Loss of karma is a punishment. It only seems like it’s not when yours is high enough to isolate you from the actual effects and any realistic chance of having it wiped out over a single disagreement. Having it cost karma to downvote would make people think twice before downvoting a post that is already out of view, or downvoting all of someone’s posts in a subthread below a post that is already out of view.
It doesn’t seem to me like I would regard it as punishment even if someone could wipe out all my karma at once, and I would not downvote less if it cost me karma to downvote (assuming that was done instead of and equivalently to the downvote cap).
I am .85+ confident that replacing the downvote cap with a policy of spending karma to downvote would result in the total number of users issuing at least one downvote in a given month dropping by at least 50%, and .6+ confident of it dropping by at least 75%.
I am less confident about what effect it would have on the downvoting patterns of users who continue to issue at least one downvote. Call it (.2) no measurable effect, (.3) increased downvote rate, (.5) decreased rate, just to put some lines in the sand.
Do you think that between the detrimental effects of giving people angry about legitimate downvotes a target, and the beneficial effects of making people accountable for actual misuse of downvoting, making vote information publicly available would be a net benefit or net harm? (if downvoting itself is a good thing, wouldn’t people be rewarded in their standing in the community if people saw them making good downvotes?)
What about the effect of being able to downvote someone multiple times in a single subthread (with real effects on their karma) discouraging people from responding to requests for clarification? I know I’m not going to make that mistake again after getting burned.
I expect making vote information public would change (>95% of) users’ processes for deciding whether to vote, introducing significantly more consideration for the signaling effects of being seen to upvote or downvote a comment/post, and therefore proportionally less consideration for the desire to have more or fewer comments/posts like that one. I expect that, in turn, to reduce the overall emphasis on post/comment quality, which would likely make this site less valuable to me.
Bulk upvoting/downvoting like you describe is a trickier business. It often seems that people do so without really evaluating the comments they are voting on, as a way of punishing individuals. The term “karmassassination” is sometimes used around here to refer to that practice, and it’s frowned upon. On the other hand, voting on multiple comments in a thread, either because one wishes to see more/fewer threads of that sort, or because one genuinely considers each one to be individually entitled to the vote, is considered perfectly acceptable. It is, of course, difficult to automate a system that allows one but not the other.
Thinking about it now, enforcing a delay period between downvotes… say, preventing me from issuing more than one downvote in a 30-second period… might be a good modification.
A common problem with positive punishment as a training mechanism is that subjects overgeneralize on the target… e.g., learn some global lesson like “don’t ever respond to requests for clarification” even if the punisher intended a more narrow lesson like “don’t make comments like this one while responding to requests for clarification”. A (positive reinforcement plus negative punishment) training program, where undesirable behavior is ignored and desired behavior is rewarded, tends to work better, but requires significant self-discipline on the part of the trainer. When the training responsibility is distributed, this is difficult to manage. On pubilc forums like this one, I’ve never seen it implemented successfully, someone always ends up rewarding the undesirable behavior.
What about randomly (1 time in 10, say) requiring downvotes to be accompanied with an explanation (which will be posted as a comment, exposed to downvotes by the rest of the community if it’s a bad reason, and upvotes if it is a good one)?
What about allowing a post to be marked as “response to clarification request” and not subject to voting by anyone but the person it is in reply to?
What about randomly (1 time in 10, say) requiring downvotes to be accompanied with an explanation (which will be posted as a comment, exposed to downvotes by the rest of the community if it’s a bad reason, and upvotes if it is a good one)?
In the face of such a mechanism, I would surely protest it by posting a minimal comment along with the downvote, and also deleting it if that’s an option. Curation already feels somewhat like work; it doesn’t need to get harder.
What about allowing a post to be marked as “response to clarification request” and not subject to voting by anyone but the person it is in reply to?
Some folks actually won’t vote on anything in a thread they’ve commented on for neutrality reasons, and the last bit there seems inharmonious with that.
Is “get thicker-skinned about downvoting” an option?
How long have you been lurking here? You seem to have a lot of opinions about how good the existing mechanisms are for someone who hasn’t been commenting for very long.
I observe that the line you’re quoting was in response to your suggestion about the proposed required-explanation-for-downvote feature, which was not being proposed in the context of a post that was already at −11.
