If one is to design a site aimed at lots of general users (as opposed to targeting a focused group, for example programmers), what would be a good algorithm/design for moderating discussions/comments with the explicit goal of having the users moderating themselves? The kind of comments that would be most necessary to remove would be the kind that would be labeled as racist/sexist/hateful/etc.

Looking at stackoverflow et al, there seems to be a working system, based on reputation. The question is: is this because the target group is a specially professional type of people, or would this work for any type of user (aka am I kidding myself thinking programmers are better than other people)? Additionally, this would require something to base the reputation/score on, which may, or may not, be possible in a general forum.

The traditional way would be to keep a list of known good users who are appointed as moderators, but this would not be very dynamical and self-going and would need constant meta-supervision.

Are there other, completely different approaches?

What I think I need are examples of successful, working examples of large scale low-maintenance moderation "algorithms". Do you have any, or do you have any insightful comments on my claims about this subject?

  • 1
    Though not strictly meta-talk. I think you'd get a better answer for this question on meta
    – Earlz
    Commented Jun 19, 2010 at 9:11
  • @Tim Post: What I mean with "reputation/score on, which may, or may not, be possible in a general forum" is that here on SO the point is to answer questions, which automatically gives something to rate. But in a discuss-your-experiences-with-plants there may be no obvious best answers and then what should we base the score on? Commented Jun 19, 2010 at 9:11
  • @Earlz: OK, I'll give it some time here, and if I get no answers I'll vote to move it there. (Actually, though the FAQ says I need 250 rep to vote for migrating my own questions, I see no such link or button.) Commented Jun 19, 2010 at 9:18
  • Seems like this wasn't really a meta subject after all. I got one good answer that I'll bring with me anyway! Commented Jun 22, 2010 at 8:04

1 Answer 1


Examples of large-scale self-moderating communities are Digg and Reddit. You'll have to decide for yourself how successful they are. By design, distributed moderation systems tend to shun fringe opinions and behaviours while rewarding or reinforcing opinions/behaviors which the average user looks favorably upon. This can be good or bad, depending on how you look at it. While this will likely get rid of trolls and spammers very effectively, it also has a homogenizing effect on the community and promotes group-think such that opinions or viewpoints that aren't situated well within the mainstream are likely to be censored. Additionally, because everything is focused towards the average, more knowledgeable or intelligent users may feel alienated.

A good compromise would be to implement a standard distributed moderation system like on Digg, but add an alternative weighted moderation system whereupon the moderations that individuals see are weighted based upon the social network they've built using a friend/foe system similar to Slashdot. So basically it could be weighted something like:

  1. Immediate Friends are weighted most heavily +10
  2. Friends of Friends (FoF), but whom aren't FoFoes, are +5
  3. Users with similar voting patterns as friends are +4
  4. Users with similar voting patterns as FoF are +3
  5. Users with opposite voting patterns as foes are +2
  6. Everyone else +1
  7. Users with opposite voting patterns as friends are -1
  8. Users with similar voting patterns as FoFoes are -2
  9. Users with similar voting patterns as foes are -3
  10. Friends of Foes (FoFoes) are -4
  11. Foes are -8
  • Very interesting! I hadn't even considered the large side effects you talk about. Commented Jun 19, 2010 at 11:10
  • 2
    so, first I must enumerate my list of friends and enemies.. can't see what could possibly go wrong there! <runs away> Commented Jun 20, 2010 at 3:54
  • No other answers, and this one was good. Accepting! :) Commented Jun 22, 2010 at 8:02

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .