Status note: This comment is written by me and reflects my views. I ran it past the other moderators, but they might have major disagreements with it.
I agree with a lot of Jason’s view here. The EA community is indeed much bigger than the EA Forum, and the Forum would serve its role as an online locus much less well if we used moderation action to police the epistemic practices of its participants.
I don’t actually think this that bad. I think it is a strength of the EA community that it is large enough and has sufficiently many worldviews that any central discussion space is going to be a bit of a mishmash of epistemologies.[1]
Some corresponding ways this viewpoint causes me to be reluctant to apply Habryka’s philosophy:[2]
Something like a judicial process is much more important to me. We try much harder than my read of LessWrong to apply rules consistently. We have the Forum Norms doc and our public history of cases forms something much closer to a legal code + case law than LW has. Obviously we’re far away from what would meet a judicial standard, but I view much of my work through that lens. Also notable is that all nontrivial moderation decisions get one or two moderators to second the proposal.
Related both to the epistemic diversity, and the above, I am much more reluctant to rely on my personal judgement about whether someone is a positive contributor to the discussion. I still do have those opinions, but am much more likely to use my power as a regular user to karma-vote on the content.
Some points of agreement:
Old users are owed explanations, new users are (mostly) not
Agreed. We are much more likely to make judgement calls in cases of new users. And much less likely to invest time in explaining the decision. We are still much less likely to ban new users than LessWrong. (Which, to be clear, I don’t think would have been tenable on LessWrong when they instituted their current policies, which was after the launch of GPT-4 and a giant influx of low quality content.)
I try really hard to not build an ideological echo chamber
Most of the work I do as a moderator is reading reports and recommending no official action. I have the internal experience of mostly fighting others to keep the Forum an open platform. Obviously that is a compatible experience with overmoderating the Forum into an echo chamber, but I will at least bring this up as a strong point of philosophical agreement.
Final points:
I do think we could potentially give more “near-ban” rate limits, such as the 1 comment/3 days. The main benefit of this I see is as allowing the user to write content disagreeing with their ban.
Controversial point! Maybe if everyone adopted my own epistemic practices the community would be better off. It would certainly gain in the ability to communicate smoothly with itself, and would probably spend less effort pulling in opposite directions as a result, but I think the size constraints and/or deference to authority that would be required would not be worth it.
I do think we could potentially give more “near-ban” rate limits, such as the 1 comment/3 days. The main benefit of this I see is as allowing the user to write content disagreeing with their ban.
I think the banned individual should almost always get at least one final statement to disagree with the ban after its pronouncement. Even the Romulans allowed (will allow?) that. Absent unusual circumstances, I think they—and not the mods—should get the last word, so I would also allow a single reply if the mods responded to the final statement.
More generally, I’d be interested in ~”civility probation,” under which a problematic poster could be placed for ~three months as an option they could choose as an alternative to a 2-4 week outright ban. Under civility probation, any “probation officer” (trusted non-mod users) would be empowered to remove content too close to the civility line and optionally temp-ban the user for a cooling-off period of 48 hours. The theory of impact comes from the criminology literature, which tells us that speed and certainty of sanction are more effective than severity. If the mods later determined after full deliberation that the second comment actually violated the rules in a way that crossed the action threshold, then they could activate the withheld 2-4 week ban for the first offense and/or impose a new suspension for the new one.
We are seeing more of this in the criminal system—swift but moderate “intermediate sanctions” for things like failing a drug test, as opposed to doing little about probation violations until things reach a certain threshold and then going to the judge to revoke probation and send the offender away for at least several months. As far as due process, the theory is that the offender received their due process (consideration by a judge, right to presumption of innocence overcome only by proof beyond a reasonable doubt) in the proceedings that led to the imposition of probation in the first place.
I want to throw in a bit of my philosophy here.
Status note: This comment is written by me and reflects my views. I ran it past the other moderators, but they might have major disagreements with it.
I agree with a lot of Jason’s view here. The EA community is indeed much bigger than the EA Forum, and the Forum would serve its role as an online locus much less well if we used moderation action to police the epistemic practices of its participants.
I don’t actually think this that bad. I think it is a strength of the EA community that it is large enough and has sufficiently many worldviews that any central discussion space is going to be a bit of a mishmash of epistemologies.[1]
Some corresponding ways this viewpoint causes me to be reluctant to apply Habryka’s philosophy:[2]
Something like a judicial process is much more important to me. We try much harder than my read of LessWrong to apply rules consistently. We have the Forum Norms doc and our public history of cases forms something much closer to a legal code + case law than LW has. Obviously we’re far away from what would meet a judicial standard, but I view much of my work through that lens. Also notable is that all nontrivial moderation decisions get one or two moderators to second the proposal.
Related both to the epistemic diversity, and the above, I am much more reluctant to rely on my personal judgement about whether someone is a positive contributor to the discussion. I still do have those opinions, but am much more likely to use my power as a regular user to karma-vote on the content.
Some points of agreement:
Agreed. We are much more likely to make judgement calls in cases of new users. And much less likely to invest time in explaining the decision. We are still much less likely to ban new users than LessWrong. (Which, to be clear, I don’t think would have been tenable on LessWrong when they instituted their current policies, which was after the launch of GPT-4 and a giant influx of low quality content.)
Most of the work I do as a moderator is reading reports and recommending no official action. I have the internal experience of mostly fighting others to keep the Forum an open platform. Obviously that is a compatible experience with overmoderating the Forum into an echo chamber, but I will at least bring this up as a strong point of philosophical agreement.
Final points:
I do think we could potentially give more “near-ban” rate limits, such as the 1 comment/3 days. The main benefit of this I see is as allowing the user to write content disagreeing with their ban.
Controversial point! Maybe if everyone adopted my own epistemic practices the community would be better off. It would certainly gain in the ability to communicate smoothly with itself, and would probably spend less effort pulling in opposite directions as a result, but I think the size constraints and/or deference to authority that would be required would not be worth it.
Note that Habryka has been a huge influence on me. These disagreements are what remains after his large influence on me.
I think the banned individual should almost always get at least one final statement to disagree with the ban after its pronouncement. Even the Romulans allowed (will allow?) that. Absent unusual circumstances, I think they—and not the mods—should get the last word, so I would also allow a single reply if the mods responded to the final statement.
More generally, I’d be interested in ~”civility probation,” under which a problematic poster could be placed for ~three months as an option they could choose as an alternative to a 2-4 week outright ban. Under civility probation, any “probation officer” (trusted non-mod users) would be empowered to remove content too close to the civility line and optionally temp-ban the user for a cooling-off period of 48 hours. The theory of impact comes from the criminology literature, which tells us that speed and certainty of sanction are more effective than severity. If the mods later determined after full deliberation that the second comment actually violated the rules in a way that crossed the action threshold, then they could activate the withheld 2-4 week ban for the first offense and/or impose a new suspension for the new one.
We are seeing more of this in the criminal system—swift but moderate “intermediate sanctions” for things like failing a drug test, as opposed to doing little about probation violations until things reach a certain threshold and then going to the judge to revoke probation and send the offender away for at least several months. As far as due process, the theory is that the offender received their due process (consideration by a judge, right to presumption of innocence overcome only by proof beyond a reasonable doubt) in the proceedings that led to the imposition of probation in the first place.
“will allow?”
very good.
Yeah seems fair.