I infer that either you lost track of the context and genuinely believed thomblake was responding in that context, or you intentionally substituted one context for another for some purpose, presumably to make thomblake seem wrong and you seem right by contrast.
The former is a more charitable assumption, so with some misgivings, I am making it.
I was responding to the general idea that downvoting is “curation”. I don’t see why the specific context is necessary for that. Are you suggesting he wouldn’t have said the same thing in the other context? That posts already at or below −2 and posts in collapsed subthreads get downvoted shows that people downvote with non-curation purposes. Maybe the site would benefit from an explanation of what purposes they do have.
That posts already at or below −2 and posts in collapsed subthreads get downvoted shows that people downvote with non-curation purposes.
No. I do not generally check whether a comment is in a collapsed subthread before downvoting it. I downvote low-quality comments. It is more efficient.
If someone says that food is tasty and I reply “I don’t see how you can consider durian fruit tasty” I have gone from the general context (food) to a specific context (durian fruit).
In much the same way, if someone says downvoting is curation and I reply “well, nobody’s explained how downvoting a post that’s already at −11 is ‘curation’” I have gone from the general context (downvoting) to a specific (downvoting highly downvoted comments).
I would consider it reasonable, if I did either of those things, for an observer to conclude that I’d changed the context intentionally, in order to make it seem as though the speaker had said something I could more compellingly disagree with.
I haven’t seen a credible argument how downvoting a post that’s already at −11, or one that’s under several layers of collapsed posts, is “curation”.
I don’t think I’ve seen one either.
Though it’s worth noting that downvotes tend to be front-loaded in time, so something that’s at −5 a little while after posting could easily rise to +6 in only about a week. So your downvotes don’t ‘stop counting’ once the comment is already at −5.
I wonder if an algorithm could be invented to reduce the front-loading in time of negative karma from downvoting that is meant to offset later potential upvotes. Such a thing might have headed off the whole incident in the other thread (he’s stated that he was “ready to fight” out of anger from seeing half his karma gone)
If I’m understanding you correctly, sure. Just truncate all reported net karma scores for comments and posts at zero (while still recording the actual score), and calculate user total karma from reported karma rather than actual.
The suggestion gets made from time to time. Some people think it’s a good idea, others don’t.
More generally, no mechanism that allows a community to communicate what they do and don’t value will serve to prevent people whose contributions the community judges as valueless (or less valuable than they consider appropriate) from being upset by that judgment being communicated.
The question becomes to what degree a given community, acknowledging this, chooses to communicate their value judgments at the potential cost of upsetting people.
That might be an interesting experiment. I’m not confident I can predict what the results would be, given the effect you mention and the large amounts of “corrective voting” I’ve seen.
I imagine the mechanism would immediately apply pending downvotes until it has reached −2, and then apply the rest of pending downvotes either any time it goes above −2 or at some specified rate over time.
But the developer in me is saying that’s a too-complicated system with questionable benefit.
Such a thing might have headed off the whole incident in the other thread (he’s stated that he was “ready to fight” out of anger from seeing half his karma gone)
For reference, the discussed thread is here and User:pleeppleep is the user in question.
“Is “get thicker-skinned about downvoting” an option?”
Not at zero karma, it’s not.
How long have you been lurking here? You seem to have a lot of opinions about how good the existing mechanisms are for someone who hasn’t been commenting for very long.
A couple weeks. Am I really less qualified to examine the current system’s actual effect on new and low-karma users, though?
A couple weeks. Am I really less qualified to examine the current system’s actual effect on new and low-karma users, though?
Yes. As far as I can tell, you don’t have hard data about the impact these things have on usage. Given that, I’m comparing my general impressions gathered over the past 4 years to your general impressions gathered over the past couple weeks.
Providing [even sporadic] explanations for downvotes will allow people who are downvoted for good reasons a clearer way to adjust their behavior.
Exposing downvote reasons to community moderation will allow bad downvoting to be punished and good downvoting to be rewarded (this last one has the additional effect of raising the user’s downvote cap so they can continue making good downvotes.)
What I expect the second one to provide is obvious: remove the problematic incentives discouraging people from providing clarifications if the original post has been downvoted.
