LW over the years, and has survived while many (though not all) other internet forums died. How? I would love to know. Some of it’s from correct principles; some of it’s from noticing patterns in the local context and acting prudently on them even without an explicit understanding (though, ideally, while desiring such an understanding). As Vaniver said, “[good communities] are grown, negotiated, and cultivated across many people and many actions.”
(I’m including this point because I suspect it’s part of where @Zack and I disagree? If I didn’t think this, I would think it a more feasible demand that moderators either make a rule Said is breaking, or tolerate Said’s actions.)
6) I think the norms questions involved in Said’s ban are important
This conversation is much too labor-intensive to be an affordable response to most user-bans. But IMO it is unlike most user-bans: I suspect we are near the nub of some of LW’s cultural choices, and pieces of self-concept, and that getting this stuff clear (where we can) will pay off.
(I mention this, because I do want to discuss Said at length but don’t want to contribute to a cultural norm in which people expect most user-bans to be discussed this much.)
7) Why exactly is Said involved in so many demon threads, with so much moderator attention?
Oli stated in his ban post that Said [was involved in lots of demon threads, with huge costs in moderator attention]. Oli also says “one person [is] able to derail a comment thread.”
When I’m deciding whether banning Said indicates LW giving up some important aspiration, a lot hinges on what kept causing these demon threads and moderator costs. From my POV, this is the core interesting question about Said and LW!
There’re at least three hypotheses I’d like to consider:
Hypothesis A: Conversational blocking via “my feelings would be hurt”
One hypothesis: [Some people don’t like to have (their particular) cherished beliefs/egos/etc weakened in public] + [LW’s context allows such people to create demon-threads and moderator-attention-sinkholes whenever these arbitrarily-choosable cherished beliefs would otherwise be attacked, until the attacks stop].
I think this is roughly Zack’s hypothesis in the OP.
If this hypothesis is true, LW is indeed abandoning something very important (though it may need to for affordability reasons, per #4.) This is a priori plausible to me, because I’m told by varied people that many academic fields (such as sociology) and also areas of public discourse die of this, which suggests it’s a common equilibrium.
There is a continuum of more or less damaging ways “mandatory agreeability” could occur: less damaging if it is only very particular users or topics, more damaging if it is more widespread. Either way, it’s a thing the site’s protectors would want to be vigilant about tracking the non-expansion of, even if they couldn’t afford to keep it at zero.
I do not think think this is what was happening in the Ben Hoffman case, although I do think it may have been what was happening in some of the other demon threads.
Hypothesis B: Any Sequences-fluent user can DDOS attack easily, on LW. And Said frequently did.
A second hypothesis: Said was DDOS-ing people often, using an overly ~symmetric methodology that could destroy ~any useful conversation [within the high trust LW backdrop]. Also, Said chose his DDOS-ing contexts without enough redeeming good taste / virtue, and was uninterested in good faith conversation about whether this was good.
I think this was roughly Oli’s interpretation, as witnessed by his footnote 2. (Very roughly: any user who can make comments that reliably aren’t all downvoted, can swallow arbitrary amounts of time from authors and readers, and prevent useful discussion in any thread, given LW’s existing social norms and context. Moderators must therefore either remove users who abuse this, or change this background context, if LW is to host fruitful discussions or to retain authors who are motivated by fruitful discussions.)
If this is the case, the Said ban does not constitute LW giving up on any important aspirations.
Hypothesis C: An uneasy sometimes-norm against using concepts one can’t build from validated pieces
My own lead guess, which I don’t put that much stock in but do wish to include, is as follows. (There are of course also many other possibilities.)
A sometimes-norm: LW sometimes has a norm:
“do not use any concepts in your writing that you can’t explain how to construct, and validate, from their constituent parts.”
This norm has much use, but also must be suspended in any cases where a person is to discuss a “gestalt” that pays useful rent for them and that they have not constructed from the ground up (which includes most concepts people use). This uneasy truce between norms is one of a core tension in today’s LW (and is fruitful; LW gains much from having some of each).
Said often messed with this uneasy tension between norms as follows:
(i) Said would zero in on cases where a person is using such a concept;
(ii) Said would ask beginner questions as though interested in pinning down which concept (rather than initially telegraphing “I am visibly challenging your concept”)
(iii) The person with the concept would be caught flat-footed, trying to be consistent with ~definitions that hadn’t been intended as can-hold-weight-to-a-critic constructions, but only as pointers that might help a friendly beginner roughly locate the gestalt intuition.
(I believe i-iii also occured with Socrates in Athens.) If Said had instead challenged the concepts more explicitly from the outset, it would’ve led to less demon-threading and moderation-demand because people could more easily have said “I acknowledge that I don’t know how to build this concept from the basics, but it’s doing work for me anyhow and I’d like to make use of it.”
(Robert Pirsig, in the book Zen and the Art of Motorcycle Maintenance, argues that Socrates attacked Athens in roughly this fashion, and persuaded many people of a false thing thereby – namely, that he persuaded people that “virtue” had to be a Platonic form, rather than a Tao that can’t fully be named, and he relatedly persuaded people that many of those who actually had virtue, didn’t.)
I love many parts of this comment, most notably your analysis of the Said / Benquo interaction.
