And “behave as if your interlocutors [or counterparties] are also aiming for convergence on truth” borders on the contradictory: if you really believed that, you wouldn’t be here!
(That is, disagreement is disrespect; the very fact that you’re disagreeing with someone implies that you think there’s something wrong with their epistemic process, and that they think there’s something wrong with your epistemic process.
I find that this post contains a non-negligible number of … questionable leaps, and big roundings-off, and various other things that would tend more to cause people to be [unable to see what’s true] rather than to [improve their ability to see what’s true]. I’m highlighting this one as an example. Another example is the second paragraph of the post (the second bullet point in the summary).
There’s the implicit assertion that “being here” is incompatible with the described behavior; that because Zack can’t think of a way to make the two of them compatible, they simply aren’t.
There’s the explicit assertion that you must disagree with someone because you think there’s something wrong with their epistemic processes (as opposed to, say, because you think they’ve seen different evidence, or haven’t processed that evidence yet).
I suspect that people who are generally nodding along will not catch such subtle maneuvers on the author’s part, and will correspondingly become less clear in their thoughts rather than moreso.
(I note for context that, when an incomplete earlier draft of the post Zack is responding to went live by accident for ten minutes, he ignored that draft’s request that he read expansions of summary bullet points before objecting to a straw understanding of the summary, and fired off a comment beginning “This is insane,” before proceeding to do lots more of the-sort-of-move I’m objecting to above.)
as opposed to, say, because you think they’ve seen different evidence, or haven’t processed that evidence yet
Hm, I’m not immediately sure how I would rewrite the offending paragraph to make the intended meaning clearer. Would adding the word “persistent” (persistent disagreement) help, or does the whole section need an overhaul?
I hope you’ll let me try again, in accordance with the Eighth Guideline: I’m trying to paint this vision of the world where—there’s only one territory, and accurate maps of that territory should all agree, and so most of the time (relative to the space of possible disagreements), people don’t disagree about things. (You won’t see a mispricing in most prediction markets; we don’t spend time talking about whether water is wet, because I can safely assume we’ve already converged on that without having to say anything.)
I agree, of course, that oftentimes, someone will say something that seems wrong, and you might mention it, expecting that the two of you “don’t ‘actually’ disagree”—that you’ll quickly converge once you have time to share evidence or clear up trivial language differences. That’s not the kind of situation this post is trying to talk about. (If that wasn’t clear from context, maybe I’m a bad writer. Sorry, and please downvote as appropriate.)
I’m saying that in the unusual situations where people persistently disagree, our guidelines for navigating that shouldn’t explicitly encourage convergence (even though ideal rational agents would end up converging), because the process by which rational agents would converge, fundamentally doesn’t look like them trying to converge. As far as the theory of normative reasoning is concerned, doing an Bayesian update based on a human’s verbal output isn’t fundamentally different from doing an update based on a photograph: you update because you think the states of the human/photo are systematically correlated with the states of some other part of reality that you want to know about. It would be very strange to talk about being “collaborative” with a photograph, or giving the photograph two more chances to demonstrate that it’s here in good faith! That’s why the Fifth Guideline sounds so strange to my ears. That’s why I’m suspicious that it’s—I apologize in advance for this uncharitable phrasing, but I think the uncharitable phrasing is warranted to communicate the nature of the suspicion—”an etiquette guideline masquerading as a rationality guideline”.
Does that make more sense?
because Zack can’t think of a way to make the two of them compatible, they simply aren’t.
I agree, of course, that my map is not the territory. In general, when I claim that two things are incompatible, it’s always possible that I’m mistaken, that the things are compatible in the territory, and the fact that my map doesn’t represent them as being compatible means that my map is wrong.
I don’t think I understand the function of appealing to the map-territory distinction here? What is being communicated by “because Zack can’t think of a way to make the two of them compatible, they simply aren’t” that couldn’t also be said as “Zack is wrong to imply the two aren’t compatible; in fact, they are compatible”, or would you consider those equivalent?
I note for context that
Thanks for adding this context. (I wasn’t going to mention it if you weren’t!)
So, from my perspective, I didn’t ignore the draft’s request to read the expansions before objecting; I read the expansions, and didn’t think my concern was addressed by the expansions, so I wrote a comment.
