I’ve commented also that the karma system, as it is currently, causes less participation on the site. Just to save time I’ll paste it here.
“The fundamental flaw that I see with LessWrong’s main site is that its karma/moderating system has the effect of silencing and banning people for being disagreed with or misunderstood. This is a major problem. You cannot mix “I don’t agree with you” or “I don’t understand you” with “you will be punished and silenced.”
People who spam, flame, or otherwise destroy conversation are the ones who need to be silenced, ignored or banned, and a lot of sites like Facebook have separate buttons to perform exactly that function. People in the other category, who are misunderstood or disagreed with, but who discuss constructively and rationally, are the ones who MOST need to be able to speak. I think the punishment and silencing, and the threat of it, contributes largely to any lack of new posters or threads that you might see. I know I personally refrain from posting theories or models I have that are counter-intuitive and would actually start good discussions specifically for this reason...and I put them on the LessWrong Facebook page or bring them up at Meetups instead, where I’ve had some great conversations and made some good friends because of it.”
For the past couple of months, I’ve found the Facebook LW group to debate more interesting subjects than the LW website. But that’s only my appraisal of what’s interesting and what’s not.
People in the other category, who are misunderstood or disagreed with, but who discuss constructively and rationally, are the ones who MOST need to be able to speak.
Do such people usually get downvoted on LW? Outside of this one downvote stalker, that is.
(This is a separate question from whether or not people think they’ll get downvoted for constructive disagreement, which is also important.)
Well I’m sure each person downvotes for their own reasons, but I have noticed several people who, when they are disagreeing with someone, tend to have a consistent series of “-1” votes showing up on the posts of the person with whom they’re disagreeing.
If they are doing what it seems, I would say this is an example of the problem. Downvoting also allows people to express disagreement without having to give reasons or even pay much attention to what’s said. I think this also goes against the purpose of the site.
Downvoting also allows people to express disagreement without having to give reasons [...] I think this also goes against the purpose of the site.
Maybe not disagreement as such, but it’s very often good to express disapproval without detailing the reasons for it. The basic issue here is that a response increases visibility (more, in fact, than an upvote does), and you generally don’t want to make things you disapprove of more visible.
The classic example would be deliberate trolling, where a lovingly crafted response detailing everything that’s wrong with the post is precisely what you don’t want: it wastes your time and encourages the troll. But it’s not much different for incoherent crankery or political diatribes or cat pictures: the author might not be encouraged by a response, but you’re still wasting other people’s time as long as the thread’s clogging up Recent Comments.
That said, while I don’t feel that downvoting your conversational partners to express disapproval is an abuse of the system in the same way that block downvoting is, I do think it’s a bad idea and wouldn’t be opposed to a feature limiting it.
I think you’re definitely right that we need to be able to control people who stop the site from being an honest exchange of ideas or good-faith discussion. It might be better to have a button to report trolling, flaming or spamming, but not an all-purpose downvote that might be used for other reasons.
The example I think about is a Religious Forum. If they had a “downvoting” feature that was implemented in the same way that the Less Wrong feature is...anyone showing up who asks too many skeptical questions could just be downvoted out of existence without anyone answering their arguments.
Perhaps this demonstrates how it could be an Anti-Rational tool or encourage groupthink...which I think is dangerous.
The example I think about is a Religious Forum. If they had a “downvoting” feature that was implemented in the same way that the Less Wrong feature is...anyone showing up who asks too many skeptical questions could just be downvoted out of existence without anyone answering their arguments.
Bidirectional voting has its disadvantages, but I don’t think this is one of them.
Sure, if you get a seed culture that’s skewed enough in one direction, karma-like systems can be used to enforce conformity with it. But that’s hardly unique; if you wander into a LiveJournal (a voteless format) or a Facebook discussion (a unidirectional format) and start spouting off opinions outside the local Overton window, you’ll quickly find yourself getting shouted down. There’s no purely technical way I know of to break what I’ll politely describe as an ideological consensus cluster.
That being the case, I find myself thinking more of the incentives karma creates in an ideologically mixed environment that values things other than conformity, like clarity and originality. Sure, offending someone’s ideology is risky; but people on the other side aren’t mindless political monsters, they care about those other values as much as you do, and if you respect them you won’t get many downvotes. But ignore those norms to dribble content-free “hooray for our side”, and the best you can hope for is a few upvotes from people suffering from halo effects.
What happens if you don’t have the option of downvoting? Well, suddenly it doesn’t matter what your opponents think, since they can’t effectively punish you for it. People don’t stop caring about discourse norms, they still have the same reactions to following them that they always did, but the thing is that being clever and polite and original is hard; it takes effort and care and some facility with the language. Repeating buzzwords for a few safe upvotes from true believers, on the other hand, doesn’t. Stripped of downside, that’s what people are going to fall back on—which of course leads to a self-perpetuating cycle of radicalization.
(Twitter and Tumblr make salient examples, although they both have other issues going on. Open Facebook comment threads are a somewhat purer case.)
It might be better to have a button to report trolling, flaming or spamming
And if some user decides to use this button to report all comments of the people with different political opinion, then what? Would it then be acceptable to ban the user, because they abused the button? Well, they are abusing the downvote button now.
At some moment you just have to use the banhammer. It could as well be now.
Trolling/flaming/spamming report buttons are clearly labeled for their purpose. The downvote button isn’t.
To Nornagest,
Here’s the big difference: On Facebook, you can’t stop OTHER people from seeing what the person has to say, no matter how much you scream at them. With the system here, you can. Their posts will be hidden and they can even lose their posting privileges when they are downvoted. And when I say that’s a big difference, I mean that’s a BIG difference. Again, think of the religious forum. This same karma system would allow them to literally stop you from speaking to or influencing people who are on the fence or more open to rationality, instead of just posting replies that highlight their own immaturity or irrationality. I think the issue here is clear.
Secondly, when you refer to (I presume) LessWrong as “an ideologically-mixed environment that values things other than conformity,” you’re assuming that everyone here views it that way. If everyone saw the downvote button in the same idealized form, we wouldn’t have a problem. The issue is that the downvote button does not have such a clear and apparent definition, and there doesn’t appear to be any actual enforced policy by the LessWrong admins to stop people from using the downvote button to simply express disagreement.
On Facebook, you can’t stop OTHER people from seeing what the person has to say, no matter how much you scream at them. With the system here, you can.
Can’t you? Eliezer cites the easiness of clicking a button and making the other person Go Away as a major perceived advantage of posting on FB rather than LW. And even if you downvote someone on LW, well, someone can undo that with an upvote.
Hi gwern, I’m not sure exactly what you mean. In Facebook groups, you can ignore someone, but the person in question can still participate in discussions that don’t involve you, or discuss what you’ve said outside of your own threads. I think this is actually a good thing, since it lets you avoid unconstructive people, but doesn’t allow you to censor people from being heard by others if that person has something valuable to add.
Regarding downvoting vs upvoting, counteracting mass downvoters (who apparently have gone to the extent of downvoting someone over 1000 times) is a huge burden on other people and not something they should have to do.
In Facebook groups, you can ignore someone, but the person in question can still participate in discussions that don’t involve you, or discuss what you’ve said outside of your own threads.
I believe Eliezer was referring to starting posts. So the question is, which is better, a banhappy omnipotent OP or gradual undoable community moderation?
counteracting mass downvoters (who apparently have gone to the extent of downvoting someone over 1000 times) is a huge burden on other people and not something they should have to do.
And indeed, it’s not something that happens often. Eugine is so far the only person to be banned for mass downvoting in the ~5 year history of a very active site.
Sorry, didn’t see this until now. In future, it works better if you put responses to a post under that post; I’m not alerted if you respond to me in another branch of the thread.
Secondly, when you refer to (I presume) LessWrong as “an ideologically-mixed environment that values things other than conformity,” you’re assuming that everyone here views it that way. If everyone saw the downvote button in the same idealized form, we wouldn’t have a problem.
I’m presuming no such thing; I was talking about the composition of LW, not the purpose of the downvote button. People’s personal downvote policies are going to vary (quite a bit, really), but as long as the forum as a whole contains people with a mix of values similar to those I mentioned, their votes are going to average out to something like the behavior I described: some votes for conformity, some for contrarianism, some for unrelated norms. Note however that this doesn’t take into account retributive downvoting; there needs to be policy in place to deal with that, but hey! Now there is, and we’ve just seen it in action.
The visibility effects of karma, I suspect, are overrated as a driver of behavior except in the case of top-level posts (where they’re taken off most of the interface and become something of a pain to get to): leaving that “downvoted below threshold” notification seems to incite people’s curiosity as much as anything. Some of my highest-ranked posts are replies to comments below the threshold; they wouldn’t have gotten there if people weren’t reading the thread.
The karma toll for replying to heavily downvoted comments does shape behavior, but I’ve only seen one person get that low for politely expressing political views, and he was a white supremacist.
Hi Nornagest, I’m used to forums with a multi-quote feature. I wasn’t aware it wouldn’t notify you if I just replied to the bottom comment.
I’m presuming no such thing; I was talking about the composition of LW, not the purpose of the downvote button. People’s personal downvote policies are going to vary (quite a bit, really), but as long as the forum as a whole contains people with a mix of values similar to those I mentioned, their votes are going to average out to something like the behavior I described: some votes for conformity, some for contrarianism, some for unrelated norms.
This doesn’t work in practice precisely because mass and retributive downvoting are disproportionately effective. One person with a skewed concept of downvoting can outweigh tons of other people who are using the functions as intended. I might vote up a comment by someone I like, but I’m not going to go through their profiles and give them hundreds (or even thousands) of upvotes, while we’ve seen the downvote-abusers do exactly this. So they won’t average out properly.
The visibility effects of karma, I suspect, are overrated as a driver of behavior except in the case of top-level posts (where they’re taken off most of the interface and become something of a pain to get to): leaving that “downvoted below threshold” notification seems to incite people’s curiosity as much as anything. Some of my highest-ranked posts are replies to comments below the threshold; they wouldn’t have gotten there if people weren’t reading the thread.
We don’t have a lot of clear data on this because an “ugh field” or people refraining from posting are often an invisible cost. I’ve had several times that I had a notion that I wanted to post about here, even considering an entire sequence or at least largely new area of discussion, then thought of some of this type of behavior and changed my mind.
Even if the “downvote below threshold” might incite curiosity, the person in question still loses privileges on site. Lastly, the Eugine_Nier news is quite encouraging and may indicate some solutions to this issue.
This doesn’t work in practice precisely because mass and retributive downvoting are disproportionately effective. [...] So they won’t average out properly.
See the next sentence of my comment.
Even if the “downvote below threshold” might incite curiosity, the person in question still loses privileges on site.
That’s a very different case. Downvoting a person into losing privileges can by done by a single user if the target’s posted a lot of marginal or controversial comments, but unless they’re very new it takes a lot of patience or a downvote script (Eugine seems to have been using patience), and AFAICT most people have karma ratios high enough that it’d take sockpuppets or other abuses that could be targeted by narrower rules. I only know of one illegitimate case, although others may emerge as the consequences of Eugine’s behavior become more apparent. Conversely, downvoting a post below the visibility threshold is much more common but can’t be done by a single user.
Yes, but I feel that problem nullifies the paragraph.
That’s a very different case. Downvoting a person into losing privileges can by done by a single user if the target’s posted a lot of marginal or controversial comments, but unless they’re very new it takes a lot of patience or a downvote script (Eugine seems to have been using patience), and AFAICT most people have karma ratios high enough that it’d take sockpuppets or other abuses that could be targeted by narrower rules.
I would have agreed that the patience required is a barrier, until I found out about the 1000 vote attacks. Also, even giving someone a smaller amount of downvotes can become a problem if it’s disproportionate to the upvotes. Such as downvoting the person’s last 30-50 comments. It simply requires a larger number of people to be doing it. When there was no indication that there would be mass downvote moderating, I actually downvoted Eugine several times in a row out of annoyance when I realized what he was doing to other people...since I figured there was no other option to control it.
