I disagree that posts from WrongBot, or others with similar failings in reasoning . . . pose a serious risk to LessWrong as a community.
How much experience have you had watching the trajectory of online communities?
Have you for example informed yourself of the case of Reddit (the original one) which is particularly relevant to this community in that the software is so similar?
I have not, but Paul Graham has (since he was an investor in Reddit) and he has stated many times that he believes that his community, Hacker News, is in constant danger of falling prey to the dynamic that rendered Reddit worthless to thoughtful busy people, and he has taken many different measures, including banning a user relatively frequently, denying new users the right to cast downvotes—or any votes at all if their karma is low enough—and disappearing the “reply” link on certain posts based on an algorithm.
How much experience have you had watching the trajectory of online communities?
Is anyone aware of any good write-ups on this topic? I’d be interested in seeing any insights as to why things happen the way they do, and what we can do to improve matters.
To answer my own question, here are a few write-ups I found about why the quality of an online community tends to decline over time, and what can be done about this problem:
Also, the book “The Virtual Community” that Richard Hollerith mentioned is available online, although I wasn’t able to find much information in it about the specific topic at hand.
Howard Reingold’s book with the string “Virtual Community” in it. Old though: 1994 or so, but very informative on pre-Internet communities. The chapter on France’s Minitel (term?) I found particularly valuable.
I’ve been participating in online communities since 1992, and most of my information has come from short comments by people trying to preserve the character of specific communities. Paul Graham’s comments on Hacker News are particularly worthwhile, but have not been collected in any one place.
How much experience have you had watching the trajectory of online communities?
Have you for example informed yourself of the case of Reddit (the original one) which is particularly relevant to this community in that the software is so similar?
I have not studied this with any rigor, although I have seen communities that I previously enjoyed enter periods of decline (sometimes recovering at a later point, sometimes not). I don’t disagree that with online communities, there is often some tipping point when the bad reasoning/noise outweighs the good. That’s why I also made this part of my comment:
From my perspective, I don’t think we’ve reached the point where LW is so crowded with posts that good ideas and posts are being crowded out by bad ones.
Perhaps I’m wrong about this. At any rate, if LW is actually in a serious period of decline, the problem is more serious than just WrongBot, and I disagree with implementing a solution where individual posters take it upon themselves to ask other posters to leave. (If EY wants to create some sort of system like Paul Graham’s or make new moderators with these sorts powers, that would be different in my view than this sort of ad hoc approach, which doesn’t seem likely to work (due to both its ad hoc nature and unenforceability) and also presents greater risks of abuse, decisions based on personality conflict, etc.)
if LW is actually in a serious period of decline, the problem is more serious than just WrongBot
Agreed. In particular, LW has successfully weathered long flurries of comments and posts by people worse than WrongBot.
The primary sign that LW is in danger of becoming the kind of place that I and those I admire no longer want to visit is the (negative) magnitude of the score on comments asking WrongBot to stop writing on things beyond his skill and the (positive) magnitude of the scores of WrongBot’s replies to those (negatively scored) comments. That is new.
Note that the vast majority of readers of LW never attempt to create evolutionary arguments relevant to human behavior or summarize novel arguments made by others. I would hope that that is because they realize that it is too difficult for them.
Nobody can downvote on Hacker News. The only vaguely analogous function is “flag” which leads to posts (not comments) being killed or marked for killing.
(Edit:rhollerithdotcom points out correctly that this is only true for submissions and that above a karma threshold comments are downvotable)
Nobody can downvote on Hacker News. The only vaguely analogous function is “flag”
If one has enough karma (ISTR the threshhold being 200 points at one point, though it has probably been raised a few times since then) one can downvote comments or else how to explain the presence of comments with negative scores in almost every comment section.
You might be right about top-level submissions though.
How much experience have you had watching the trajectory of online communities?
Have you for example informed yourself of the case of Reddit (the original one) which is particularly relevant to this community in that the software is so similar?
I have not, but Paul Graham has (since he was an investor in Reddit) and he has stated many times that he believes that his community, Hacker News, is in constant danger of falling prey to the dynamic that rendered Reddit worthless to thoughtful busy people, and he has taken many different measures, including banning a user relatively frequently, denying new users the right to cast downvotes—or any votes at all if their karma is low enough—and disappearing the “reply” link on certain posts based on an algorithm.
Is anyone aware of any good write-ups on this topic? I’d be interested in seeing any insights as to why things happen the way they do, and what we can do to improve matters.
To answer my own question, here are a few write-ups I found about why the quality of an online community tends to decline over time, and what can be done about this problem:
A Group Is Its Own Worst Enemy by Clay Shirky
What I’ve Learned from Hacker News by Paul Graham
Trolls by Paul Graham
Well-Kept Gardens Die By Pacifism by Eliezer Yudkowsky
Also, the book “The Virtual Community” that Richard Hollerith mentioned is available online, although I wasn’t able to find much information in it about the specific topic at hand.
Howard Reingold’s book with the string “Virtual Community” in it. Old though: 1994 or so, but very informative on pre-Internet communities. The chapter on France’s Minitel (term?) I found particularly valuable.
I’ve been participating in online communities since 1992, and most of my information has come from short comments by people trying to preserve the character of specific communities. Paul Graham’s comments on Hacker News are particularly worthwhile, but have not been collected in any one place.
I have not studied this with any rigor, although I have seen communities that I previously enjoyed enter periods of decline (sometimes recovering at a later point, sometimes not). I don’t disagree that with online communities, there is often some tipping point when the bad reasoning/noise outweighs the good. That’s why I also made this part of my comment:
Perhaps I’m wrong about this. At any rate, if LW is actually in a serious period of decline, the problem is more serious than just WrongBot, and I disagree with implementing a solution where individual posters take it upon themselves to ask other posters to leave. (If EY wants to create some sort of system like Paul Graham’s or make new moderators with these sorts powers, that would be different in my view than this sort of ad hoc approach, which doesn’t seem likely to work (due to both its ad hoc nature and unenforceability) and also presents greater risks of abuse, decisions based on personality conflict, etc.)
Agreed. In particular, LW has successfully weathered long flurries of comments and posts by people worse than WrongBot.
The primary sign that LW is in danger of becoming the kind of place that I and those I admire no longer want to visit is the (negative) magnitude of the score on comments asking WrongBot to stop writing on things beyond his skill and the (positive) magnitude of the scores of WrongBot’s replies to those (negatively scored) comments. That is new.
Note that the vast majority of readers of LW never attempt to create evolutionary arguments relevant to human behavior or summarize novel arguments made by others. I would hope that that is because they realize that it is too difficult for them.
Nobody can downvote on Hacker News. The only vaguely analogous function is “flag” which leads to posts (not comments) being killed or marked for killing.
(Edit:rhollerithdotcom points out correctly that this is only true for submissions and that above a karma threshold comments are downvotable)
Useful essay on online communities http://www.kuro5hin.org/story/2009/3/12/33338/3000
If one has enough karma (ISTR the threshhold being 200 points at one point, though it has probably been raised a few times since then) one can downvote comments or else how to explain the presence of comments with negative scores in almost every comment section.
You might be right about top-level submissions though.