As it happens, I … still don’t think the concern has been addressed? The “Bayesian reasoners aren’t trying to converge” thing seems very fundamental to me (I’m proud of that second summary bullet point!), and I still don’t know what your reply to that is.
In retrospect, given that I offended you so much (which was not the outcome I was hoping for), I definitely wish both that I had used a nicer tone, and that I had explicitly included a sentence to the effect of “I read the expansion, but I still don’t think my objection has been addressed.” (I think I would have taken more care if I were commenting on your personal Facebook wall, rather than on Less Wrong.)
I’m aware that “I’m sorry you were offended” isn’t really an apology. The reason I can’t offer you a better sincere apology, is because I … don’t particularly think I did anything wrong? (I can make more of an effort to conform with your preferred communication norms when I’m talking with you, in order to try to be on good terms with you, but that would be me trying to be sensitive to your preferences, rather than me recognizing those norms qua norms.)
To gesture at where I’m coming from (without expecting you to conform to my norms), in my culture, “This is insane” was the least interesting part of the comment, and that (in Zackistan, though not in the world of Duncans), harping on it would reflect poorly on you. In my culture, if someone like (say) Said Achmiz leaves me a comment starting, “This is insane,” followed by an intellectually substantive counterargument to the post, I don’t consider that a norm violation. I think, “Gee, sounds like Said really didn’t like my post,” and then (if I care and have time), I respond to the counterargument. I don’t think of myself as having the “right” to request that people engage with my writing in a particular way; if I think the counterargument was already addressed by something I said in the post, I’ll say, “I think I already covered this in this-and-such paragraph; does that address your objection?” I would definitely never tell a critic that they’ve failed to pass the ITT of the post they think they’re objecting to; I think passing an ITT, while desirable, is a high bar, not something you can reasonably expect of anyone before they react to a post!
Basically, to Zackistani eyes, it looks like Duncans are prone to getting unreasonably offended and shutting down productive conversations over perceived norm violations that Zackistani people just don’t recognize as enforceable norms, and instead see as part of the “cost of doing business” of having intellectually substantive discussions. It’s definitely annoying when (for example) critics seem to motivatedly misunderstand (“strawman”) your work, but in Zackistan, the culturally normative response is to just keep arguing (correct the misunderstanding; don’t stress about whether it was in “good faith”); as uncivilized and anarchic as it must seem to visitors from Duncanland, we don’t have a book of guidelines that everyone has agreed to be bound by.
Does that make sense? I really think Zackistan and the world of Duncans should be able to have friendly diplomatic relations—that there should be some way for us to cooperate despite apparently having different conceptions of what cooperation looks like.
“You don’t need to apologize for being blunt! Let me be equally blunt. The sense you’re getting is accurate: no, I am not treating you as a cooperative partner in this conversation. I think your arguments are bad, and I feel very motivated to explain the obvious counterarguments to you in public, partially for the education of third parties, and partially to raise my status at the expense of yours.”
I consider this a good faith reply. It’s certainly not a polite thing to say. But politeness is bad faith. (That’s why someone might say in response to a compliment, “Do you really mean it, or are you just being polite?”) Given that someone actually in fact thinks my arguments are bad, and actually in fact feels motivated to explain why to me in public in order to raise their status at expense of mine, I think it’s fine for them to tell me so. How would me expecting them to lie about their motives help anyone? Complying with such an expectation really would be in bad faith!
I suppose such a person would not be engaging in the “collaborative truth-seeking” that the “Basics of Rationalist Discourse” guideline list keeps talking about.
Zack supposes wrong, and this is an excellent demonstration that he cannot pass the ITT of the post he thinks he is objecting to, and is in fact objecting to a strawman of his own construction. The imagined reply is not particularly nice, and is not the sort of comment I would tend to write when there are other, better ways to convey the same accurate information, but it doesn’t break any of the guidelines listed (and note also that the guidelines are guidelines, not rules, and that they are explicitly described as only being an 80⁄20 of good discourse anyway).
Wait, sorry, on further thought, I don’t think I understand why this doesn’t break the Fifth Guideline (“[...] behave as if your interlocutors are also aiming for convergence on truth”)?
