I am not able to reply to messages on Less Wrong. If you need to contact me, send email to [my first name] [at] [my full name] [dot] [net].
Said Achmiz
Karma: 17,432
When discussing rationality, I typically use the word normative to refer to what idealized Bayesian reasoners would do, often in contrast to what humans do.
Understood. However, I am not sure that I approve of this usage; and it is certainly not how I use the word (or, to a first approximation, any words) myself. My comments are, unless specified otherwise, generally intended to refer to actually-existing humans.[1]
As Schopenhauer observes, the entire concept of adversarial debate is non-normative!
Indeed, so either we take this to mean that any normative claims about how to conduct such debates are necessarily meaningless, or else we allow for a concept of normativity that is not restricted to idealized Bayesian reasoners (which, I must remind you, are not actually real things that exist). Now, I am not saying that we should not identify an ideal and try to approach it asymptotically, but surely it makes no sense to behave as if we have already reached that ideal. And until we have (which seems unlikely to happen anytime soon or possibly ever), adversarial debate is a form of epistemic inquiry we will always have with us. So there must be right and wrong ways to go about doing it.
“[N]ot demand[ing] [...] that a compelling argument be immediately accepted” is normatively correct insofar as even pretty idealized Bayesian reasoners would face computational constraints, but a “stubborn defense of one’s starting position—combined with a willingness [...] to change one’s mind later” isn’t normatively correct, because the stubbornness part comes from humans’ innate vanity rather than serving any functional purpose. You could just say, “Let me think about that and get back to you later.”
“Stubbornness” is just the refusal to immediately update. Whether it makes sense to continue defending a point, or whether it makes more sense to say “let me think about it and get back to you”, is contingent on various circumstantial aspects of the situation, the course of the discussion, etc. It does not seem to me like this point can make any substantive difference.
- ↩︎
Perhaps not necessarily endorsing the actually existing distributions of certain traits in humans, perhaps generalizing slightly to “actually-existing humans but also very similar entities, humans under small plausible modifications, etc.”, but essentially still “actual humans”, and definitely not “hypothetical idealized Bayesian reasoners, which don’t exist and who maybe (probably?) can’t exist at all”.
- ↩︎
And yet here you demand I immediately change my mind in response to reason and evidence.
I think this is an improperly narrow interpretation of the word now in the grandparent’s “I’ll take that retraction and apology now.” A retraction and apology in a few days after you’ve taken some time to cool down and reflect would be entirely in line with Schopenhauer’s advice. I await the possibility with cautious optimism.
I endorse this interpretation.
My actual position is subtler: I think Schopenhauer is correct to point out that it’s possible to concede an argument too early and that good outcomes often result from being obstinate in the heat of an argument and then reflecting at leisure later, but I think describing the obstinacy behavior as “normatively correct” is taking it way too far
I think that this position is reasonable, but wrong. On the other hand, perhaps we do not actually disagree on this point, as such, because of the next point:
that’s not what the word normative means
I disagree. Elaborating:
Suppose that we are considering some class of situations, and two possible behaviors, A and B, in such a situation; and we are discussing which is the correct behavior in a situation of the given class. It may be the case (and we may claim) that any of the following hold:
Behavior A is always correct; behavior B is never correct.
Behavior B is always correct; behavior A is never correct.
In all cases, either A or B is fine; both are acceptable, neither is wrong.
In certain situations of the given class, A is correct and B is wrong; in other situations of the given class, B is correct and A is wrong.
In certain situations of the given class, A is correct and B is wrong; in other situations of the given class, B is correct and A is wrong; in yet other situations of the given class, either A or B is fine.
In certain situations of the given class, A is correct and B is wrong; in other situations of the given class, either A or B is fine.
In certain situations of the given class, B is correct and A is wrong; in other situations of the given class, either A or B is fine.
In which of these scenarios would you assent to the claim that “A is normatively correct”?
My own position is that the answer is “all of the above except #2 and possibly #7”. (I can see a definitional argument based on #7, but I am not strongly committed to including it in the definition of “normative”.)
And yet here you demand I immediately change my mind in response to reason and evidence.
We are not talking, here, about some subtle point of philosophy, or some complicated position on the facts of some difficult and specialized subject. You made a claim about my views. I disclaimed it. Either you have some support for your claim, or it is unsubstantiated. It would seem that you have no support for your claim.