I was not describing bulk downvoting that could reasonably be called “karmassassination” or anything like that. This is limited to one subthread. The point is that you’re downvoting someone twice for the same thing for no better reason than that it’s across two posts. It discourages people from answering replies to their posts (and rewards editing answers into the original post [which doesn’t notify the person you’re responding to] or simply not engaging in discussion), which stifles discussion, because then the downvoter (who is not engaging in discussion to explain why they do not like the comments) has an extra opportunity to “legitimately” strike again, even though the downvotes are individually legitimate under the “want to have less posts like this one” theory of why people vote.
tl;dr:
On the other hand, voting on multiple comments in a thread, either because one wishes to see more/fewer threads of that sort, or because one genuinely considers each one to be individually entitled to the vote, is considered perfectly acceptable.
What I am suggesting is that it is considered “perfectly acceptable” in part because people have not fully considered this effect.
Maybe if the votes were allowed but the karma effect reduced?
P.S.
“learn some global lesson like “don’t ever respond to requests for clarification” even if the punisher intended a more narrow lesson like “don’t make comments like this one while responding to requests for clarification”.”
The point is that the punishment is for failing to change your mind. If you continue the discussion with anything but a full retraction, it’s likely that whatever the silent downvoter disliked is not fixed. So, no, I won’t be replying to requests for clarification—people can accept the inconvenience of watching the original post for additions as a cost of the current system.
And it is a global lesson: fewer posts on a topic unpopular enough to draw downvotes always means fewer downvotes, because if you stick to one post the downvoters can’t hit you twice while remaining within the “downvoting rules”.
I don’t see how that’s relevant, I’m talking about responses to requests for clarification. Controlled for original posts that had a negative score—any downvotes that were due to disagreement with someone’s position are obviously unlikely to change with clarification, and the response will get another downvote.
any downvotes that were due to disagreement with someone’s position are obviously unlikely to change with clarification, and the response will get another downvote
You continue to imply that voting behavior is entirely a function of whether voters agree with the commenter’s position. This continues to not match my experience.
It’s certainly possible that you’re correct and that I’m drawing the wrong lesson from my experience, of course.
My assumption is that disagreement is one of several reasons that people downvote, and that people are more likely to volunteer explanations (especially to new users) for the other reasons than for disagreement. Therefore, I assumed that the downvotes I got with no explanation were for disagreement. The one person who provided an alternate theory of why I was getting downvoted denied being one of the downvoters, and when I took his advice and clarified something from an earlier post, the new comment was also downvoted.
When I said I had observed a spoiler being stated “numerous” times in the thread, as evidence that the spoiler policy wasn’t preventing this effectively, someone replied asking for a list of links to specific comments; I replied with nine, and that post was downvoted three times.
It doesn’t seem to me like I would regard it as punishment even if someone could wipe out all my karma at once, and I would not downvote less if it cost me karma to downvote (assuming that was done instead of and equivalently to the downvote cap).
After about 5 minutes of thought...
I am .85+ confident that replacing the downvote cap with a policy of spending karma to downvote would result in the total number of users issuing at least one downvote in a given month dropping by at least 50%, and .6+ confident of it dropping by at least 75%.
I am less confident about what effect it would have on the downvoting patterns of users who continue to issue at least one downvote. Call it (.2) no measurable effect, (.3) increased downvote rate, (.5) decreased rate, just to put some lines in the sand.
None of that sounds blatantly unreasonable to me.
Do you think that between the detrimental effects of giving people angry about legitimate downvotes a target, and the beneficial effects of making people accountable for actual misuse of downvoting, making vote information publicly available would be a net benefit or net harm? (if downvoting itself is a good thing, wouldn’t people be rewarded in their standing in the community if people saw them making good downvotes?)
What about the effect of being able to downvote someone multiple times in a single subthread (with real effects on their karma) discouraging people from responding to requests for clarification? I know I’m not going to make that mistake again after getting burned.
A few related but distinct things here.
I expect making vote information public would change (>95% of) users’ processes for deciding whether to vote, introducing significantly more consideration for the signaling effects of being seen to upvote or downvote a comment/post, and therefore proportionally less consideration for the desire to have more or fewer comments/posts like that one. I expect that, in turn, to reduce the overall emphasis on post/comment quality, which would likely make this site less valuable to me.
