What is it that causes you to believe a commenter should be penalized 20 or 30 karma points at a time? If it is that they make a lot of worthless comments, then you have no shortage of comments to down vote, and there is no need to down vote their comments indiscriminately. If it is the they made an exceptionally worthless comment, it is my experience that these get down vote pretty fast by many people, so they will still lose a lot of karma even though you only contribute one point to their loss.
In short, I don’t see what you gain by this strategy that justifies the decrease in correlation between a comments displayed karma score and the value the community assigns it that occurs when you down vote a comment not because it is a problem, but because the author had written other comments that are a problem. If a normally good contributor has a bad day and makes some bad comments, it does not make sense to devalue their previous high quality comments.
I concur; the points-values attached to individual comments have a larger impact on what LW-readers see than do users’ karma values, and are therefore more important to retain as accurate indicators of comment quality.
If a particular user has a pattern of making comments that impair LW in a particular way, you might explicitly comment on this, rhollerith, with detailed, concrete language describing what the pattern is, what specific comments fit that pattern, and why it may impair LW conversation. You could do this by public comment or private message. This has the following advantages over blanket user-downvoting:
It does not impair quality-indicators on the user’s other comments;
The user can understand where you are coming from, and so can integrate information instead of just finding it unfair;
It publicly states community norms (in the public-message version), and so may help others of us retool our comments in more useful ways as well (as well as making us less likely to feel there are random invisible grudges disrupting LW karma);
If you are mistaken about what is and is not useful, others can respond by explicitly sharing conflicting impressions.
ETA: My comment here was slightly mis-directed, in that Hollerith above said he would send the user a message explaining his reasoning.
JGWeissman writes, “I don’t see what you gain by this strategy that justifies the decrease in correlation between a comments displayed karma score and the value the community assigns it that occurs when you down vote a comment not because it is a problem, but because the author had written other comments that are a problem.”
Vladimir Nesov writes, “If you are downvoting indiscriminately, not separating the better comments from the worse ones, without even bothering to understand them, you are abusing the system.”
Anna writes, “This has the following advantages over blanket user-downvoting: . . . It does not impair quality-indicators on the user’s other comments”
The objection is valid. I retract my proposal and will say so in an addendum to my original comment.
The problem with my proposal is the part where the voter goes to a commenter’s lesswrong.com/user/ page and votes down 20 or 30 or so comments in a row. That dilutes or cancels out useful information, namely, votes from those who used the system the way it was intended.
If there were a way for a voter to reduce the karma of a person without reducing the point-score of any substantive comment, then my proposal might still have value, but without that, my proposal will have a destructive effect on the community, so of course I withdraw my proposal.
the points-values attached to individual comments have a larger impact on what LW-readers see than do users’ karma value
True, but if the behavior of the voters change so that the former becomes less informative, the site will tend to change so that the user’s karma will come to have a larger impact. In competent software development, changes in people’s behavior will cause maor changes in the software more often than changes in the software will cause major changes in the behavior of people. (Consequently, assuming the software developers are competent, most changes to the system are best initiated as changes to behavior rather than changes to the software—and if the software developers are not competent, then the site is probably doomed anyway.) Or so it seems to me.
A normally good contributor’s having a bad day is not going to be enough to trigger any downvoting of any of his comments under the policy I contemplate. Th policy I contemplate makes use of a general skill that I hypothesize that most participants on this site have: the ability to reserve judgement on someone till one has seen at least a dozen communications from that person and then to make a determination as to whether the person is worth continuing to pay attention to.
The people who have the most to contribute to a site like this are very busy. As Eliezer has written recently on this site, all that is need for this site to die is for these busy people to get discouraged because they see that the contributions of the worthwhile people are difficult to find among the contributions of the people who are not worth reading—and I stress that the people who are not worth reading often have a lot of free time which they use to generate many contributions.
Well, the voting is supposed to be the main way that the worthwhile contributions “float to the top” or float to where people are more likely to see them than to see the mediocre contributions. But that only works if the people who can distinguish a worthwhile contribution from a mediocre contribution bother to vote. So let us consider whether they do. For example, has Patri Friedman or Shane Legg bothered to vote? They both have made a few comments here. But they are both very busy people. I’ll send them both emails, referencing this conversation and asking them if they remember actually voting on comments here, and report back to y’all. (Eliezer is not a good person to ask in this regard because he has a big investment in the idea that a social web site based on voting will win, so of course he has been voting on the contributions here.)
