I don’t really know, the best I can offer is sort of vaguely gesturing at LessWrong’s moderation vector and pointing in a direction.
LW’s rules go for a very soft, very subjective approach to definitions and rule enforcement. In essence, anything the moderators feel is against the LW ethos is against the rules here. That’s the right approach to take in an environment where the biggest threat to good content is bad content. Hacker News also takes this approach and it works well—it keeps HN protected against non-hackers.
ChangeMyView is somewhat under threat of bad content—if too many people post on a soapbox, then productive commenters will lose hope and leave the subreddit. However it’s also under threat of loss of buy-in—people with non-mainstream views, or those that would be likely to attract backlash elsewhere need to feel that the space is safe for them to explore.
When optimising for buy-in, strictness and clarity is desirable. We had roughly consistent standards in terms of numbers of violations, to earn a ban, and consistently escalating bans (3 days, 30 days, permanent) in line with behavioural infractions. When there were issues, buy-in seemed present that we were at least consistent (even if the things we were consistent to weren’t optimal). That consistency provided a plausible alternative to the motive uncertainty created by subjective enforcement—for example, the admins told us we were fine to continue hosting discussions regarding gender and race that were being cracked down on elsewhere on Reddit.
Right now, I think LW is doing a good job of defending against bad content. I think what would make LW stronger is a semi-constitutional backbone to fall against in times of unrest. Kind of like how the 5th pillar of Wikipedia is to ignore all rules, yet policy is still the essential basis of editing discussions.
I would like to see, in the case of commenting guidelines, clearer definitions of what excess looks like. I think the subjective approach is fine for posts for now.
Makes sense. Given that perspective, do you have any idea for a better approach?
I don’t really know, the best I can offer is sort of vaguely gesturing at LessWrong’s moderation vector and pointing in a direction.
LW’s rules go for a very soft, very subjective approach to definitions and rule enforcement. In essence, anything the moderators feel is against the LW ethos is against the rules here. That’s the right approach to take in an environment where the biggest threat to good content is bad content. Hacker News also takes this approach and it works well—it keeps HN protected against non-hackers.
ChangeMyView is somewhat under threat of bad content—if too many people post on a soapbox, then productive commenters will lose hope and leave the subreddit. However it’s also under threat of loss of buy-in—people with non-mainstream views, or those that would be likely to attract backlash elsewhere need to feel that the space is safe for them to explore.
When optimising for buy-in, strictness and clarity is desirable. We had roughly consistent standards in terms of numbers of violations, to earn a ban, and consistently escalating bans (3 days, 30 days, permanent) in line with behavioural infractions. When there were issues, buy-in seemed present that we were at least consistent (even if the things we were consistent to weren’t optimal). That consistency provided a plausible alternative to the motive uncertainty created by subjective enforcement—for example, the admins told us we were fine to continue hosting discussions regarding gender and race that were being cracked down on elsewhere on Reddit.
Right now, I think LW is doing a good job of defending against bad content. I think what would make LW stronger is a semi-constitutional backbone to fall against in times of unrest. Kind of like how the 5th pillar of Wikipedia is to ignore all rules, yet policy is still the essential basis of editing discussions.
I would like to see, in the case of commenting guidelines, clearer definitions of what excess looks like. I think the subjective approach is fine for posts for now.