I understand that guidelines are not rules, and that you wrote an additional 900 words explaining the shorthand summary of the Fifth Guideline. But if the Fifth Guideline isn’t trying to get people to not say this kind of thing, then I’m not sure what the Fifth Guideline is saying? Is my hypothetical mean person in the clear because he’s sticking to the object-level (“your arguments are bad”) and not explicitly making any claims about his interlocutor’s motivations (e.g., “you’re here in bad faith”)?
Thanks for clarifying! I agree that I’m not yet passing your ITT. (The post itself explicitly says that I’m not sure I understand how you use some words, so this shouldn’t be surprising.) I don’t think passing an ITT is or should be a prerequisite for replying to a post (although passing is definitely desirable).
I find that this post contains a non-negligible number of … questionable leaps, and big roundings-off, and various other things that would tend more to cause people to be [unable to see what’s true] rather than to [improve their ability to see what’s true]. I’m highlighting this one as an example. Another example is the second paragraph of the post (the second bullet point in the summary).
There’s the implicit assertion that “being here” is incompatible with the described behavior; that because Zack can’t think of a way to make the two of them compatible, they simply aren’t.
There’s the explicit assertion that you must disagree with someone because you think there’s something wrong with their epistemic processes (as opposed to, say, because you think they’ve seen different evidence, or haven’t processed that evidence yet).
I suspect that people who are generally nodding along will not catch such subtle maneuvers on the author’s part, and will correspondingly become less clear in their thoughts rather than moreso.
(I note for context that, when an incomplete earlier draft of the post Zack is responding to went live by accident for ten minutes, he ignored that draft’s request that he read expansions of summary bullet points before objecting to a straw understanding of the summary, and fired off a comment beginning “This is insane,” before proceeding to do lots more of the-sort-of-move I’m objecting to above.)
Thanks for commmenting!
Hm, I’m not immediately sure how I would rewrite the offending paragraph to make the intended meaning clearer. Would adding the word “persistent” (persistent disagreement) help, or does the whole section need an overhaul?
I hope you’ll let me try again, in accordance with the Eighth Guideline: I’m trying to paint this vision of the world where—there’s only one territory, and accurate maps of that territory should all agree, and so most of the time (relative to the space of possible disagreements), people don’t disagree about things. (You won’t see a mispricing in most prediction markets; we don’t spend time talking about whether water is wet, because I can safely assume we’ve already converged on that without having to say anything.)
I agree, of course, that oftentimes, someone will say something that seems wrong, and you might mention it, expecting that the two of you “don’t ‘actually’ disagree”—that you’ll quickly converge once you have time to share evidence or clear up trivial language differences. That’s not the kind of situation this post is trying to talk about. (If that wasn’t clear from context, maybe I’m a bad writer. Sorry, and please downvote as appropriate.)
I’m saying that in the unusual situations where people persistently disagree, our guidelines for navigating that shouldn’t explicitly encourage convergence (even though ideal rational agents would end up converging), because the process by which rational agents would converge, fundamentally doesn’t look like them trying to converge. As far as the theory of normative reasoning is concerned, doing an Bayesian update based on a human’s verbal output isn’t fundamentally different from doing an update based on a photograph: you update because you think the states of the human/photo are systematically correlated with the states of some other part of reality that you want to know about. It would be very strange to talk about being “collaborative” with a photograph, or giving the photograph two more chances to demonstrate that it’s here in good faith! That’s why the Fifth Guideline sounds so strange to my ears. That’s why I’m suspicious that it’s—I apologize in advance for this uncharitable phrasing, but I think the uncharitable phrasing is warranted to communicate the nature of the suspicion—”an etiquette guideline masquerading as a rationality guideline”.
Does that make more sense?
I agree, of course, that my map is not the territory. In general, when I claim that two things are incompatible, it’s always possible that I’m mistaken, that the things are compatible in the territory, and the fact that my map doesn’t represent them as being compatible means that my map is wrong.
I don’t think I understand the function of appealing to the map-territory distinction here? What is being communicated by “because Zack can’t think of a way to make the two of them compatible, they simply aren’t” that couldn’t also be said as “Zack is wrong to imply the two aren’t compatible; in fact, they are compatible”, or would you consider those equivalent?
