If I had a vote, I’d vote for getting rid of strong votes altogether. Here’s another downside from my perspective: I actually don’t like getting strong upvotes on my comments, because if that person didn’t do a strong upvote, in most cases others would eventually (weakly) upvote that comment to around the same total (because people don’t bother to upvote if they think the comment’s karma is already what it deserves), and (at least for me) it feels more rewarding and more informative to know that several people upvoted a comment than to know that one person strongly upvoted a comment.
Also strong upvotes always make me think “who did that?”, which is pointless because it’s too hard to guess based on the available information but I can’t help myself. (Votes that are 3 points also make me think this.) (I’ve complained about this before, but from the voter perspective as opposed to the commenter perspective.) I think I’d be happier if everyone just had either 1 or 2 point votes.
The 3-point votes are an enormous entropy leak: only 13 users have a 3-point weak upvote (only 8-ish of which I’d call currently “active”), and probably comparatively few 3-point votes are strong-upvotes from users with 100–249 karma. (In contrast, about 400 accounts have 2-point weak upvotes, which I think of as “basically everyone.”)
Gah, this makes me even more reluctant to vote. I didn’t realize there are so few active 3-point members. (Didn’t know about Issa Rice’s karma list.) Seriously, there have already been multiple instances since you wrote this that I thought about voting and then stopped myself.
I’m not sure why the LW team hasn’t made a change about this, but if they really want to keep the 3-point votes, maybe drop the threshold a bit so that there are at least several tens of users with 3-point votes?
Yeah this discussion had me update that we should probably just drop 3-point smallvotes. (dropping the threshold would solve this problem, but not the problem I personally experience most, which is ‘a lot of comments feel worth upvoting a tiny bit, but 3-karma feels excessive’).
Yesterday the team discussed some weirder ideas, such as:
Just don’t display karma for comments. Instead, just use it to silently sort things in the background. This might also make people more willing to downvote (since people often find it unpleasantly mean to downvote things below 0). It might also curtail some of the “voting as yay/boo”. This is what hackernews currently does AFAIK. We might also copy hackernews’s thing of “downvoted things start to fade away based on how downvoted they are.
On the flipside, sometimes it’s actually good to see when things are highly upvoted (such as an important criticism or question)
Alternately: maybe karma doesn’t get displayed until it has at least 3 votes (possibly in addition to the OP’s auto-upvote?). This might help obfuscate who’s been doing which upvoting. (I personally find it most noticeable when the karma score and voter-count is low)
I prefer to see the karma, because “sometimes it’s actually good to see when things are highly upvoted (such as an important criticism or question)”.
While we’re on the topic of voting, when I look at my old LW1 comments I occasionally see 10-20 people vote up one of my comments. Now my comments often get voted up to 10-20 karma by 1-4 people (besides my own default upvote), but almost never receive more than 10 votes. This makes me worried that I’m reaching a lot fewer people with my content compared to those days. Is this true, or do people just vote less frequently now?
It is (alas) definitely the case that there are fewer site participants now than in Ye Old Golden Days, although the metrics have been trending upwards for the past year(ish). (sometime we’ll do an updated analytics post to give a clearer picture of that)
I do also think that in addition to that, people also just vote less. If I remember correctly, number of people voting in a given week is about 60% of what it was at the peak, but total number of votes per week is closer to 35% or something like that. There are also a bunch less comments, so you likely get some quadratic effects that at least partially explain this.
I’d get rid of strong upvotes as well, or perhaps make voting nonlinear, such that a weak/strong vote changes in value based on how many voters expressed an opinion (as it kind of does over time—strong votes only matter a small bit when there are 20+ votes cast, but if they’re one of the first or only few to vote, they’re HUGE). Or perhaps only display the ordinal value of posts and comments (relative to others shown on the page), with the actual vote values hidden in the same way we do number of voters.
The vast majority of my comments get 5 or fewer voters. This is data in itself, of course, but it means that I react similarly to Wei when I see an outsided change.
In this case this was actually me removing a weak upvote, presumably at the same time someone else cast a regular weak downvote? (I had originally upvoted as a general reward for providing information about what users might care about, then realized I kinda didn’t want to make it look like the object-level idea have tons of support. Which is relevant. In any case apologies for confusion. :p)
If I had a vote, I’d vote for getting rid of strong votes altogether. Here’s another downside from my perspective: I actually don’t like getting strong upvotes on my comments, because if that person didn’t do a strong upvote, in most cases others would eventually (weakly) upvote that comment to around the same total (because people don’t bother to upvote if they think the comment’s karma is already what it deserves), and (at least for me) it feels more rewarding and more informative to know that several people upvoted a comment than to know that one person strongly upvoted a comment.