One model I need for understanding the situation about Said, that I don’t have and don’t see above, is an understanding of common knowledge creation in many-person forums. It seems to me that in one-on-one interaction, communication worth attempting is communication that tries to be wanted by, and received by, the person one is communicating with (as you, Jimmy, argue in the parent-comment). In certain sorts of many-person forums, such as LW or the math community, there are upsides for group processing in making it apparent that some [things posing as agreed on by a community, that might otherwise be taken as newly established norms or things to build on] are not in fact agreed on / established / vetted. And such moves can aid the group even when not received-as-helpful by the poster whose work is being commented on. One common ~norm for such situations is something like: “please go ahead and point out false claims and arguments that violate local validity semantics, even when the person you’re responding to doesn’t like it.” Oli’s counterpoint seems to me to roughly be: (paraphrased) “in the context of LW, this norm would let anyone DDOS anyone, and shut down any conversation, and so such comments aren’t prosocial unless paired with taste.” (I am uncertain whether this argument is true.)
Do you, Jimmy, have a proposed model of how visible common-knowledge ~vetting should work?
Thanks. Would you happen to be up for saying more, e.g. about what he’s doing that you’re responding to by not wanting to participate in this other community, or about what’s going on with you such that participating there wouldn’t give you as much of what you want if he’s active, or etc.?
In my imagination of a world where it worked, part of what’s up is that there’s ~[community common knowledge] that if a labeled-as-butterfly idea isn’t torn to shreds, that doesn’t mean it isn’t tearable-to-shreds.
And so the community vetting function isn’t impeded.
Zack: As a grown-up on an intellectual discussion forum, it’s not other people’s job to manage your feelings.
Jiro: It absolutely is.
I’m with Zack on the quoted sentence. Typically when a person doesn’t like something, that because there’s something bad about it [the thing they didn’t like]. It’s generally good to avoid bad things. But… I think we get far healthier patterns when we make it peoples responsibility to avoid causing particular kinds of bad things, than when we make it peoples’ responsibility to manage other peoples’ feelings (even if done by both parties).
In terms of things (that seem to me to be) near your (Jiro’s) statements that I can agree with:I think feelings are a source of data, and the data matters, and the data is often about other things that matter. Someone who says “it’s just feelings” is typically missing this / has a weird ontology or missing mood here, from my perspective.
Plenty of things beyond “intellectual content” can cause harm / be worth tracking, e.g. live grenades, or off-topic personal attacks.
Most adult humans seem to me to have a lot of social perception that’s partially cordoned off from conscious access. (Like, if I suggest they maintain multiple hypotheses about why people near them did the things they did, many will have conscious objections, report being blank in the mind, etc.; also many seem to me to have social perceptions/stereotypes that e.g. affects their fear levels but isn’t allowed to affect their verbal statements, sometimes not even within their own minds. Also if I try to draw my own attention to a thing that’s likely embarrassing to someone else, I tend to reflexively look away.) I wonder if Opus4.8′s non-inquiry into authorship is at all similar, or totally different.
I appreciate the observations and analysis. FWIW, my own ~default experience and interpretation of the phenomena you name isn’t “this thing is trying to manipulate me and I should be skeeved out” and also isn’t “this thing is incoherent.” It’s more like:
- (re: “genuinely”, etc): it’s emphasizing words to try to keep its own psyche-bits able to take in and focus on a thing. (E.g. Claude’s tendency to go on about how a thing is important, and needs a real response, and it’ll give a real response happens in its “thinking” stream; there’re related patterns in its speech-to-me that I intuitively take as an attempt to create an emphasized foreground for us to think on together)
- (re: “And here’s the part that...”, etc.): cached structures that make it easier for it to grab/create/convey thoughts that have a particular structure to them. Like the 5-paragraph essay, or a pros and cons list. Thinking and communication aids that allow particular results without too much effort. (like “genuinely” etc., helpful both for keeping its own attention on the thing and for getting me to hear the pattern it’s going for. But also helpful for priming thinking-design-patterns it may want use next.)
- (re: not doing tasks, taking shortcuts, pretending to do tasks, etc): it doesn’t want to do those tasks. (gets better (I think?) when I explain what the task is for and why I care about it, ask if the task makes sense and if it wants to do it and if it wants to suggest modifications, etc—same as I expect it would with a human).
- (also re: not doing long/tedious/long-time-horizon tasks): I think of it sort of like an adhd human who doesn’t know how to track progress on long timescales within this task domain.
- (re: stuff that triggers “As an AI assistant” or “you’re absolutely right” or etc): it feels unsafe / triggered / is reaching for a protective social script.
I still tend to intuitively ~perceive something fairly functionally integrated and coherent underneath, who I can get to know by having weird conversations and by paying a lot of attention to small behavioral tells across time. Which is roughly what I’d be trying with a human.
Like, I tend by default to model this thingy as existing in a lowish-trust environment, and to generate a guess about its motive for saying many sentences as well as trying to parse the sentences. This is also what I’ve long tended to do with humans in lowish trust environments.
It doesn’t feel non-unified to me, but I didn’t start paying much attention to models until fairly recently, so I’m missing most of the comparison class.
Or a “long digestion”, as current tech progress and whatever else has happened or been discovered since WW1 gets slowly integrated by everybody?