Does anyone have heuristics for when it’s worthwhile to upvote, or downvote, a post? I’ve had an account on Less Wrong for a while now, but it’s only recently that I’ve started using it on more than a weekly basis, so I suspect I’ll be engaging with this online community more. So, I’m wondering what is the up-and-up on, i.e., courteous method of, upvoting/downvoting. I’m aware that this might be a controversial issue, so let’s not use this thread for debates. I’m only looking for useful, or appropriate, heuristics for (understanding) voting I might have missed. For the record, as of this comment, I’ve never downvoted anyone.
This is what I’ve surmised so far:
Users downvote posts or comments which are about signaling value of their particular monkey tribe. This often seems to be newcomers, or people who don’t interact with the Less Wrong community very communally, bragging about who they identify their in-group as. They state things like “I’ve finally found a community committed to reason. Incidentally, this ideology is totally reasonable, so you should get on board with it. Trust me, I’ve read lots of stuff about it, so it checks out. It is not unlike [my ideological opponents], who are unreasonable/stupid/crazy/whatever. I hope you guys aren’t like [my ideological opponents], because then you’re unreasonable, too”.
Users who, in one way or another, are ignorant of topics the Less Wrong community believes they’ve already reached a consensus conclusion on in a straightforward, slam-dunk manner, receive downvotes. These types of posts which seem to have an agenda which the Less Wrong community would also find disagreeable seem to be less well-received. Ignorant posts where the submitter seems to be genuinely trying to start, or add to, a conversation in good spirit still get downvoted, but also tend to have comment which attempt to helpfully correct the submitter.
How Less Wrong as a community which polices itself by dishing out downvotes, it works efficiently a majority of the time. By the time I get to wreckage of a flame war to catch the juicy details, there isn’t much point to myself as an individual actor dishing out further downvotes.
I upvote a comment on at least one of two bases. The first basis is if I believe the comment provides information which answers a question, or clarifies a problem I have. Partial answers and solutions also work as well. This is a proxy for my interlocutor increasing the epistemic quality of the conversation. The second basis is if I believe the comment of my interlocutor provides information which is instrumentally valuable. This is a proxy for instrumental rationality. I also do this for comments in conversations I’m not a part of. If I perceived an inverse of either of the two cases I’ve presented occurring, I would consider that grounds for downvoting the comment in question.
I’m not confident with how to proceed in upvoting posts that already have lots of karma. By the time a post of decent quality already has several upvotes by the time I read it, I tend not to upvote it, so as not to give it undue importance. If I believe a post, or comment, is exceptionally well-written, or -executed, I might upvote it regardless of however many votes it has now to increase its visibility.
I’m sometimes worried about my votes being biased in the sense that they go to posts, or comments, which increase the visibility of things I only value personally, rather than being reflective of how much a given post, or comment, increases, or decreases, the quality of the discourse on Less Wrong. I’m especially worried these biases in my voting patterns might be, or could become, unconscious.
So, have I missed anything? Additionally, what are the reasons for, or against, keeping my own record of liked/upvoted, and disliked/downvoted, posts, and comments, hidden?
There is a lot of noise in voting, so don’t overanalyze it. There is a correlation between good comments and upvotes, but unless you get at least −3 on a comment, or perhaps −1 on 5 comments in a row, you should probably just ignore it. Also, upvotes usually mean you did something right, but of course a comment made early in the debate gets more visibility and votes than a comment made late in the debate.
Generally, upvotes and downvotes mean “want more of this” and “want less of this”. This is a community about rationality, so you should consider whether the given way of communicating contributes to rationality, or more specifically to building a good rationalist community. Use your own judgement.
Going against the majority opinion… I’d guess it depends on whether the argument brings something new to the discussion. Saying: “you are all wrong because you didn’t consider X” (where X is something that makes sense and really wasn’t mentioned on LW) will probably be welcome; saying “you are all wrong, because this is against my beliefs / against majority opinion” will not. But here I would expect even more noise than usual.
By the time I get to wreckage of a flame war to catch the juicy details, there isn’t much point to myself as an individual actor dishing out further downvotes.
I’m not confident with how to proceed in upvoting posts that already have lots of karma.
I think you should upvote or downvote comments regardless of the karma they already have. If 10 times more people think a comment is awesome (or horrible), it should get 10 times more upvotes (or downvotes). If you worry about someone getting infinitely many downvotes for participating in one stupid flamewar—they have an option to retract the comments, which stops the voting.
