If dunbar's limit is the issue, if you feel the community is too large to feel like a community, or if you feel that the community has become a granfalloon, not as united as it might like to think, there's a fairly straightforward solution. Controlled mitosis. Demarcate houses, build ourselves a sorting hat(possibly in the form of a quiz although asking meaningful questions in these things is much harder than you might imagine), send people on to their respective common rooms and tell them to make friends with everyone.
It would be deleterious to demarcate the houses with ideological boundaries. Better that they're drawn along practical specializations to aid in developing rationalist outlooks in specific fields of praxis.
Demarcate houses, build ourselves a sorting hat(possibly in the form of a quiz although asking meaningful questions in these things is much harder than you might imagine), send people on to their respective common rooms and tell them to make friends with everyone.
I actually think this is an excellent idea, especially if it could be written in time for the end of HPMOR. Additionally, it could combine with the yearly census Yvain does. Pre-existing questionnaires could be used partially as well.
One possibility would be to do the questionnaire first, and then run a clustering algorithm over it to determine the groupings.
The biggest problem I can see is that this might make LW look even sillier then it already does.
As much as importing themes from the HP universe sounds fun, I don't think anything as fanciful as that would work for us. If the system works, it's unlikely that the splits we end up with would seem silly.
I think it'd be interesting to make a system that tries to cluster people into small, somewhat arbitrary groups of 8-10 people, under the expectation that everyone will get to know everyone else and nobody will slip between the cracks. A group any larger than that can have little in the way of intimacy.
I think a lot of this is a technological issue. Build the right system of sorting hats and virtual common rooms, and it'll just happen. Until you do, it can't happen.
it's unlikely that the splits we end up with would seem silly.
I was rather meaning that the whole sorting-hat thing is pretty silly.
I think a lot of this is a technological issue.
I'm never run an online questionnaire, but I imagine there are simple out of the box ways to do it. Similerley, I believe that LW's code is adapted from reddit, and so I would guess it would be easy enough to add the code for subreddits. The clustering algorithms are simple too and have many open-source implmentations.
In fact, just splitting discussion into, say 'technical maths', 'rationality', 'singularity stuff', 'lifehacks', 'fun stuff' could be an improvement maybe?
I guess there's two problems here: community building and useful work.
As far as useful work goes, its probably inevitable and for the best that it moves elsewhere, especially things that will scare off new people, such as highly technical maths and all politics.
As far as community building goes, perhaps this should be treated as orthogonal? Incidentally, if anyone wants to play dungeons & discourse or over-analysed game-theory diplomacy I'd be up for a game.
On a personal note, I ran into transhumanism, and SIAI and EY over a decade ago, but I didn't think I was smart enough to contribute anything, so I didn't get involved. Its only recently that I decided I was ready to have something useful to say, and I'm kinda annoyed that I've got involved just as the community seems to be dying.
I suppose part of this is that I want the hipster street cred of joining "before it was cool" and that maybe this isn't the most productive attitude.
An old joke;
"What bar should we go to?"
"Oh, I like Bar X"
"No-one goes there anymore, it's too crowded"
I suspect that LessWrong is getting too crowded for many people to come here. We're well above Dunbar's number, and most of us can't get to know the other active users without a large investment of time. Other than Eliezer, and maybe a couple of other very frequent past contributors (Yvain, Luke, Anna) the site seems fractured - there are too many people to keep track of socially, and the local groups are kind of doing their own thing. The early heavy users are largely gone (as noted by inferential, http://lesswrong.com/lw/le5/welcome_to_less_wrong_7th_thread_december_2014/bsyj) and the site has split; MIRI, CFAR, and related work seem to have (usefully) siphoned off much of the energy that used to go into the site. The influx of HPMOR readers continues, and I suspect will spike again after it finishes, but Lesswrong is beginning to seem more useful as an archive for the sequences, and an extended chat group, than a source of new and useful ideas.
I think there is a simple reason: either you're very in the know, or you can't participate much. The bar for participating has gotten higher as the community has evolved and become well versed in the basics; now there is less to say that's accessible without a heavy background, and many of those things are technical, deep, or uninteresting to casual readers. Adding to this, it seems that if you're very in the know, the utility of the site goes down significantly (or other options open up,) so you eventually leave - to better things, probably. I'm not suggesting that this is an unmitigated bad thing, but I'd like to hear what others think about the idea that LessWrong as a community has largely outgrown this forum - it's useful as a touchstone for meetups and reference to older material, but not nearly as useful as a living site.
What say you? Am I wrong? Are there things we can or should do as a community to change this, or should we try to continue to transition to more meetups, and to MIRI/CFAR-oriented endeavors?