Note that we really should have an around here. Ah, if only there was an as well! Technology saves the day again!
Killfiles are shit. The incentive structure they create is all screwed up: not only would they contribute to fragmenting the community into little incestuous clumps of people all vigorously pandering to each other (if someone killfiles you, you don’t have to worry about their votes), but they raise barriers to entry (by making it necessary for new users to killfile every troll and douchebag in the community) and don’t materially discourage trolling (because people need to read your stuff to killfile you, and because hundreds of clicks are a lot more effort than a few sentences of drivel). At best they can function as a patch over an inadequate moderation policy, which at least I’ll confess we’ve historically had.
“Ignore votes” is actually kind of interesting, but it doesn’t solve the problem of people’s privs getting affected by mass downvotes, makes voting a lot less anonymous if it’s reversible, and still creates somewhat ugly incentives. It should never be harder, as a sum of effort, for a forum to correct for the presence of a problem user than it is for that user to create problems.
I don’t think they work too well in a diverse community, either: I used to moderate such a community, on a codebase that introduced killfile features during my tenure, and its only substantial effect on moderation seemed to be cutting down on complaints from long-term users that had well-developed killfiles. (I’ve gone into more detail elsewhere in this thread on its cultural effects.) Since all communities are mostly newer/transient people during the active phase of their lifecycle, this wasn’t much consolation.
That is a much harder administrative problem, though, and I’ve never found a solution that works other than “have a good seed culture, create strong norms against empty rhetoric and generally being a dick, and choose your mods very carefully”. With the LW experience in mind I’m actually kind of a fan of karma as a self-moderation tool, but it introduces some problems of its own (see: Recent Unpleasantness), isn’t anywhere close to a panacea (see: half of Reddit), and doesn’t completely eliminate the need for good people with higher perm levels.
Some people are concerned about Signal/Noise. Filter some people, and poof, signal/noise is improved for you according to your tastes.
Some people liked walled gardens. Great! is your personal wall. Throw everyone you don’t want to over that wall.
not only would they contribute to fragmenting the community into little incestuous clumps of people
You mean, like life, where people associate with the people they like, and don’t associate with those they don’t?
making it necessary for new users to killfile every troll and douchebag in the community
It’s hardly necessary, as it’s impossible to do now. It merely gives you an option to do so.
because people need to read your stuff to killfile you, and because hundreds of clicks are a lot more effort than a few sentences of drivel
What? I read a post, get annoyed, and click, that person drops into my bit bucket never to be seen again. Nothing could possibly be easier.
At best they can function as a patch over an inadequate moderation policy
It’s a personal moderation policy that you control. I would rather have Eugine on the list. From his 9000 karma, I doubt that I’m alone. But I don’t have that option.
It should never be
This is simple a category error. What is the “it” that has a moral duty to “never be”?
for a forum to correct for the presence of a problem user than it is for that user to create problems.
Problem user, according to who? Eugine’s downvoting was a minor problem in the karma system. That’s hardly the only deficiency of it. That’s hardly the only problem around here.
My problem is having a valuable poster banned from the list.
It’s hardly necessary, as it’s impossible to do now. It merely gives you an option to do so.
Options have a habit of becoming mandatory over time as norms adjust to their presence. Make it possible to ignore people and I guarantee that a year later, when the next white supremacist or militant Maoist or Randroid or whatever shows up, you’ll get people saying that it’s not a problem, everyone just needs to ignore them and they’ll never need to see them again. I further guarantee that said white supremacists etc. will respond to this by settling down and carving out hateful little niches for themselves in the forum ecosystem, as the people that care start dropping them into their killfiles and stop downvoting their posts or leaving angry responses or, y’know, actually proving them wrong.
All of which comes to a huge waste of effort, because...
What? I read a post, get annoyed, and click, that person drops into my bit bucket never to be seen again. Nothing could possibly be easier.
...you should now imagine that process being repeated by some large fraction of the two thousand users on this forum, every time a problem (excuse me, controversial) user shows up or creates a new sockpuppet. Doesn’t look so trivial now, does it?
the next white supremacist or militant Maoist or Randroid
Some people want centrally enforced ideological litmus tests, and some don’t.
Doesn’t look so trivial now, does it?
Scales linearly. Click , and they’re gone, for everyone who doesn’t want to see them. Nothing could be simpler. An order of magnitude (or two) less sound and fury than we’ve spent on Eugine.
The recent ban was executed through administrative action. That’s O(1), albeit apparently with a high constant factor if Kaj’s posts are to be trusted. There’s been a lot of drama surrounding it, but that doesn’t have anything to do with scalability.
