Bram Cohen on Slashdot
“I’ve generally avoided slashdot discussion, due to the poor quality of discussion there.The other day I fiddled with it a bit, and set the value for ‘funny’ to be -4, the minimum to display to 3, and display to nested, and suddenly the quality of discussion is altogether more reasonable.
This presents a bit of a theoretical problem. I’ve long held the opinion that in trust metrics someone else’s positive cert shouldn’t be counted as a negative cert, because it’s trivially gameable, and yet here it is, exactly that technique is working quite well. I’m not sure what to make of it.”
Some discussion followed this entry:
Uke : “funny” isn’t positive or negative–it’s a dimension. (well, at least one dimension.)if you admit the possibility that opinions could be negatively correlated, then you get a lot better results from reputation systems when you reliably disagree with someone!
Bram: I might buy that if I didn’t like things which are funny. I do like things which are funny, the problem is that the vast bulk of things labelled ‘funny’ on slashdot aren’t funny at all, they’re obnoxious.
Uke : That’s why I said that it was at least one dimension. On slashdot, “funny” maps to a positive opinion for some people, and a negative one for some other people. “Funny” has no universal meaning on slashdot, just like it has no universal meaning anywhere else in the realm of human language.
My views; well sometimes what modded “funny” might not funny at all, but then there is meta moderation, which eventually eliminates bad/incompetent moderators. Isn’t it here even the wiki concept shares grounds with this. The system eventually corrects itself (or gets extinct, I must add). And obvious but obligatory, “funny” is an individual perception. You can’t lay all the reliability on the moderation system, as it’ll always have all the shortcomings of a democratic system(read anonymous moderation, in this case). IMDb‘s rating system is much more flawed in that sense, and no-one should take it as the ultimate truth, yet these scores do give a general idea of things.
Perhaps, a more reliable system would be weighted moderation, everyone who rates has a weight associated with him/her (Google page rank?), or even custom weights assigned by the reader. Would it be scalable? Yes! certainly.