The highest-scoring comment I know of is Shane Legg’s description of an anti-procrastination technique, which currently has 16 points. But there are thousands of readers of this site. Now it is possible that a lot more readers of Shane’s comment would have voted it up if it did not already have a high score, but I humbly suggest that it is more likely that only one or two or three percent of the readers of a comment would have bothered to vote on the comment regardless of its score.
Whether this site lives or dies seems to depend on the frequency with which the people who can tell a worthwhile comment from a non-worthwhile comment bother to vote. But like I said, these people tend to be very busy.
Hence my suggestion of adopting a policy of voting on commenters rather than coment—because that is going to save some of the busy person’s time.
There is a strong ethic in American society (and probably in other societies) that it is contributions and not individuals that should be judged. Well, I humbly suggest that since being able to contribute comments and posts here is not a basic human need, like housing or education or the opportunity to compete on an equal footing with other workers for income, the application of that admirable ethic to the decision of who gets to comment and post here is not worth the risk of this site’s going downhill to the point where the people who could have carried the site decide it is not worth the time out of their busy lives.
EDIT. If no other participants on this site declare their intention to use commenter-based voting, then I probably will not use commenter-based voting either because of what the economists call network effects. The only reason I suggested it in the first place is that conchis’s comment is not the first time someone here has indicated that voters other than me are already using commenter-based voting.
The policy I contemplate makes use of a general skill that I hypothesize that most participants on this site have: the ability to reserve judgment on someone till one has seen at least a dozen communications from that person and then to make a determination as to whether the person is worth continuing to pay attention to.
If you are basing your judgment on at least a dozen communications from the commenter, then, as I explained, you have already identified plenty of comments that should be downloaded. If you base your decision on seeing the 10 of the 12 observed comments are problems, then you can dock the bad commenter 10 points. And if you are right, you will not be the only one. You do not need to personally dock the user 20 or 30 points.
Whether this site lives or dies seems to depend on the frequency with which the people who can tell a worthwhile comment from a non-worthwhile comment bother to vote. But like I said, these people tend to be very busy.
If a person has excellent judgment to distinguish which comments should be up voted and which person should be down voted, but does not have the time to actually use that judgment, then that person is not going to be a successful protector of this site. Either take the time to do it right, or leave it to those who have the good judgment and the time, who there seems to be plenty of, giving that the system is working.
What is it that causes you to believe a commenter should be penalized 20 or 30 karma points at a time? If it is that they make a lot of worthless comments, then you have no shortage of comments to down vote, and there is no need to down vote their comments indiscriminately. If it is the they made an exceptionally worthless comment, it is my experience that these get down vote pretty fast by many people, so they will still lose a lot of karma even though you only contribute one point to their loss.
In short, I don’t see what you gain by this strategy that justifies the decrease in correlation between a comments displayed karma score and the value the community assigns it that occurs when you down vote a comment not because it is a problem, but because the author had written other comments that are a problem. If a normally good contributor has a bad day and makes some bad comments, it does not make sense to devalue their previous high quality comments.
I concur; the points-values attached to individual comments have a larger impact on what LW-readers see than do users’ karma values, and are therefore more important to retain as accurate indicators of comment quality.
If a particular user has a pattern of making comments that impair LW in a particular way, you might explicitly comment on this, rhollerith, with detailed, concrete language describing what the pattern is, what specific comments fit that pattern, and why it may impair LW conversation. You could do this by public comment or private message. This has the following advantages over blanket user-downvoting:
It does not impair quality-indicators on the user’s other comments;
The user can understand where you are coming from, and so can integrate information instead of just finding it unfair;
It publicly states community norms (in the public-message version), and so may help others of us retool our comments in more useful ways as well (as well as making us less likely to feel there are random invisible grudges disrupting LW karma);
If you are mistaken about what is and is not useful, others can respond by explicitly sharing conflicting impressions.
ETA: My comment here was slightly mis-directed, in that Hollerith above said he would send the user a message explaining his reasoning.
JGWeissman writes, “I don’t see what you gain by this strategy that justifies the decrease in correlation between a comments displayed karma score and the value the community assigns it that occurs when you down vote a comment not because it is a problem, but because the author had written other comments that are a problem.”
Vladimir Nesov writes, “If you are downvoting indiscriminately, not separating the better comments from the worse ones, without even bothering to understand them, you are abusing the system.”