When one makes objectionable factual claims about another person, and is unable to substantiate those claims, the correct thing to do is to retract it and apologize. (This does not preclude making the claim again in the future, should it so happen that you acquire previously unavailable support for the claim! But currently, you have nothing—and indeed, less than nothing—namely, a statement from me disclaiming your characterization, and nothing from you to support it.)
If you refuse to do so, the only appropriate conclusion is that you are someone who knowingly lies about other people’s views.
You go on to praise Schopenhauer when he writes about how to have discourse, including (for example) this line:
As a rule, then, every man will insist on maintaining whatever he has said, even though for the moment he may consider it false or doubtful.
Schopenhauer was here describing human behavior, having just two sentences prior (in a section which I bolded for emphasis) characterized said behavior as “the weakness of our intellect and the perversity of our will”. To say of this merely that it is “Schopenhauer when he writes about how to have discourse” is disingenuous.
You guys have got to decide whether the position is laughable or obviously correct!
I am not a “you guys” and I reject the notion that I have to decide anything for anyone else. Zack is perfectly capable of speaking for himself, as I am capable of speaking for myself. If I endorse someone’s point, I’ll say so.
What is “normatively correct” is what I described in the section I quoted in the grandparent. I have been completely clear about this view, never wavering from it in the slightest. The idea that there is some sort of ambiguity or vaccilation here is entirely of your own false invention.
Your characterization of me as “an LLM in whose system prompt it was written that it should not be able to either agree with or understand your point” is obviously insulting and, more importantly, unambiguously and verifiably false.[1], insofar as I have agreed with people often.
you go on to link to yourself repeatedly endorsing not changing your mind in comment sections
This again is an erroneous and deceptive characterization.
The bottom line is that, once again, your claim about my views is demonstrably false, and you have no support for it whatsoever. You should retract it and apologize to me.
- ↩︎
And not just in the trivial “actually I am a biological human and not a large language model” sense.
- ↩︎
My philosophy is no more “totalizing” than that which is described in, say… the Sequences. (Or, indeed, basically any other normative view on almost any intellectual topic.) Do you consider Eliezer to have constantly been “making dominance threats” in all of his posts?
EDIT: Uh… not sure what happened here. The parent comment was deleted, and now this comment is in the middle of nowhere…?
My best guess is that you just wrote that in order to write something that reads as a definitive slap-down
It would seem that you didn’t follow the link in that text. My best guess is that you just wanted to score a point against me, and didn’t bother to check or figure out what it was that I was actually saying.
If you had, you would have read the comment that I linked to, the key section of which I will now quote:
Schopenhauer is saying that—to put it in modern terms—we do not have the capability to instantly evaluate all arguments put to us, to think in the moment through all their implications, to spot flaws, etc., and to perform exactly the correct update (or lack of update). So if we immediately admit that our interlocutor is right and we are wrong, as soon as this seems to be the case, then we can very easily be led into error!
So we don’t do that. We defend our position, as it stands at the beginning. And then, after the dispute concludes, we can consider the matter at leisure, and quite possibly change our minds.
As you can see, this is very much not “epistemically committed to not changing his mind in the face of evidence and argument”.
I’ll take that retraction and apology now, please.
I like to have conversations where we both toss back and forth 99 vaguely truthy-sounding ideas and one of them happens to be a deep insight, and the other 98 are irrelevant or verifiably false and immediately brushed under the rug. However, if I try to converse with Said like this, every comment I make is directed into an scrutinization of the 98 irrelevant/false things. In my world, if I have produced one true, interesting insight in all of this, I’ve made progress. In my model of Said’s, I have sinned 98 times.
Your model of my view bears very little resemblance to my actual view.
said also makes dominance threats and those suck
What in the world is this about…?
In response to a comment, moderator @Ben Pace describes me as:
A commenter that is epistemically committed to not changing his mind in the face of evidence and argument
I consider this to be a libelous characterization. To say that it is false is an understatement.
Ben Pace should either support this accusation with cited quotes from me (which he will be unable to do, of course), or else retract it and apologize.