Also strong upvotes always make me think “who did that?”, which is pointless because it’s too hard to guess based on the available information but I can’t help myself. (Votes that are 3 points also make me think this.) (I’ve complained about this before, but from the voter perspective as opposed to the commenter perspective.) I think I’d be happier if everyone just had either 1 or 2 point votes.
The 3-point votes are an enormous entropy leak: only 13 users have a 3-point weak upvote (only 8-ish of which I’d call currently “active”), and probably comparatively few 3-point votes are strong-upvotes from users with 100–249 karma. (In contrast, about 400 accounts have 2-point weak upvotes, which I think of as “basically everyone.”)
Gah, this makes me even more reluctant to vote. I didn’t realize there are so few active 3-point members. (Didn’t know about Issa Rice’s karma list.) Seriously, there have already been multiple instances since you wrote this that I thought about voting and then stopped myself.
I’m not sure why the LW team hasn’t made a change about this, but if they really want to keep the 3-point votes, maybe drop the threshold a bit so that there are at least several tens of users with 3-point votes?
Looks like the weak 3-votes are gone now!
Yep, it didn’t seem worth the cost of the chilling effects that were discussed in this thread.
Yeah. Even if Wei_Dai is the only one chilled then that’s still a huge fraction of the 3-point members.
I think we probably should have announced this with more fanfare but a series of distracting things happened and we forgot. Alas!
Yeah this discussion had me update that we should probably just drop 3-point smallvotes. (dropping the threshold would solve this problem, but not the problem I personally experience most, which is ‘a lot of comments feel worth upvoting a tiny bit, but 3-karma feels excessive’).
Yesterday the team discussed some weirder ideas, such as:
Just don’t display karma for comments. Instead, just use it to silently sort things in the background. This might also make people more willing to downvote (since people often find it unpleasantly mean to downvote things below 0). It might also curtail some of the “voting as yay/boo”. This is what hackernews currently does AFAIK. We might also copy hackernews’s thing of “downvoted things start to fade away based on how downvoted they are.
On the flipside, sometimes it’s actually good to see when things are highly upvoted (such as an important criticism or question)
Alternately: maybe karma doesn’t get displayed until it has at least 3 votes (possibly in addition to the OP’s auto-upvote?). This might help obfuscate who’s been doing which upvoting. (I personally find it most noticeable when the karma score and voter-count is low)
I prefer to see the karma, because “sometimes it’s actually good to see when things are highly upvoted (such as an important criticism or question)”.
While we’re on the topic of voting, when I look at my old LW1 comments I occasionally see 10-20 people vote up one of my comments. Now my comments often get voted up to 10-20 karma by 1-4 people (besides my own default upvote), but almost never receive more than 10 votes. This makes me worried that I’m reaching a lot fewer people with my content compared to those days. Is this true, or do people just vote less frequently now?
It is (alas) definitely the case that there are fewer site participants now than in Ye Old Golden Days, although the metrics have been trending upwards for the past year(ish). (sometime we’ll do an updated analytics post to give a clearer picture of that)
I do also think that in addition to that, people also just vote less. If I remember correctly, number of people voting in a given week is about 60% of what it was at the peak, but total number of votes per week is closer to 35% or something like that. There are also a bunch less comments, so you likely get some quadratic effects that at least partially explain this.
Aren’t there also people for whom 3 points is a strong upvote that you can’t distinguish from those where 3 point is a weak upvote?
True, but I think you can usually tell what sort of things might-have-gotten strong upvoted
I’d get rid of strong upvotes as well, or perhaps make voting nonlinear, such that a weak/strong vote changes in value based on how many voters expressed an opinion (as it kind of does over time—strong votes only matter a small bit when there are 20+ votes cast, but if they’re one of the first or only few to vote, they’re HUGE). Or perhaps only display the ordinal value of posts and comments (relative to others shown on the page), with the actual vote values hidden in the same way we do number of voters.
The vast majority of my comments get 5 or fewer voters. This is data in itself, of course, but it means that I react similarly to Wei when I see an outsided change.
Someone strong-voted down my comment, from 11 to 7. (Normally I wouldn’t mention this, but it seems relevant here. :)
In this case this was actually me removing a weak upvote, presumably at the same time someone else cast a regular weak downvote? (I had originally upvoted as a general reward for providing information about what users might care about, then realized I kinda didn’t want to make it look like the object-level idea have tons of support. Which is relevant. In any case apologies for confusion. :p)