The intuition behind this is approximately that if voting is better than not voting (which based on my experience with web fora I consider obvious) then more voting is better than less voting. Also, some fraction of users will abuse the voting system, so we need more votes from the nice users to balance this. Actually, this is also a reason why the nice users should vote more often.
I’m sometimes worried about my votes being biased in the sense that they go to posts, or comments, which increase the visibility of things I only value personally, rather than being reflective of how much a given post, or comment, increases, or decreases, the quality of the discourse on Less Wrong.
You are a human and to some degree this is inevitable. You should try to do as well as you can, but don’t try to reverse stupidity and extract signal from noise. Let’s assume that 20% of your votes are biased, but 80% correctly estimate what improves the discourse. How could you improve this ratio by voting less? You can’t; because the assumption was that you don’t know which votes belong to the 20%. Voting less frequently is equivalent to giving your votes a multiplier 0.5 or 0.1 or maybe 0.01. If you have a reason to believe that you are significantly more biased than an average voter (which is not the same as an average commenter), you should abstain from voting completely; otherwise there is no reason to tune down your voting. The mere fact that you care about this all is an evidence that you should vote.
I think you should upvote or downvote comments regardless of the karma they already have.
I’m not convinced.
Should an unambiguously baddish comment really attract infinitely many downvotes, and an unambiguously goodish one infinitely many upvotes? -- Where “infinitely many” means “potentially a lot, with the number depending on how many people see the comment rather than on its quality”.
Suppose (perhaps more realistically than my use of the word “infinitely” above would suggest) a typical comment is voted on only a smallish number of times. If a smallish fraction of voters are crazy or stupid or evil, then with “independent voting” (on average) every comment gets a smallish amount of noise in its score, which in practice means that a few comments get scored completely wrong. Whereas with “vote towards what you think is the right score for this comment”, provided most of the most recent users to see a comment are sane its score should be sane.
My own practice, for what it’s worth, is somewhere intermediate between “vote according to merit, ignoring existing score” and “vote towards what seems like the score merited by the comment’s quality”, nearer the latter than the former.
I’m bad at predicting how my comments get rated, too. However, funny comments seem to get a lot of upvotes.
I suspect most people are voting on the basis of what they do or don’t like, but the community has good enough taste that it works out to a useful feedback system—or maybe it’s just a self-reinforcing hall of mirrors which suits my taste.
Does anyone have heuristics for when it’s worthwhile to upvote, or downvote, a post? I’ve had an account on Less Wrong for a while now, but it’s only recently that I’ve started using it on more than a weekly basis, so I suspect I’ll be engaging with this online community more. So, I’m wondering what is the up-and-up on, i.e., courteous method of, upvoting/downvoting. I’m aware that this might be a controversial issue, so let’s not use this thread for debates. I’m only looking for useful, or appropriate, heuristics for (understanding) voting I might have missed. For the record, as of this comment, I’ve never downvoted anyone.
This is what I’ve surmised so far:
Users downvote posts or comments which are about signaling value of their particular monkey tribe. This often seems to be newcomers, or people who don’t interact with the Less Wrong community very communally, bragging about who they identify their in-group as. They state things like “I’ve finally found a community committed to reason. Incidentally, this ideology is totally reasonable, so you should get on board with it. Trust me, I’ve read lots of stuff about it, so it checks out. It is not unlike [my ideological opponents], who are unreasonable/stupid/crazy/whatever. I hope you guys aren’t like [my ideological opponents], because then you’re unreasonable, too”.
Users who, in one way or another, are ignorant of topics the Less Wrong community believes they’ve already reached a consensus conclusion on in a straightforward, slam-dunk manner, receive downvotes. These types of posts which seem to have an agenda which the Less Wrong community would also find disagreeable seem to be less well-received. Ignorant posts where the submitter seems to be genuinely trying to start, or add to, a conversation in good spirit still get downvoted, but also tend to have comment which attempt to helpfully correct the submitter.
Posts, or comments, which are seen as trolling are downvoted. Posts, or comments, which take a meta-contrarian/intellectual-hipster stance, or go against the grain of the majority/plurality opinion(s) on Less Wrong will be volatile, but tend to get more downvotes. A recent post on life-extension and death is an example. An exception to this tendency is if the post, or comment, in question is executed very well.
How Less Wrong as a community which polices itself by dishing out downvotes, it works efficiently a majority of the time. By the time I get to wreckage of a flame war to catch the juicy details, there isn’t much point to myself as an individual actor dishing out further downvotes.