(Personally, I’d say most of the drama has to do with preexisting cultural and administrative issues that this has dragged squirming into the light, and takes the late unpleasantness as a proximate rather than an ultimate cause, but we may reasonably disagree on that point.)
Killfiles are shit. The incentive structure they create is all screwed up: not only would they contribute to fragmenting the community into little incestuous clumps of people all vigorously pandering to each other (if someone killfiles you, you don’t have to worry about their votes), but they raise barriers to entry (by making it necessary for new users to killfile every troll and douchebag in the community) and don’t materially discourage trolling (because people need to read your stuff to killfile you, and because hundreds of clicks are a lot more effort than a few sentences of drivel). At best they can function as a patch over an inadequate moderation policy, which at least I’ll confess we’ve historically had.
“Ignore votes” is actually kind of interesting, but it doesn’t solve the problem of people’s privs getting affected by mass downvotes, makes voting a lot less anonymous if it’s reversible, and still creates somewhat ugly incentives. It should never be harder, as a sum of effort, for a forum to correct for the presence of a problem user than it is for that user to create problems.
Killfiles are not efficient for communities of people who think alike. They are pretty good for collections of radically diverse people.
I understand your point about scaling. But I am also highly suspicious of one-size-fits-all solutions.
I don’t think they work too well in a diverse community, either: I used to moderate such a community, on a codebase that introduced killfile features during my tenure, and its only substantial effect on moderation seemed to be cutting down on complaints from long-term users that had well-developed killfiles. (I’ve gone into more detail elsewhere in this thread on its cultural effects.) Since all communities are mostly newer/transient people during the active phase of their lifecycle, this wasn’t much consolation.
That is a much harder administrative problem, though, and I’ve never found a solution that works other than “have a good seed culture, create strong norms against empty rhetoric and generally being a dick, and choose your mods very carefully”. With the LW experience in mind I’m actually kind of a fan of karma as a self-moderation tool, but it introduces some problems of its own (see: Recent Unpleasantness), isn’t anywhere close to a panacea (see: half of Reddit), and doesn’t completely eliminate the need for good people with higher perm levels.
is awesome.
Some people are concerned about Signal/Noise. Filter some people, and poof, signal/noise is improved for you according to your tastes.
Some people liked walled gardens. Great! is your personal wall. Throw everyone you don’t want to over that wall.
You mean, like life, where people associate with the people they like, and don’t associate with those they don’t?
It’s hardly necessary, as it’s impossible to do now. It merely gives you an option to do so.
What? I read a post, get annoyed, and click, that person drops into my bit bucket never to be seen again. Nothing could possibly be easier.
It’s a personal moderation policy that you control. I would rather have Eugine on the list. From his 9000 karma, I doubt that I’m alone. But I don’t have that option.
This is simple a category error. What is the “it” that has a moral duty to “never be”?
Problem user, according to who? Eugine’s downvoting was a minor problem in the karma system. That’s hardly the only deficiency of it. That’s hardly the only problem around here.
My problem is having a valuable poster banned from the list.
Options have a habit of becoming mandatory over time as norms adjust to their presence. Make it possible to ignore people and I guarantee that a year later, when the next white supremacist or militant Maoist or Randroid or whatever shows up, you’ll get people saying that it’s not a problem, everyone just needs to ignore them and they’ll never need to see them again. I further guarantee that said white supremacists etc. will respond to this by settling down and carving out hateful little niches for themselves in the forum ecosystem, as the people that care start dropping them into their killfiles and stop downvoting their posts or leaving angry responses or, y’know, actually proving them wrong.
All of which comes to a huge waste of effort, because...
...you should now imagine that process being repeated by some large fraction of the two thousand users on this forum, every time a problem (excuse me, controversial) user shows up or creates a new sockpuppet. Doesn’t look so trivial now, does it?
Some people want centrally enforced ideological litmus tests, and some don’t.
Scales linearly. Click , and they’re gone, for everyone who doesn’t want to see them. Nothing could be simpler. An order of magnitude (or two) less sound and fury than we’ve spent on Eugine.
The entire point is that we can and should do a lot better than O(n).
We haven’t. To quote myself:
The recent ban was executed through administrative action. That’s O(1), albeit apparently with a high constant factor if Kaj’s posts are to be trusted. There’s been a lot of drama surrounding it, but that doesn’t have anything to do with scalability.
(Personally, I’d say most of the drama has to do with preexisting cultural and administrative issues that this has dragged squirming into the light, and takes the late unpleasantness as a proximate rather than an ultimate cause, but we may reasonably disagree on that point.)