Being more interested in voting as community feedback and less interested in getting the karmic high score, I’ve taken to removing the auto-upvote on any of my own comments that I would not have upvoted had someone else posted them, then continued to monitor the vote scores for the next few days. I’ve come to a couple conclusions:
Most people seem to be voting based on what they think the final score should be, not their personal approval independent of existing score. Most posts ended up at 1 or 2, independent of the default self-upvote.
Given roughly similar information density, people are more likely to upvote short comments.
People are far too likely to upvote bad jokes. For a while I think my highest-scored comment was an “Eliezer Yudkowsky Fact”. Oy vey.
I believe the behavior is mostly as follows. If the comment is voted neutral (currently, 1 point), people only upvote or downvote if they feel strongly about whether it’s good or bad (according to whatever valuation). If the comment is already upvoted or downvoted, people are more likely to cast an opposite vote if they feel that the comment is rated incorrectly (in the wrong direction), to “fix” its rating. Thus, if someone upvotes a 0-point comment, it doesn’t mean that the comment is considered worthy of an upvote (to reach, say, 9 points), it only means that the person thought that it didn’t deserve a downvote from 1 point.
Thus, there seems to be 2 modes of voting: hard voting and soft voting.
Hard vote is supposed to push the comment all the way in the specified direction.
Soft vote is supposed to push the comment in the given direction, but only towards neutral rating, and not beyond that. It’s intended only to eliminate the opposing votes, not to set the vote.
So the solution is either to change the system’s design, or change the user’s behavior? The latter seems unlikely, so what would a system designed to utilize soft voting look like?
My highest-scored comment was also my worst. (It wouldn’t have been awful in a different context; but I might not have gotten so many up-votes in an appropriate context.)
My comment just above, though, sets a new personal speed record. It was up for about 10 seconds before someone down-voted it.
I am curious as to why they down-voted it. It would be nice to require people to give a brief (possibly anonymous; possibly not displayed by default) explanation when they downvote something.
A ranking preference is expressed as a vote. An explanation is expressed as a reply. In the system as it stands, these are two very discrete actions. How often and in what circumstances do people use them in combination? What would be the effects of explicitly linking them?
Being more interested in voting as community feedback and less interested in getting the karmic high score, I’ve taken to removing the auto-upvote on any of my own comments that I would not have upvoted had someone else posted them, then continued to monitor the vote scores for the next few days. I’ve come to a couple conclusions:
Most people seem to be voting based on what they think the final score should be, not their personal approval independent of existing score. Most posts ended up at 1 or 2, independent of the default self-upvote.
Given roughly similar information density, people are more likely to upvote short comments.
People are far too likely to upvote bad jokes. For a while I think my highest-scored comment was an “Eliezer Yudkowsky Fact”. Oy vey.
I believe the behavior is mostly as follows. If the comment is voted neutral (currently, 1 point), people only upvote or downvote if they feel strongly about whether it’s good or bad (according to whatever valuation). If the comment is already upvoted or downvoted, people are more likely to cast an opposite vote if they feel that the comment is rated incorrectly (in the wrong direction), to “fix” its rating. Thus, if someone upvotes a 0-point comment, it doesn’t mean that the comment is considered worthy of an upvote (to reach, say, 9 points), it only means that the person thought that it didn’t deserve a downvote from 1 point.
Thus, there seems to be 2 modes of voting: hard voting and soft voting.
Hard vote is supposed to push the comment all the way in the specified direction.
Soft vote is supposed to push the comment in the given direction, but only towards neutral rating, and not beyond that. It’s intended only to eliminate the opposing votes, not to set the vote.
Yes. The quandary seems, to me, that the voting system is designed for hard voting, but in practice more people are using soft voting.
So the solution is either to change the system’s design, or change the user’s behavior? The latter seems unlikely, so what would a system designed to utilize soft voting look like?
Hacker News solved the problem by not displaying comment karma, except to the owner of the comment.
My highest-scored comment was also my worst. (It wouldn’t have been awful in a different context; but I might not have gotten so many up-votes in an appropriate context.)
My comment just above, though, sets a new personal speed record. It was up for about 10 seconds before someone down-voted it.
I am curious as to why they down-voted it. It would be nice to require people to give a brief (possibly anonymous; possibly not displayed by default) explanation when they downvote something.
A ranking preference is expressed as a vote. An explanation is expressed as a reply. In the system as it stands, these are two very discrete actions. How often and in what circumstances do people use them in combination? What would be the effects of explicitly linking them?