Thanks for adding this context. (I wasn’t going to mention it if you weren’t!)
So, from my perspective, I didn’t ignore the draft’s request to read the expansions before objecting; I read the expansions, and didn’t think my concern was addressed by the expansions, so I wrote a comment.
As it happens, I … still don’t think the concern has been addressed? The “Bayesian reasoners aren’t trying to converge” thing seems very fundamental to me (I’m proud of that second summary bullet point!), and I still don’t know what your reply to that is.
In retrospect, given that I offended you so much (which was not the outcome I was hoping for), I definitely wish both that I had used a nicer tone, and that I had explicitly included a sentence to the effect of “I read the expansion, but I still don’t think my objection has been addressed.” (I think I would have taken more care if I were commenting on your personal Facebook wall, rather than on Less Wrong.)
I’m aware that “I’m sorry you were offended” isn’t really an apology. The reason I can’t offer you a better sincere apology, is because I … don’t particularly think I did anything wrong? (I can make more of an effort to conform with your preferred communication norms when I’m talking with you, in order to try to be on good terms with you, but that would be me trying to be sensitive to your preferences, rather than me recognizing those norms qua norms.)
To gesture at where I’m coming from (without expecting you to conform to my norms), in my culture, “This is insane” was the least interesting part of the comment, and that (in Zackistan, though not in the world of Duncans), harping on it would reflect poorly on you. In my culture, if someone like (say) Said Achmiz leaves me a comment starting, “This is insane,” followed by an intellectually substantive counterargument to the post, I don’t consider that a norm violation. I think, “Gee, sounds like Said really didn’t like my post,” and then (if I care and have time), I respond to the counterargument. I don’t think of myself as having the “right” to request that people engage with my writing in a particular way; if I think the counterargument was already addressed by something I said in the post, I’ll say, “I think I already covered this in this-and-such paragraph; does that address your objection?” I would definitely never tell a critic that they’ve failed to pass the ITT of the post they think they’re objecting to; I think passing an ITT, while desirable, is a high bar, not something you can reasonably expect of anyone before they react to a post!
Basically, to Zackistani eyes, it looks like Duncans are prone to getting unreasonably offended and shutting down productive conversations over perceived norm violations that Zackistani people just don’t recognize as enforceable norms, and instead see as part of the “cost of doing business” of having intellectually substantive discussions. It’s definitely annoying when (for example) critics seem to motivatedly misunderstand (“strawman”) your work, but in Zackistan, the culturally normative response is to just keep arguing (correct the misunderstanding; don’t stress about whether it was in “good faith”); as uncivilized and anarchic as it must seem to visitors from Duncanland, we don’t have a book of guidelines that everyone has agreed to be bound by.
Does that make sense? I really think Zackistan and the world of Duncans should be able to have friendly diplomatic relations—that there should be some way for us to cooperate despite apparently having different conceptions of what cooperation looks like.
Separately:
Zack supposes wrong, and this is an excellent demonstration that he cannot pass the ITT of the post he thinks he is objecting to, and is in fact objecting to a strawman of his own construction. The imagined reply is not particularly nice, and is not the sort of comment I would tend to write when there are other, better ways to convey the same accurate information, but it doesn’t break any of the guidelines listed (and note also that the guidelines are guidelines, not rules, and that they are explicitly described as only being an 80⁄20 of good discourse anyway).
Wait, sorry, on further thought, I don’t think I understand why this doesn’t break the Fifth Guideline (“[...] behave as if your interlocutors are also aiming for convergence on truth”)?
I understand that guidelines are not rules, and that you wrote an additional 900 words explaining the shorthand summary of the Fifth Guideline. But if the Fifth Guideline isn’t trying to get people to not say this kind of thing, then I’m not sure what the Fifth Guideline is saying? Is my hypothetical mean person in the clear because he’s sticking to the object-level (“your arguments are bad”) and not explicitly making any claims about his interlocutor’s motivations (e.g., “you’re here in bad faith”)?
Thanks for clarifying! I agree that I’m not yet passing your ITT. (The post itself explicitly says that I’m not sure I understand how you use some words, so this shouldn’t be surprising.) I don’t think passing an ITT is or should be a prerequisite for replying to a post (although passing is definitely desirable).