Anyway, it may be of course that Eugine is the first person to be outed for this behavior and it will become a regular thing. In which case this issue may cease to be a problem at all.
Eugine may be the only person to have (recently) been using this as a tool of policy, aside from a couple people downvoting him in retribution. If you look at the patterns of people targeted for retributive downvoting (here, here, and here, plus this thread and its relatives), most of the situations seem to fit his MO and apparent set of grievances. Perhaps most tellingly, I don’t know of anyone besides Eugine himself who’s been mass-downvoted by two users (which is easy to tell from karma on obscure or unremarkable posts).
(I’m not sure about Will_Newsome, but that was three years ago.)
For what it’s worth, I haven’t noticed that myself, and I don’t think it’s ever happened to me here. But I agree that when it happens, it’s an example of the problem you’re talking about.
Downvoting also allows people to express disagreement without having to give reasons or even pay much attention to what’s said. I think this also goes against the purpose of the site.
I agree with this too. I think maybe we have just different intuitions of how commonly it’s actually used like that.
I’ve commented also that the karma system, as it is currently, causes less participation on the site. Just to save time I’ll paste it here.
“The fundamental flaw that I see with LessWrong’s main site is that its karma/moderating system has the effect of silencing and banning people for being disagreed with or misunderstood. This is a major problem. You cannot mix “I don’t agree with you” or “I don’t understand you” with “you will be punished and silenced.”
People who spam, flame, or otherwise destroy conversation are the ones who need to be silenced, ignored or banned, and a lot of sites like Facebook have separate buttons to perform exactly that function. People in the other category, who are misunderstood or disagreed with, but who discuss constructively and rationally, are the ones who MOST need to be able to speak. I think the punishment and silencing, and the threat of it, contributes largely to any lack of new posters or threads that you might see. I know I personally refrain from posting theories or models I have that are counter-intuitive and would actually start good discussions specifically for this reason...and I put them on the LessWrong Facebook page or bring them up at Meetups instead, where I’ve had some great conversations and made some good friends because of it.”
Are you seriously implying that the facebook group for LessWrong has better discussions than the site? I can’t say that I agree.
For the past couple of months, I’ve found the Facebook LW group to debate more interesting subjects than the LW website. But that’s only my appraisal of what’s interesting and what’s not.
Good criticism is frequently upvoted on LW. But overall, I agree with you that this is an issue.
Do such people usually get downvoted on LW? Outside of this one downvote stalker, that is.
(This is a separate question from whether or not people think they’ll get downvoted for constructive disagreement, which is also important.)
Well I’m sure each person downvotes for their own reasons, but I have noticed several people who, when they are disagreeing with someone, tend to have a consistent series of “-1” votes showing up on the posts of the person with whom they’re disagreeing.
If they are doing what it seems, I would say this is an example of the problem. Downvoting also allows people to express disagreement without having to give reasons or even pay much attention to what’s said. I think this also goes against the purpose of the site.
Maybe not disagreement as such, but it’s very often good to express disapproval without detailing the reasons for it. The basic issue here is that a response increases visibility (more, in fact, than an upvote does), and you generally don’t want to make things you disapprove of more visible.
The classic example would be deliberate trolling, where a lovingly crafted response detailing everything that’s wrong with the post is precisely what you don’t want: it wastes your time and encourages the troll. But it’s not much different for incoherent crankery or political diatribes or cat pictures: the author might not be encouraged by a response, but you’re still wasting other people’s time as long as the thread’s clogging up Recent Comments.
That said, while I don’t feel that downvoting your conversational partners to express disapproval is an abuse of the system in the same way that block downvoting is, I do think it’s a bad idea and wouldn’t be opposed to a feature limiting it.
I think you’re definitely right that we need to be able to control people who stop the site from being an honest exchange of ideas or good-faith discussion. It might be better to have a button to report trolling, flaming or spamming, but not an all-purpose downvote that might be used for other reasons.
The example I think about is a Religious Forum. If they had a “downvoting” feature that was implemented in the same way that the Less Wrong feature is...anyone showing up who asks too many skeptical questions could just be downvoted out of existence without anyone answering their arguments.
