Eugine’s primary karma engine was karma-mining the Rationalist Quotes page
Nah. The quotes make up <1/5th of his top-ranked comments, and you can see for yourself: load http://www.ibiblio.org/weidai/lesswrong_user.php?u=Eugine_Nier , wait for it to fetch all his comments, “sort by: points”, “hide parents”, copy-paste down to, say, his comments with +9 karma, and then look at the composition:
Of his comments ranked >= 9 points, 20⁄108 or <1/5 were on rationality quote pages. I suppose he could be getting much more karma from masses of lower-ranked comments on quotes pages, but that seems a bit unlikely and more work than I want to do at the moment.
Even a casual inspection of his comments page will reveal that he posted a lot in threads other than quote threads, that his comments were of reasonably good quality, and that they were frequently upvoted (and occasionally downvoted). I don’t think there could be any system that would have stopped him from mass downvoting people by manipulating what counts for karma, as he was basically a contributing member of comment society.
He did have a lot of good comments, but he also had a lot of very negative comments. Hypothetically a system could look for people with a very wide range of scores and flag them for deeper inspection.
Nah. The quotes make up <1/5th of his top-ranked comments, and you can see for yourself: load http://www.ibiblio.org/weidai/lesswrong_user.php?u=Eugine_Nier , wait for it to fetch all his comments, “sort by: points”, “hide parents”, copy-paste down to, say, his comments with +9 karma, and then look at the composition:
Of his comments ranked >= 9 points, 20⁄108 or <1/5 were on rationality quote pages. I suppose he could be getting much more karma from masses of lower-ranked comments on quotes pages, but that seems a bit unlikely and more work than I want to do at the moment.
gwern: Testing our hypotheses since 2009.
Thanks for the info; I was not expecting the data to show that. It does indicate that the problem will be smaller than I feared.
Even a casual inspection of his comments page will reveal that he posted a lot in threads other than quote threads, that his comments were of reasonably good quality, and that they were frequently upvoted (and occasionally downvoted). I don’t think there could be any system that would have stopped him from mass downvoting people by manipulating what counts for karma, as he was basically a contributing member of comment society.
He did have a lot of good comments, but he also had a lot of very negative comments. Hypothetically a system could look for people with a very wide range of scores and flag them for deeper inspection.