Bulk upvoting/downvoting like you describe is a trickier business. It often seems that people do so without really evaluating the comments they are voting on, as a way of punishing individuals. The term “karmassassination” is sometimes used around here to refer to that practice, and it’s frowned upon. On the other hand, voting on multiple comments in a thread, either because one wishes to see more/fewer threads of that sort, or because one genuinely considers each one to be individually entitled to the vote, is considered perfectly acceptable. It is, of course, difficult to automate a system that allows one but not the other.
Thinking about it now, enforcing a delay period between downvotes… say, preventing me from issuing more than one downvote in a 30-second period… might be a good modification.
A common problem with positive punishment as a training mechanism is that subjects overgeneralize on the target… e.g., learn some global lesson like “don’t ever respond to requests for clarification” even if the punisher intended a more narrow lesson like “don’t make comments like this one while responding to requests for clarification”. A (positive reinforcement plus negative punishment) training program, where undesirable behavior is ignored and desired behavior is rewarded, tends to work better, but requires significant self-discipline on the part of the trainer. When the training responsibility is distributed, this is difficult to manage. On pubilc forums like this one, I’ve never seen it implemented successfully, someone always ends up rewarding the undesirable behavior.
What about randomly (1 time in 10, say) requiring downvotes to be accompanied with an explanation (which will be posted as a comment, exposed to downvotes by the rest of the community if it’s a bad reason, and upvotes if it is a good one)?
What about allowing a post to be marked as “response to clarification request” and not subject to voting by anyone but the person it is in reply to?
In the face of such a mechanism, I would surely protest it by posting a minimal comment along with the downvote, and also deleting it if that’s an option. Curation already feels somewhat like work; it doesn’t need to get harder.
Some folks actually won’t vote on anything in a thread they’ve commented on for neutrality reasons, and the last bit there seems inharmonious with that.
Is “get thicker-skinned about downvoting” an option?
How long have you been lurking here? You seem to have a lot of opinions about how good the existing mechanisms are for someone who hasn’t been commenting for very long.
I haven’t seen a credible argument how downvoting a post that’s already at −11, or one that’s under several layers of collapsed posts, is “curation”.
I observe that the line you’re quoting was in response to your suggestion about the proposed required-explanation-for-downvote feature, which was not being proposed in the context of a post that was already at −11.
I infer that either you lost track of the context and genuinely believed thomblake was responding in that context, or you intentionally substituted one context for another for some purpose, presumably to make thomblake seem wrong and you seem right by contrast.
The former is a more charitable assumption, so with some misgivings, I am making it.
I was responding to the general idea that downvoting is “curation”. I don’t see why the specific context is necessary for that. Are you suggesting he wouldn’t have said the same thing in the other context? That posts already at or below −2 and posts in collapsed subthreads get downvoted shows that people downvote with non-curation purposes. Maybe the site would benefit from an explanation of what purposes they do have.
No. I do not generally check whether a comment is in a collapsed subthread before downvoting it. I downvote low-quality comments. It is more efficient.
If someone says that food is tasty and I reply “I don’t see how you can consider durian fruit tasty” I have gone from the general context (food) to a specific context (durian fruit).
In much the same way, if someone says downvoting is curation and I reply “well, nobody’s explained how downvoting a post that’s already at −11 is ‘curation’” I have gone from the general context (downvoting) to a specific (downvoting highly downvoted comments).
I would consider it reasonable, if I did either of those things, for an observer to conclude that I’d changed the context intentionally, in order to make it seem as though the speaker had said something I could more compellingly disagree with.
I don’t think I’ve seen one either.
Though it’s worth noting that downvotes tend to be front-loaded in time, so something that’s at −5 a little while after posting could easily rise to +6 in only about a week. So your downvotes don’t ‘stop counting’ once the comment is already at −5.
I wonder if an algorithm could be invented to reduce the front-loading in time of negative karma from downvoting that is meant to offset later potential upvotes. Such a thing might have headed off the whole incident in the other thread (he’s stated that he was “ready to fight” out of anger from seeing half his karma gone)
If I’m understanding you correctly, sure. Just truncate all reported net karma scores for comments and posts at zero (while still recording the actual score), and calculate user total karma from reported karma rather than actual.
The suggestion gets made from time to time. Some people think it’s a good idea, others don’t.
More generally, no mechanism that allows a community to communicate what they do and don’t value will serve to prevent people whose contributions the community judges as valueless (or less valuable than they consider appropriate) from being upset by that judgment being communicated.