Perhaps this demonstrates how it could be an Anti-Rational tool or encourage groupthink...which I think is dangerous.
Bidirectional voting has its disadvantages, but I don’t think this is one of them.
Sure, if you get a seed culture that’s skewed enough in one direction, karma-like systems can be used to enforce conformity with it. But that’s hardly unique; if you wander into a LiveJournal (a voteless format) or a Facebook discussion (a unidirectional format) and start spouting off opinions outside the local Overton window, you’ll quickly find yourself getting shouted down. There’s no purely technical way I know of to break what I’ll politely describe as an ideological consensus cluster.
That being the case, I find myself thinking more of the incentives karma creates in an ideologically mixed environment that values things other than conformity, like clarity and originality. Sure, offending someone’s ideology is risky; but people on the other side aren’t mindless political monsters, they care about those other values as much as you do, and if you respect them you won’t get many downvotes. But ignore those norms to dribble content-free “hooray for our side”, and the best you can hope for is a few upvotes from people suffering from halo effects.
What happens if you don’t have the option of downvoting? Well, suddenly it doesn’t matter what your opponents think, since they can’t effectively punish you for it. People don’t stop caring about discourse norms, they still have the same reactions to following them that they always did, but the thing is that being clever and polite and original is hard; it takes effort and care and some facility with the language. Repeating buzzwords for a few safe upvotes from true believers, on the other hand, doesn’t. Stripped of downside, that’s what people are going to fall back on—which of course leads to a self-perpetuating cycle of radicalization.
(Twitter and Tumblr make salient examples, although they both have other issues going on. Open Facebook comment threads are a somewhat purer case.)
And if some user decides to use this button to report all comments of the people with different political opinion, then what? Would it then be acceptable to ban the user, because they abused the button? Well, they are abusing the downvote button now.
At some moment you just have to use the banhammer. It could as well be now.
To Viliam,
Trolling/flaming/spamming report buttons are clearly labeled for their purpose. The downvote button isn’t.
To Nornagest,
Here’s the big difference: On Facebook, you can’t stop OTHER people from seeing what the person has to say, no matter how much you scream at them. With the system here, you can. Their posts will be hidden and they can even lose their posting privileges when they are downvoted. And when I say that’s a big difference, I mean that’s a BIG difference. Again, think of the religious forum. This same karma system would allow them to literally stop you from speaking to or influencing people who are on the fence or more open to rationality, instead of just posting replies that highlight their own immaturity or irrationality. I think the issue here is clear.
Secondly, when you refer to (I presume) LessWrong as “an ideologically-mixed environment that values things other than conformity,” you’re assuming that everyone here views it that way. If everyone saw the downvote button in the same idealized form, we wouldn’t have a problem. The issue is that the downvote button does not have such a clear and apparent definition, and there doesn’t appear to be any actual enforced policy by the LessWrong admins to stop people from using the downvote button to simply express disagreement.
Can’t you? Eliezer cites the easiness of clicking a button and making the other person Go Away as a major perceived advantage of posting on FB rather than LW. And even if you downvote someone on LW, well, someone can undo that with an upvote.
Hi gwern, I’m not sure exactly what you mean. In Facebook groups, you can ignore someone, but the person in question can still participate in discussions that don’t involve you, or discuss what you’ve said outside of your own threads. I think this is actually a good thing, since it lets you avoid unconstructive people, but doesn’t allow you to censor people from being heard by others if that person has something valuable to add.
Regarding downvoting vs upvoting, counteracting mass downvoters (who apparently have gone to the extent of downvoting someone over 1000 times) is a huge burden on other people and not something they should have to do.
I believe Eliezer was referring to starting posts. So the question is, which is better, a banhappy omnipotent OP or gradual undoable community moderation?
And indeed, it’s not something that happens often. Eugine is so far the only person to be banned for mass downvoting in the ~5 year history of a very active site.
Sorry, didn’t see this until now. In future, it works better if you put responses to a post under that post; I’m not alerted if you respond to me in another branch of the thread.
