Various people raised concerns that growth might ruin the culture after reading my "LessWrong could grow a lot" thread. There has been some discussion about whether endless September, a phenomenon that kills online discussion groups, is a significant threat to LessWrong and what can be done. I really care about it, so I volunteered to code a solution myself for free if needed. Luke invited debate on the subject (the debate is here) and will be sent the results of this poll and asked to make a decision. It was suggested by him in an email that I wait a little while and then post my poll (meta threads are apparently annoying to some, so we let people cool off). Here it is, preceded by a Cliff's notes summary of the concerns.
Why this is worth your consideration:
- Yvain and I checked the IQ figures in the survey against other data this time, and the good news is that it's more believable that the average LessWronger is gifted. The bad news is that LessWrong's IQ average has decreased on each survey. It can be argued that it's not decreasing by a lot or we don't have enough data, but if the data is good, LessWrong's average has lost 52% of it's giftedness since March of 2009.
- Eliezer documented the arrival of poseurs (people who superficially copycat cultural behaviors - they are reported to over-run subcultures) which he termed "Undiscriminating Skeptics".
- Efforts to grow LessWrong could trigger an overwhelming deluge of newbies.
- LessWrong registrations have been increasing fast and it's possible that growth could outstrip acculturation capacity. (Chart here)
- The Singularity Summit appears to cause a deluge of new users that may have similar effect to the September deluges of college freshman that endless September is named after. (This chart shows a spike correlated with the 2011 summit where 921 users joined that month, which is roughly equal to the total number of active users LW tends to have in a month if you go by the surveys or Vladmir's wget.)
- A Slashdot effect could result in a tsunami of new users if a publication with lots of readers like the Wall Street Journal (they used LessWrong data in this article) decides to write an article on LessWrong.
- The sequences contain a lot of the culture and are long meaning that "TLDR" may make LessWrong vulnerable to cultural disintegration. (New users may not know how detailed LW culture is or that the sequences contain so much culture. I didn't.)
- Eliezer said in August that the site was "seriously going to hell" due to trolls.
- A lot of people raised concerns.
Two Theories on How Online Cultures Die:
Overwhelming user influx.
There are too many new users to be acculturated by older members, so they form their own, larger new culture and dominate the group.
Trending toward the mean.
A group forms because people who are very different want a place to be different together. The group attracts more people that are closer to mainstream than people who are equally different because there are more mainstream people than different people. The larger group attracts people who are even less different in the original group's way for similar reasons. The original group is slowly overwhelmed by people who will never understand because they are too different.
Poll Link:
Request for Feedback:
In addition to constructive criticism, I'd also like the following:
-
Your observations of a decline or increase in quality, culture or enjoyment at LessWrong, if any.
-
Ideas to protect the culture.
-
Ideas for tracking cultural erosion.
- Ways to test the ideas to protect the culture.
It has occurred to me to wonder whether the poll might be biased. I wanted to add a summary of things that protect LessWrong against endless September when I wrote this post. However, I couldn't think of even one. I figured my thread to debate whether we should have better protection would have turned up any compelling reasons to think LessWrong is protected but it didn't.
I became curious about this just now wondering whether there really isn't a single reason to think that LessWrong is protected, and I re-read all of the comments (though not the replies to the comments) to see if I had forgotten any. This comment by AndrewHickey was the closest thing I found to an argument that there is something protecting LessWrong:
The registration numbers showed that LessWrong is gaining members fast, so the echo chamber idea does not appear to be supported.
As for the "steep learning curve" idea, the 2012 Survey Results show that only 1/4 of the survey respondents have read the Sequences, and that 60% of those who have participated either have not read them or have not finished them. Considering that the majority of participants haven't finished the sequences, I think LessWrong's steep learning curve is more likely to add to the risk than to have any protective benefits because if most people are going "Your culture is TLDR, I'm commenting anyway." then they're going to be participating without all the cultural knowledge.
One reflex is to think that the current karma system will protect LessWrong against endless September but thinking about that strategy further, one realizes that there is a limit to how many new posts older users can read and vote on, so this would not help if there were enough new users or users closer to the mean to overwhelm their voting capacity.
As far as I can tell, there's currently nothing that is likely to protect LessWrong from eternal September.