I have two questions:
If you found the discussion fruitful in the end, why is that not the bottom line? (Especially if this fruitfulness involved “reasonable conclusions” being reached?)
(Here I am talking about “the bottom line” only with respect to your interaction with me directly, ignoring any effects like the benefit of a comment exchange to other commenters or to readers, etc.)
You say that you “had the feeling that to interact with [me] at all was to an invitation to be drawn into an vortex of fact-checking and quibbling”. But as we can see from the linked examples, there generally was not, in fact, any “vortex of fact-checking and quibbling”.[1] So it would seem that the “feeling” you had was false-to-fact. Do you agree with this evaluation?
- ↩︎
Indeed, in the exchange at the first link, the putative roles were reversed—you were questioning me about what I believe, etc. Of course, I have no objection to this! But it hardly serves as an example of me drawing anyone into any vortices of quibbling…
Commenter @alkjash writes:
It’s been many years since I’ve been active on LW, but while I was, Said was the source of a plurality of my unpleasant interactions on this site. Many other commenters leveraged serious criticisms of my writing, but only Said consistently ruined my day while doing so.
This would seem to be a highly dubious claim, at best.
I have looked through the entirety of @alkjash’s posting/commenting history and used the site search feature, and have found only the following interactions involving the two of us:
Obviously nothing day-ruining or even unpleasant here.
Ditto.
A couple of interactions in comments on post “New moderation tools and moderation guidelines” (concerning UX of records/traces of moderation actions)
Not really “criticism of writing”, but rather a discussion of costs/benefits of certain aspects of site design.
Technically “criticism”, I guess, but more like “agreeing with part of what was said, disagreeing with another part”. Nothing unpleasant here either.
These four cases seem to be the totality of all my interactions with @alkjash, throughout the entirety of my tenure here on Less Wrong.
So where are these “unpleasant interactions” that “ruined [your] day”…?
I do think this is less like “two people disagree” and more like “two cultures are clashing”
Sure; my point was just that it’s more like either “two people disagree” or “two cultures are clashing” than it is like “physicists are explaining Newtonian mechanics to the Time Cube guy”.
I think the following framing is different from yours but hopefully still seems valid to you
Yes, that would also be basically fine.
But some things that look similarly are gesturing at fruitful puzzles, which are too difficult for the author to solve by the time they’ve written the post, or possibly ever. This shouldn’t of course involve the author claiming to have a coherent picture already.
Of course. I wholly agree with this.
The difficulty to distinguish from worthless nonsense is already too much of a punishment
Empirically, this is clearly false. The track record of LW in the past ~8 years makes this very clear.
So to be clear would it be accurate to say that you would choose (a) rather than (b) in my previous question? Perhaps with some amendments or caveats?
Right.
I was thinking that there might be 3-5 rules that are most relevant and that would be easy to rattle off. Is that not the case?
I guess I’d have to think about it. The “rules” that are relevant to this sort of situation have always seemed to me to be both very obvious and also continuous with general principles of how to live and act, so separating them out is not easy.
It sounds like here you have rules you are following that clearly apply to this decision to post the tenth comment and you are not thinking about expected consequences. Is that correct? If not would you mind clarifying what is true?
Sure, that’s basically true. Let’s say, provisionally, that this is a reasonable description.
I would appreciate it if you could outline 1) what the rules are and 2) why you have selected them.
I’m talking about stuff like this:
I say and write things[3] because I consider those things to be true, relevant, and at least somewhat important.
Now, is that the only rule that applies to situations like this (i.e., “writing comments on a discussion forum”)? No, of course not. Many other rules apply. It’s not really reasonable to expect me to enumerate the entirety of my moral and practical views in a comment.
As for why I’ve selected the rules… it’s because I think that they’re the right ones, of course.
Like, at this point we’ve moved into “list and explain all of your opinions about morality and also about everything else”. And, man, that is definitely a “we’re gonna be here all day or possibly all year or maybe twelve years” sort of conversation.
So in this hypothetical calculation which you allude to, “the effects on Bob” (in the sense that we are discussing) should be weighted at exactly zero.
Hm. I’d like to clarify something here. This seems important.
It’s one thing to say that 1) “tough love” is good because despite being painful in the short term, it is what most benefits the person in the long term. But it is another thing to say 2) that if someone is “soft” then their experiences don’t matter.
