How do people decide what comments to upvote? I see two kinds of possible strategies:
Use my approval level of the comment to decide how to vote (up, down or neutral). Ignore other people’s votes on this comment.
Use my approval level to decide what total voting score to give the comment. Vote up or down as needed to move towards that target.
My own initial approach belonged to the first class. However, looking at votes on my own comments, I get the impression most people use the second approach. I haven’t checked this with enough data to be really certain, so would value more opinions & data.
Here’s what I found: I summed the votes from the last 4 pages of my own comments (skipping the most recent page because recent comments may yet be voted on):
Score <0: 2
Score =0: 36
Score =0: 39
Score =2: 14
Score =3: 5
Score >3: 6
35% of my comments are voted 0, and 52% are voted 1 or 2. There are significantly more than 1 or 2 people participating in the same threads as me. It is not likely that for each of these comments, just one or two people happened to like it, and the rest didn’t. It is even less likely that for each of these comments, up- and down-votes balanced so as to leave +1 or +2.
So it’s probable that many people use the second approach: they see a comment, think “that’s nice, deserves +1 but no more”, and then if it’s already at +1, they don’t vote.
How do you vote? And what do you see as the goal of the voting process?
I self-identify as using the first one, with a caveat.
The second is obviously awful for communicating any sort of information given that only the sum of votes is displayed rather than total up and total down. The second is order dependent and often means you’ll want to change your vote later based purely on what others think of the post.
My “strategy” is to vote up and down based on whether I’d have wanted others with more insight than me to vote to bring my attention to or away from a comment, unless I feel I have special insight, in which case it’s based on whether I want to bring others’ attention to or away from a comment.
This is because I see the goal of the voting process that readers’ independent opinions on how much a comment is worth readers’ attention be aggregated and used to bring readers’ attention to or away from a comment. As a side effect, the author of a comment can use the aggregated score to determine whether her readers felt the comment was worth their collective attention.
Furthermore since each reader’s input comes in distinct chunks of exactly −1, 0, or +1, it’s wildly unlikely that voting very often results in the best aggregation: instead I leave a comment alone unless I feel it was(is) significantly worth or not worth my(your) attention.
The caveat: there is a selection effect in which comments I vote on, since my attention will be drawn away from comments with very negative karma. There is also undoubtedly an unconscious bias away from voting up a comment with very high karma: since I perceive the goal to be to shift attention, once a comment has very high karma I know it’s going to attract attention so my upvote is in fact worth fewer attention-shift units. But I haven’t yet consciously noticed that kick in until about +10 or so.
At home I use the Anti-Kibitzer, which enforces 1. I’ve been on vacation for a couple days and noticed the temptation to use 2. Gave in on one occasion, I’m afraid. On balance I’ll stick to 1, as 2 seems too vulnerable to information cascades.
It would be also interesting to check the difference between comments on quotes threads and comments on substantive posts—at least one person has proposed that quotations are disproportionately subject to strategy 1 voting over strategy 2.
How do people decide what comments to upvote? I see two kinds of possible strategies:
Use my approval level of the comment to decide how to vote (up, down or neutral). Ignore other people’s votes on this comment.
Use my approval level to decide what total voting score to give the comment. Vote up or down as needed to move towards that target.
My own initial approach belonged to the first class. However, looking at votes on my own comments, I get the impression most people use the second approach. I haven’t checked this with enough data to be really certain, so would value more opinions & data.
Here’s what I found: I summed the votes from the last 4 pages of my own comments (skipping the most recent page because recent comments may yet be voted on):
Score <0: 2
Score =0: 36
Score =0: 39
Score =2: 14
Score =3: 5
Score >3: 6
35% of my comments are voted 0, and 52% are voted 1 or 2. There are significantly more than 1 or 2 people participating in the same threads as me. It is not likely that for each of these comments, just one or two people happened to like it, and the rest didn’t. It is even less likely that for each of these comments, up- and down-votes balanced so as to leave +1 or +2.
So it’s probable that many people use the second approach: they see a comment, think “that’s nice, deserves +1 but no more”, and then if it’s already at +1, they don’t vote.
How do you vote? And what do you see as the goal of the voting process?
I self-identify as using the first one, with a caveat.
The second is obviously awful for communicating any sort of information given that only the sum of votes is displayed rather than total up and total down. The second is order dependent and often means you’ll want to change your vote later based purely on what others think of the post.
My “strategy” is to vote up and down based on whether I’d have wanted others with more insight than me to vote to bring my attention to or away from a comment, unless I feel I have special insight, in which case it’s based on whether I want to bring others’ attention to or away from a comment.
This is because I see the goal of the voting process that readers’ independent opinions on how much a comment is worth readers’ attention be aggregated and used to bring readers’ attention to or away from a comment. As a side effect, the author of a comment can use the aggregated score to determine whether her readers felt the comment was worth their collective attention.
Furthermore since each reader’s input comes in distinct chunks of exactly −1, 0, or +1, it’s wildly unlikely that voting very often results in the best aggregation: instead I leave a comment alone unless I feel it was(is) significantly worth or not worth my(your) attention.
The caveat: there is a selection effect in which comments I vote on, since my attention will be drawn away from comments with very negative karma. There is also undoubtedly an unconscious bias away from voting up a comment with very high karma: since I perceive the goal to be to shift attention, once a comment has very high karma I know it’s going to attract attention so my upvote is in fact worth fewer attention-shift units. But I haven’t yet consciously noticed that kick in until about +10 or so.
At home I use the Anti-Kibitzer, which enforces 1. I’ve been on vacation for a couple days and noticed the temptation to use 2. Gave in on one occasion, I’m afraid. On balance I’ll stick to 1, as 2 seems too vulnerable to information cascades.
It is worth noting that people have explicitly claimed to be following strategy 2 here. Edit: This is far from the only example; just the one I found by searching for “upvoted to”.
It would be also interesting to check the difference between comments on quotes threads and comments on substantive posts—at least one person has proposed that quotations are disproportionately subject to strategy 1 voting over strategy 2.