The question becomes to what degree a given community, acknowledging this, chooses to communicate their value judgments at the potential cost of upsetting people.
That might be an interesting experiment. I’m not confident I can predict what the results would be, given the effect you mention and the large amounts of “corrective voting” I’ve seen.
I imagine the mechanism would immediately apply pending downvotes until it has reached −2, and then apply the rest of pending downvotes either any time it goes above −2 or at some specified rate over time.
But the developer in me is saying that’s a too-complicated system with questionable benefit.
For reference, the discussed thread is here and User:pleeppleep is the user in question.
Not at zero karma, it’s not.
A couple weeks. Am I really less qualified to examine the current system’s actual effect on new and low-karma users, though?
Yes. As far as I can tell, you don’t have hard data about the impact these things have on usage. Given that, I’m comparing my general impressions gathered over the past 4 years to your general impressions gathered over the past couple weeks.
What do you expect the results of those changes to be?
Providing [even sporadic] explanations for downvotes will allow people who are downvoted for good reasons a clearer way to adjust their behavior.
Exposing downvote reasons to community moderation will allow bad downvoting to be punished and good downvoting to be rewarded (this last one has the additional effect of raising the user’s downvote cap so they can continue making good downvotes.)
What I expect the second one to provide is obvious: remove the problematic incentives discouraging people from providing clarifications if the original post has been downvoted.
OK. Thanks for clarifying that.
I was not describing bulk downvoting that could reasonably be called “karmassassination” or anything like that. This is limited to one subthread. The point is that you’re downvoting someone twice for the same thing for no better reason than that it’s across two posts. It discourages people from answering replies to their posts (and rewards editing answers into the original post [which doesn’t notify the person you’re responding to] or simply not engaging in discussion), which stifles discussion, because then the downvoter (who is not engaging in discussion to explain why they do not like the comments) has an extra opportunity to “legitimately” strike again, even though the downvotes are individually legitimate under the “want to have less posts like this one” theory of why people vote.
tl;dr:
What I am suggesting is that it is considered “perfectly acceptable” in part because people have not fully considered this effect.
Maybe if the votes were allowed but the karma effect reduced?
P.S.
“learn some global lesson like “don’t ever respond to requests for clarification” even if the punisher intended a more narrow lesson like “don’t make comments like this one while responding to requests for clarification”.”
The point is that the punishment is for failing to change your mind. If you continue the discussion with anything but a full retraction, it’s likely that whatever the silent downvoter disliked is not fixed. So, no, I won’t be replying to requests for clarification—people can accept the inconvenience of watching the original post for additions as a cost of the current system.
And it is a global lesson: fewer posts on a topic unpopular enough to draw downvotes always means fewer downvotes, because if you stick to one post the downvoters can’t hit you twice while remaining within the “downvoting rules”.
That’s inconsistent with my experience here.
You are, of course, free to do that.
Ditto.
Requests for clarification, in particular, are often upvoted on net after a few days.
Except, in some cases, when they are blatantly passive aggressive requests.
I don’t see how that’s relevant, I’m talking about responses to requests for clarification. Controlled for original posts that had a negative score—any downvotes that were due to disagreement with someone’s position are obviously unlikely to change with clarification, and the response will get another downvote.
You continue to imply that voting behavior is entirely a function of whether voters agree with the commenter’s position. This continues to not match my experience.
It’s certainly possible that you’re correct and that I’m drawing the wrong lesson from my experience, of course.
It isn’t obvious, though.
My assumption is that disagreement is one of several reasons that people downvote, and that people are more likely to volunteer explanations (especially to new users) for the other reasons than for disagreement. Therefore, I assumed that the downvotes I got with no explanation were for disagreement. The one person who provided an alternate theory of why I was getting downvoted denied being one of the downvoters, and when I took his advice and clarified something from an earlier post, the new comment was also downvoted.
When I said I had observed a spoiler being stated “numerous” times in the thread, as evidence that the spoiler policy wasn’t preventing this effectively, someone replied asking for a list of links to specific comments; I replied with nine, and that post was downvoted three times.
I agree with this
I suppose this is possible, but I doubt the size of the effect is significant.
Aha—I’d missed that bit.
Sure, though I usually avoid downvoting for disagreement, and I’ve gotten the impression that’s still a norm around here.
ETA: And actually countered somewhat by the tendency of several frequent users to upvote for disagreement.