What amount of bad comments would be a reasonable threshold for downvoting someone's every comment? 50 percent? 20 percent?
My own standard for downvoting a user as a category is "Would Less Wrong be better off if this user went away?" It's possible that there's some threshold percentage that causes me to arrive at that judgement, but if so, I don't know what that threshold is.
Should there be guidelines for that?
My suggested guideline is: if LessWrong would be better off if user X went away, downvote user X's comments.
You don't simply happen to notice most old comments, but if you're so inclined and have the time, clickfest awaits.
Sure, that's true. And I certainly agree that it's easier to retroactively downvote all of a single user's comments than it is to retroactively downvote all the comments in various other categories. It is consequently true that if downvoting all the comments in a category I want less of is a bad thing, doing so for the category "user X's comments" is particularly bad because it's both bad and easy.
Silence already is the norm. "By the way I downvoted all your comments because of X." How do you expect that to go?
Sorry, I was unclear. By "silence" I don't mean the absence of English sentences, I mean the absence of signal.
To rephrase... if failing to downvote comments in a category that's of negative value to the site becomes (or remains) the norm, it becomes (remains) true that users won't know that they should change their behaviour. (Corresponding things are true of failing to upvote comments in a positive-value category.)
If that norm is replaced by up/downvoting such comments as I advocate, you're right that the user doesn't suddenly become aware of what the problem/benefit is. But they weren't aware of that information before implementing that norm-replacement, either.
Looked at the other way: if our goal is to maximize the amount of information people get about what's wrong (or right) with their comments, discussing how we ought to be using the karma system is a waste of time, because karma is a deeply flawed mechanism for achieving that goal.
If that norm is replaced by up/downvoting such comments as I advocate, you're right that the user doesn't suddenly become aware of what the problem/benefit is. But they weren't aware of that information before implementing that norm-replacement, either.
If all your comments are downvoted because someone deemed that 20% of them are damaging, it's much more difficult to deduce why that happened than if voting happens per comment.
...if our goal is to maximize the amount of information people get about what's wrong (or right) with their comments [...] karma i
I'm just tired of the signal pollution, and would like to be able to use karma to honestly appraise the worth of my articles and posts, without seeing 80% of my downvotes come in chunks that correspond precisely to how many posts I've made since the last massive downvote spree.
EDIT to add data points:
Spurious downvoting stopped soon after I named a particular individual (not ALL downvoting stopped, but the downvotes I got all seemed on-the-level.)
One block of potentially spurious downvoting occurred approximately one week ago, but then karma patterns returned to expected levels. I consider this block dubious, because it reasonably matches what I'd expect to see if someone noticed several of my posts together and disagreed with all of them, and did not match the usual pattern of starting with the earliest or latest post that I had made and downvoting everything (it downvoted all posts in a few threads, but not in other threads), so I'm just adding for completeness.
Spurious, indiscriminate downvoting started up again approximately half an hour ago on Sunday (12/1/2013), around noon MDT.
Edit: And now on Tuesday, 12/3/2013, at 10 AM, I'm watching my karma go down again... about 30 points so far.
Edit: And now on Saturday, 12/14/2013, at 2 PM, I'm watching my karma go down again... about 15 points so far, at a rate of about 1-2 points per second.