Anna writes, “This has the following advantages over blanket user-downvoting: . . . It does not impair quality-indicators on the user’s other comments”
The objection is valid. I retract my proposal and will say so in an addendum to my original comment.
The problem with my proposal is the part where the voter goes to a commenter’s lesswrong.com/user/ page and votes down 20 or 30 or so comments in a row. That dilutes or cancels out useful information, namely, votes from those who used the system the way it was intended.
If there were a way for a voter to reduce the karma of a person without reducing the point-score of any substantive comment, then my proposal might still have value, but without that, my proposal will have a destructive effect on the community, so of course I withdraw my proposal.
True, but if the behavior of the voters change so that the former becomes less informative, the site will tend to change so that the user’s karma will come to have a larger impact. In competent software development, changes in people’s behavior will cause maor changes in the software more often than changes in the software will cause major changes in the behavior of people. (Consequently, assuming the software developers are competent, most changes to the system are best initiated as changes to behavior rather than changes to the software—and if the software developers are not competent, then the site is probably doomed anyway.) Or so it seems to me.
A normally good contributor’s having a bad day is not going to be enough to trigger any downvoting of any of his comments under the policy I contemplate. Th policy I contemplate makes use of a general skill that I hypothesize that most participants on this site have: the ability to reserve judgement on someone till one has seen at least a dozen communications from that person and then to make a determination as to whether the person is worth continuing to pay attention to.
The people who have the most to contribute to a site like this are very busy. As Eliezer has written recently on this site, all that is need for this site to die is for these busy people to get discouraged because they see that the contributions of the worthwhile people are difficult to find among the contributions of the people who are not worth reading—and I stress that the people who are not worth reading often have a lot of free time which they use to generate many contributions.
Well, the voting is supposed to be the main way that the worthwhile contributions “float to the top” or float to where people are more likely to see them than to see the mediocre contributions. But that only works if the people who can distinguish a worthwhile contribution from a mediocre contribution bother to vote. So let us consider whether they do. For example, has Patri Friedman or Shane Legg bothered to vote? They both have made a few comments here. But they are both very busy people. I’ll send them both emails, referencing this conversation and asking them if they remember actually voting on comments here, and report back to y’all. (Eliezer is not a good person to ask in this regard because he has a big investment in the idea that a social web site based on voting will win, so of course he has been voting on the contributions here.)
The highest-scoring comment I know of is Shane Legg’s description of an anti-procrastination technique, which currently has 16 points. But there are thousands of readers of this site. Now it is possible that a lot more readers of Shane’s comment would have voted it up if it did not already have a high score, but I humbly suggest that it is more likely that only one or two or three percent of the readers of a comment would have bothered to vote on the comment regardless of its score.
Whether this site lives or dies seems to depend on the frequency with which the people who can tell a worthwhile comment from a non-worthwhile comment bother to vote. But like I said, these people tend to be very busy.
Hence my suggestion of adopting a policy of voting on commenters rather than coment—because that is going to save some of the busy person’s time.
There is a strong ethic in American society (and probably in other societies) that it is contributions and not individuals that should be judged. Well, I humbly suggest that since being able to contribute comments and posts here is not a basic human need, like housing or education or the opportunity to compete on an equal footing with other workers for income, the application of that admirable ethic to the decision of who gets to comment and post here is not worth the risk of this site’s going downhill to the point where the people who could have carried the site decide it is not worth the time out of their busy lives.
EDIT. If no other participants on this site declare their intention to use commenter-based voting, then I probably will not use commenter-based voting either because of what the economists call network effects. The only reason I suggested it in the first place is that conchis’s comment is not the first time someone here has indicated that voters other than me are already using commenter-based voting.
EDIT. I have backed down from the whole idea of downvoting many comments in one go. I do not delete this comment only because someone already replied to it.
If you are basing your judgment on at least a dozen communications from the commenter, then, as I explained, you have already identified plenty of comments that should be downloaded. If you base your decision on seeing the 10 of the 12 observed comments are problems, then you can dock the bad commenter 10 points. And if you are right, you will not be the only one. You do not need to personally dock the user 20 or 30 points.
If a person has excellent judgment to distinguish which comments should be up voted and which person should be down voted, but does not have the time to actually use that judgment, then that person is not going to be a successful protector of this site. Either take the time to do it right, or leave it to those who have the good judgment and the time, who there seems to be plenty of, giving that the system is working.