I upvote a comment on at least one of two bases. The first basis is if I believe the comment provides information which answers a question, or clarifies a problem I have. Partial answers and solutions also work as well. This is a proxy for my interlocutor increasing the epistemic quality of the conversation. The second basis is if I believe the comment of my interlocutor provides information which is instrumentally valuable. This is a proxy for instrumental rationality. I also do this for comments in conversations I’m not a part of. If I perceived an inverse of either of the two cases I’ve presented occurring, I would consider that grounds for downvoting the comment in question.
I’m not confident with how to proceed in upvoting posts that already have lots of karma. By the time a post of decent quality already has several upvotes by the time I read it, I tend not to upvote it, so as not to give it undue importance. If I believe a post, or comment, is exceptionally well-written, or -executed, I might upvote it regardless of however many votes it has now to increase its visibility.
I’m sometimes worried about my votes being biased in the sense that they go to posts, or comments, which increase the visibility of things I only value personally, rather than being reflective of how much a given post, or comment, increases, or decreases, the quality of the discourse on Less Wrong. I’m especially worried these biases in my voting patterns might be, or could become, unconscious.
So, have I missed anything? Additionally, what are the reasons for, or against, keeping my own record of liked/upvoted, and disliked/downvoted, posts, and comments, hidden?
There is a lot of noise in voting, so don’t overanalyze it. There is a correlation between good comments and upvotes, but unless you get at least −3 on a comment, or perhaps −1 on 5 comments in a row, you should probably just ignore it. Also, upvotes usually mean you did something right, but of course a comment made early in the debate gets more visibility and votes than a comment made late in the debate.
Generally, upvotes and downvotes mean “want more of this” and “want less of this”. This is a community about rationality, so you should consider whether the given way of communicating contributes to rationality, or more specifically to building a good rationalist community. Use your own judgement.
Going against the majority opinion… I’d guess it depends on whether the argument brings something new to the discussion. Saying: “you are all wrong because you didn’t consider X” (where X is something that makes sense and really wasn’t mentioned on LW) will probably be welcome; saying “you are all wrong, because this is against my beliefs / against majority opinion” will not. But here I would expect even more noise than usual.
I think you should upvote or downvote comments regardless of the karma they already have. If 10 times more people think a comment is awesome (or horrible), it should get 10 times more upvotes (or downvotes). If you worry about someone getting infinitely many downvotes for participating in one stupid flamewar—they have an option to retract the comments, which stops the voting.
The intuition behind this is approximately that if voting is better than not voting (which based on my experience with web fora I consider obvious) then more voting is better than less voting. Also, some fraction of users will abuse the voting system, so we need more votes from the nice users to balance this. Actually, this is also a reason why the nice users should vote more often.
You are a human and to some degree this is inevitable. You should try to do as well as you can, but don’t try to reverse stupidity and extract signal from noise. Let’s assume that 20% of your votes are biased, but 80% correctly estimate what improves the discourse. How could you improve this ratio by voting less? You can’t; because the assumption was that you don’t know which votes belong to the 20%. Voting less frequently is equivalent to giving your votes a multiplier 0.5 or 0.1 or maybe 0.01. If you have a reason to believe that you are significantly more biased than an average voter (which is not the same as an average commenter), you should abstain from voting completely; otherwise there is no reason to tune down your voting. The mere fact that you care about this all is an evidence that you should vote.
I’m not convinced.
Should an unambiguously baddish comment really attract infinitely many downvotes, and an unambiguously goodish one infinitely many upvotes? -- Where “infinitely many” means “potentially a lot, with the number depending on how many people see the comment rather than on its quality”.
Suppose (perhaps more realistically than my use of the word “infinitely” above would suggest) a typical comment is voted on only a smallish number of times. If a smallish fraction of voters are crazy or stupid or evil, then with “independent voting” (on average) every comment gets a smallish amount of noise in its score, which in practice means that a few comments get scored completely wrong. Whereas with “vote towards what you think is the right score for this comment”, provided most of the most recent users to see a comment are sane its score should be sane.
My own practice, for what it’s worth, is somewhere intermediate between “vote according to merit, ignoring existing score” and “vote towards what seems like the score merited by the comment’s quality”, nearer the latter than the former.
Noted. I will update my voting behavior on this basis. Thanks.
I can never predict how my comments would be rated, so I gave up on looking for voting criteria and do what feels right at the moment.
I’m bad at predicting how my comments get rated, too. However, funny comments seem to get a lot of upvotes.
I suspect most people are voting on the basis of what they do or don’t like, but the community has good enough taste that it works out to a useful feedback system—or maybe it’s just a self-reinforcing hall of mirrors which suits my taste.