I think your “go down this rabbit hole where you’re not actually trying to talk about” story could potentially fit the improv game framework, actually?
Absolutely not, no. I hope the following can help explain.
you should have more clearly flagged that you were commenting in an “adversarial” role rather than a “cooperative” one with respect to my research program
Yes, IMO this is a much more prosocial pattern and tends to go better for everyone involved. You’re saying this is “controlling”, but again, there’s a simple matter of matching intents. Like a traffic light.
Suppose this happens:
I write a big post that assumes X and goes on to discuss Y. Then you come in and say “It seems like this post assumes X; Is that right? Maybe this is a bit off topic / off of your interests, but I think that X is false, and I think this affects/invalidates your conclusions about Y in reality.”.
Now, I could be annoyed or something that I’m not getting the engagement I want, or annoyed that you’re getting more attention, or something. But I think in this case that annoyance is probably largely my problem, or possibly a problem with the local discourse [i.e. collective user behavior] (e.g. because it’s not able to allocate its cognitive resources properly). It’s a perfectly valid way for you to engage in discourse in a public forum; it’s interesting + relevant to discuss whether X is true, whether it affects my conclusions about Y, and whether my post does assume X.
However, suppose you come in and make an ambiguous comment, e.g. “just asking questions”, while in fact you mostly already have an opinion which you are for some reason not stating. Specifically, highly ambiguous about what you care about / what you’re trying to talk about / what topic you’re trying to get at. (And perhaps, further, the comment is elliptical, or seems to presume something about what I said incorrectly, or seems to signal that I “ought” to engage or have a response or something.) Well, now I might want to engage on the assumption that you’re talking about Y. There might also be attentional presumptions that I ought to engage. For example, it could be that most readers / the community broadly agrees that X is true, and doesn’t view it as that worthwhile to debate whether X is true unless there’s important new evidence. But if your comment is ambiguous, then I as the author and especially the audience would not have common knowledge that you’re not arguing against my claims about Y (presuming X).
If you signpost what you’re doing, then IMO what’s supposed to happen is this:
If your arguments are bad, you get ignored.
If your arguments are kind good / interesting, you get upvoted and discussed, but I may ignore because I believe X.
If your arguments are really good, then I’m eventually “forced” (within some ideal discourse community) to grapple with what you’re saying somehow.
This seems to be like both feasible and a much better sort of attentional traffic-light coordination. (Though plausibly I’m confused / missing important stuff, and I’m confident this is far from comprehensively addressing the overall conflict at hand.)
As a random illustration (forgive my poor memory), for something like a year or two (? not sure), I had various “debates / debate skirmishes” with various other people doing agent foundations research at MIRI (circa 2019-2021 I guess, though I’m quite fuzzy on that). This would generally take the form of me saying stuff like “this whole research program is dooomed because it is too slow / it looks at the wrong data / it’s asking the wrong questions / it’s presuming away too much of the important stuff / it’s not sufficiently grappling with our lack of the appropriate concepts / etc.” or similar, and then trying to communicate about that, mostly by waving my arms around. IDK how much of an impression it made, but anyway, eventually I realized that I and the other researchers just had “fundamentally” different background presumptions about how to approach the whole problem of alignment / research taste / how to think / what questions to ask / how important methodology is / etc.; and therefore they weren’t trying to orient to the same things I was trying to orient towards; and therefore it was somewhat of a waste of effort to invest too much (beyond more initial fun versions) in these debates, and generally stop trying to attack the core parts of the problem together as a team with them. If I had been able to more quickly figure out, and state explicitly, these fundamental positions / orientations, then I could have avoided derailing other people’s research conversations as much. (I don’t think this is a direct analogy in most respects to the topic; my intent is to exemplify the structure of “unexpressed basic difference of orientation --> wasted effort; expressed basic difference --> more efficiently parallel / independent processing” or something like that. My guess is that in the cases at hand, it would be much easier to just state the disagreement directly, though maybe that’s incorrect?)
Absolutely not, no. I hope the following can help explain.
Yes, IMO this is a much more prosocial pattern and tends to go better for everyone involved. You’re saying this is “controlling”, but again, there’s a simple matter of matching intents. Like a traffic light.
Suppose this happens:
Now, I could be annoyed or something that I’m not getting the engagement I want, or annoyed that you’re getting more attention, or something. But I think in this case that annoyance is probably largely my problem, or possibly a problem with the local discourse [i.e. collective user behavior] (e.g. because it’s not able to allocate its cognitive resources properly). It’s a perfectly valid way for you to engage in discourse in a public forum; it’s interesting + relevant to discuss whether X is true, whether it affects my conclusions about Y, and whether my post does assume X.
However, suppose you come in and make an ambiguous comment, e.g. “just asking questions”, while in fact you mostly already have an opinion which you are for some reason not stating. Specifically, highly ambiguous about what you care about / what you’re trying to talk about / what topic you’re trying to get at. (And perhaps, further, the comment is elliptical, or seems to presume something about what I said incorrectly, or seems to signal that I “ought” to engage or have a response or something.) Well, now I might want to engage on the assumption that you’re talking about Y. There might also be attentional presumptions that I ought to engage. For example, it could be that most readers / the community broadly agrees that X is true, and doesn’t view it as that worthwhile to debate whether X is true unless there’s important new evidence. But if your comment is ambiguous, then I as the author and especially the audience would not have common knowledge that you’re not arguing against my claims about Y (presuming X).
If you signpost what you’re doing, then IMO what’s supposed to happen is this:
If your arguments are bad, you get ignored.
If your arguments are kind good / interesting, you get upvoted and discussed, but I may ignore because I believe X.
If your arguments are really good, then I’m eventually “forced” (within some ideal discourse community) to grapple with what you’re saying somehow.
This seems to be like both feasible and a much better sort of attentional traffic-light coordination. (Though plausibly I’m confused / missing important stuff, and I’m confident this is far from comprehensively addressing the overall conflict at hand.)
As a random illustration (forgive my poor memory), for something like a year or two (? not sure), I had various “debates / debate skirmishes” with various other people doing agent foundations research at MIRI (circa 2019-2021 I guess, though I’m quite fuzzy on that). This would generally take the form of me saying stuff like “this whole research program is dooomed because it is too slow / it looks at the wrong data / it’s asking the wrong questions / it’s presuming away too much of the important stuff / it’s not sufficiently grappling with our lack of the appropriate concepts / etc.” or similar, and then trying to communicate about that, mostly by waving my arms around. IDK how much of an impression it made, but anyway, eventually I realized that I and the other researchers just had “fundamentally” different background presumptions about how to approach the whole problem of alignment / research taste / how to think / what questions to ask / how important methodology is / etc.; and therefore they weren’t trying to orient to the same things I was trying to orient towards; and therefore it was somewhat of a waste of effort to invest too much (beyond more initial fun versions) in these debates, and generally stop trying to attack the core parts of the problem together as a team with them. If I had been able to more quickly figure out, and state explicitly, these fundamental positions / orientations, then I could have avoided derailing other people’s research conversations as much. (I don’t think this is a direct analogy in most respects to the topic; my intent is to exemplify the structure of “unexpressed basic difference of orientation --> wasted effort; expressed basic difference --> more efficiently parallel / independent processing” or something like that. My guess is that in the cases at hand, it would be much easier to just state the disagreement directly, though maybe that’s incorrect?)