I’m presuming no such thing; I was talking about the composition of LW, not the purpose of the downvote button. People’s personal downvote policies are going to vary (quite a bit, really), but as long as the forum as a whole contains people with a mix of values similar to those I mentioned, their votes are going to average out to something like the behavior I described: some votes for conformity, some for contrarianism, some for unrelated norms. Note however that this doesn’t take into account retributive downvoting; there needs to be policy in place to deal with that, but hey! Now there is, and we’ve just seen it in action.
The visibility effects of karma, I suspect, are overrated as a driver of behavior except in the case of top-level posts (where they’re taken off most of the interface and become something of a pain to get to): leaving that “downvoted below threshold” notification seems to incite people’s curiosity as much as anything. Some of my highest-ranked posts are replies to comments below the threshold; they wouldn’t have gotten there if people weren’t reading the thread.
The karma toll for replying to heavily downvoted comments does shape behavior, but I’ve only seen one person get that low for politely expressing political views, and he was a white supremacist.
Hi Nornagest, I’m used to forums with a multi-quote feature. I wasn’t aware it wouldn’t notify you if I just replied to the bottom comment.
This doesn’t work in practice precisely because mass and retributive downvoting are disproportionately effective. One person with a skewed concept of downvoting can outweigh tons of other people who are using the functions as intended. I might vote up a comment by someone I like, but I’m not going to go through their profiles and give them hundreds (or even thousands) of upvotes, while we’ve seen the downvote-abusers do exactly this. So they won’t average out properly.
We don’t have a lot of clear data on this because an “ugh field” or people refraining from posting are often an invisible cost. I’ve had several times that I had a notion that I wanted to post about here, even considering an entire sequence or at least largely new area of discussion, then thought of some of this type of behavior and changed my mind.
Even if the “downvote below threshold” might incite curiosity, the person in question still loses privileges on site. Lastly, the Eugine_Nier news is quite encouraging and may indicate some solutions to this issue.
See the next sentence of my comment.
That’s a very different case. Downvoting a person into losing privileges can by done by a single user if the target’s posted a lot of marginal or controversial comments, but unless they’re very new it takes a lot of patience or a downvote script (Eugine seems to have been using patience), and AFAICT most people have karma ratios high enough that it’d take sockpuppets or other abuses that could be targeted by narrower rules. I only know of one illegitimate case, although others may emerge as the consequences of Eugine’s behavior become more apparent. Conversely, downvoting a post below the visibility threshold is much more common but can’t be done by a single user.
Yes, but I feel that problem nullifies the paragraph.
I would have agreed that the patience required is a barrier, until I found out about the 1000 vote attacks. Also, even giving someone a smaller amount of downvotes can become a problem if it’s disproportionate to the upvotes. Such as downvoting the person’s last 30-50 comments. It simply requires a larger number of people to be doing it. When there was no indication that there would be mass downvote moderating, I actually downvoted Eugine several times in a row out of annoyance when I realized what he was doing to other people...since I figured there was no other option to control it.
Anyway, it may be of course that Eugine is the first person to be outed for this behavior and it will become a regular thing. In which case this issue may cease to be a problem at all.
Eugine may be the only person to have (recently) been using this as a tool of policy, aside from a couple people downvoting him in retribution. If you look at the patterns of people targeted for retributive downvoting (here, here, and here, plus this thread and its relatives), most of the situations seem to fit his MO and apparent set of grievances. Perhaps most tellingly, I don’t know of anyone besides Eugine himself who’s been mass-downvoted by two users (which is easy to tell from karma on obscure or unremarkable posts).
(I’m not sure about Will_Newsome, but that was three years ago.)
For what it’s worth, I haven’t noticed that myself, and I don’t think it’s ever happened to me here. But I agree that when it happens, it’s an example of the problem you’re talking about.
I agree with this too. I think maybe we have just different intuitions of how commonly it’s actually used like that.
You probably have more experience than I do with how people as a whole do the voting. I’m just concerned with potential problems.
In my personal experience, I have posted things that are quite critical of LW ideas, but if I show I’ve done my homework they get upvotes.