Well, yes, those are indeed two different things. But also, neither of them are things that I’ve said, so neither of them seems relevant…?
Do you think there is something useful here, perhaps with some amendments?
I think that you’re reading things into my comments that are not the things that I wrote in those comments. I’m not sure what the source of the confusion is.
I’ll also try to ask a more concrete question here. Are you saying a) by taking the effects on Bob into account it will lead to less good consequences for society as a whole (ie. Bob + everyone else), and thus we shouldn’t take the effects of Bob into account? Or are you saying b), that the effects on Bob simply don’t matter at all?
Well, things don’t just “matter” in the abstract, they only matter to specific people. I’m sure that the effects on Bob of Bob reading my comments matter to Bob. This is fine! Indeed, it’s perfect: the effects matter to Bob, and Bob is the one who knows best what the effects are, and Bob is the one best capable of controlling the effects, so a policy of “the effects on Bob of Bob reading my comments are Bob’s to take care of” is absolutely ideal in every way.
And, yes indeed, it would be very bad for society as a whole (and relevant subsets thereof, such as “the participants in this discussion forum”) if we were to adopt the opposite policy. (Indeed, we can see that it is very bad for society, almost every time we do adopt the opposite policy.)
Like, very straightforwardly, a society that takes the position that I have described is just better than a society that takes the opposite position. That’s the rule consequentialist reasoning here.
I’m realizing that I’ve been presuming that you are at least roughly consequentialist and are trying to take actions that lead to good consequences for affected parties. Maybe that’s not true though.
“Roughly consequentialist” is a basically apt label. But as I have written a few times, act consequentialism is pretty obviously non-viable; the only reasonable way to be a consequentialist is rule consequentialism.
This makes your the reasoning you outline in your second paragraph inapplicable and inappropriate.
I describe my views on this a bit in the thread I linked earlier. Some more relevant commentary can be found in this comment (Cmd-F “I say and write things” for the relevant ~3 paragraphs, although that entire comment thread is at least partly relevant to this discussion, as it talks about consequentialism and how to implement it, etc.).
Commenter @Lukas_Gloor writes:
I feel like Said … has a personal distaste of that sort of “post that contains bits that aren’t pinned down,”
This impression is mistaken. I have no such “distaste”.
On the contrary, my comments are often aimed at helping to “pin down” those bits. Asking probing questions, asking for examples, asking authors to explain how they are using certain words, etc., is precisely the correct way to do such “pinning down”.
@Lukas_Gloor continues by saying:
… it also seemed like he wouldn’t get any closer to seeing the point of those posts or comments when it was explained in additional detail. (Or, in case he did eventually see the points, he’d rarely say thanks or acknowledged that he got it now). That’s pretty frustrating to deal with for authors and other commenters.
The unacknowledged possibility here, for any given post in this category, is that the post had no coherent point, and was in fact confused, nonsensical, simply wrong, or some combination thereof. In such a case, it is entirely correct that I should not “get closer to seeing the point”, and anyone who did “get closer to seeing the point” of such a post would be making a mistake—becoming more wrong instead of less wrong. In other words: if “there is no there there”, then “getting there” is wrong, and “not getting there” is correct.
The way that we can distinguish between this possibility, and the possibility that there is something there but it’s difficult to verbalize or to characterize coherently, is precisely via discussion, conceptual analysis, examination of intent behind word choices, examination of examples (or trying to think of examples), etc. And if we find “something there”, the same methods are the means by which we can develop and refine it.
Well, right now my comment saying what I think “top author” means to most LW readers is on +12/+4 while yours saying what you think it means to most readers is on −18/-10. LW karma is a pretty poor measure of quality, but it does give some indication of what LW readers think, no?
Don’t be ridiculous. Of course it doesn’t give any indication. My comment is that low because of two LW mods strong-downvoting it. That’s literally, precisely the reason: two strength-10 downvotes, from the mods. This says nothing about what “LW readers” think.
Almost every single one of my comments under this post has been getting strong downvotes from at least one mod. Judging what “LW readers” think on this basis is obviously absurd.
Who could possibly be disagree-voting with this comment? What does it even mean to disagree with me saying that I endorse someone’s interpretation of my own words?