The occasional phenomenon where people go downvote every comment by someone they disagree with could be limited by only allowing people to downvote comments made within the last week.
Or limit the number of votes one person can give to another within a time period. I think most vendetta voting happens in the heat of the moment. I don’t like not being able to vote old comments, or skewing the voting on either side.
In the only cases I’ve seen where I’ve had grounds for suspicion about who was doing the karmassassination, the person I thought was the culprit was a high-karma long-established LWer.
(But there is some bias here; such people are likely to be more salient as candidates for who the culprit is. And in no case have I been very sure who was responsible.)
I always wondered if an algorithm could be implemented akin to the Page rank algorithm. A vote from someone counts more if the person votes seldom and it counts more if the person is upvoted frequently by people with high vote weight.
The assumption is that it is that we will capture the variable of “how well do they know lesswrong” by measuring how much they are upvoted. I think the most important part is that votes by people with high karma give more karma. The best kind of upvote is one by someone who is very very popular on lesswrong because they say lots of important stuff, but almost never thinks anything is worth upvoting.
Ah. If that’s the goal, I suggest increasing the impact of votes the more upvoted someone is, and increasing the upness of votes the more often she downvotes relative to upvoting. If I’m popular and upvote a whole lot of things, that seems like a possible reason to weight my downvotes more strongly. But If I’m popular and don’t vote for much of anything at all, it’s not as clear to me why that’s a reason to take my vote more seriously than if I were equally popular but participated in the voting system more. The latter just seems to discourage popular people from voting very much.
If we want to encourage our popular people to vote more, we should increase the power of their votes the more votes they make, rather than decreasing it.
I did not know this was a thing, but I do not think this is a worthwhile fix. If a user experiences a sudden drop in karma, and a lot of −1 posts, they should be able to report the user, and a mod should be able to check and punish them and fix the problem. We do not want a fix which shows up as an inconvenience often for a problem which is only rarely a problem.
If a user experiences a sudden drop in karma, and a lot of −1 posts, they should be able to report the user, and a mod should be able to check and punish them and fix the problem.
I’ve never seen a mod capable of checking who downvoted what reacting in any way when thishascomeup.
Having been on both sides of a flash downvote (guilty!), I can tell you that these vendettas are not very effective on a forum of this size. Enough people read old comments and tend to upvote comments they otherwise wouldn’t if they feel it’s unfairly penalized, even if the comment is old. It’s a lot more effective to post a quality reply which convinces other readers that the comment in question deserves a downvote.
Hmmm. It’s true that one of the cases where I’m most likely to upvote is where I see a comment that looks fine to me that has a negative point total. But flash downvoting won’t necessarily produce any negative point totals, and I’m a lot less likely to spring into action for a post that merely doesn’t have quite as many positive points as it should (since I rarely have a very firm idea of how many positive points anything should have). Then again, perhaps in cases where nothing actually goes negative, not much harm is really being done anyway. So I may agree with you, but I’m not sure you’ve got exactly the right reason.
I guess mostly my own feeling is that the karma system seems to work pretty well as is. When I see a comment getting downvoted to oblivion, it usually seems to deserve it, and the quality of conversation around here usually seems above the internet average. I’m sure karma doesn’t precisely measure what it’s supposed to measure (whatever that is anyway), but I’m inclined to suspect that trying to make it do so is likely to end up being more trouble than it’s worth.
The occasional phenomenon where people go downvote every comment by someone they disagree with could be limited by only allowing people to downvote comments made within the last week.
Or limit the number of votes one person can give to another within a time period. I think most vendetta voting happens in the heat of the moment. I don’t like not being able to vote old comments, or skewing the voting on either side.
I like this fix. If the mass voters tend to have low karma, you could also make this a fix that only applies to people below some karma threshold.
In the only cases I’ve seen where I’ve had grounds for suspicion about who was doing the karmassassination, the person I thought was the culprit was a high-karma long-established LWer.
(But there is some bias here; such people are likely to be more salient as candidates for who the culprit is. And in no case have I been very sure who was responsible.)
I think even the best of us are susceptible to the keyboard warrior berserk mode.
I always wondered if an algorithm could be implemented akin to the Page rank algorithm. A vote from someone counts more if the person votes seldom and it counts more if the person is upvoted frequently by people with high vote weight.
Could you explain this bit? I’d expect someone who votes seldom to have lower quality votes, because ey’re likely to read less of LW.
The assumption is that it is that we will capture the variable of “how well do they know lesswrong” by measuring how much they are upvoted. I think the most important part is that votes by people with high karma give more karma. The best kind of upvote is one by someone who is very very popular on lesswrong because they say lots of important stuff, but almost never thinks anything is worth upvoting.
Ah. If that’s the goal, I suggest increasing the impact of votes the more upvoted someone is, and increasing the upness of votes the more often she downvotes relative to upvoting. If I’m popular and upvote a whole lot of things, that seems like a possible reason to weight my downvotes more strongly. But If I’m popular and don’t vote for much of anything at all, it’s not as clear to me why that’s a reason to take my vote more seriously than if I were equally popular but participated in the voting system more. The latter just seems to discourage popular people from voting very much.
If we want to encourage our popular people to vote more, we should increase the power of their votes the more votes they make, rather than decreasing it.
I did not know this was a thing, but I do not think this is a worthwhile fix. If a user experiences a sudden drop in karma, and a lot of −1 posts, they should be able to report the user, and a mod should be able to check and punish them and fix the problem. We do not want a fix which shows up as an inconvenience often for a problem which is only rarely a problem.
I’ve never seen a mod capable of checking who downvoted what reacting in any way when this has come up.
I would guess that a mod being capable of checking that would be an easier or at least not much harder fix than a time limit on voting down.
Having been on both sides of a flash downvote (guilty!), I can tell you that these vendettas are not very effective on a forum of this size. Enough people read old comments and tend to upvote comments they otherwise wouldn’t if they feel it’s unfairly penalized, even if the comment is old. It’s a lot more effective to post a quality reply which convinces other readers that the comment in question deserves a downvote.
Hmmm. It’s true that one of the cases where I’m most likely to upvote is where I see a comment that looks fine to me that has a negative point total. But flash downvoting won’t necessarily produce any negative point totals, and I’m a lot less likely to spring into action for a post that merely doesn’t have quite as many positive points as it should (since I rarely have a very firm idea of how many positive points anything should have). Then again, perhaps in cases where nothing actually goes negative, not much harm is really being done anyway. So I may agree with you, but I’m not sure you’ve got exactly the right reason.
I guess mostly my own feeling is that the karma system seems to work pretty well as is. When I see a comment getting downvoted to oblivion, it usually seems to deserve it, and the quality of conversation around here usually seems above the internet average. I’m sure karma doesn’t precisely measure what it’s supposed to measure (whatever that is anyway), but I’m inclined to suspect that trying to make it do so is likely to end up being more trouble than it’s worth.