An idea: a rationality hackathon.
From what I see, it seems like rationalists don't act on ideas often enough. To help people get the motivation to act on ideas, I sense that a hackathon would be effective. People would talk and group together to prototype different ideas, and at the end of the hackathon, participants would vote on the best ideas, and hopefully this would spark some action.
I guess what makes this different from the typical hackathon is:
1) Participants would be Less Wrong readers, or people part of other rationality-minded communities.
2) The goal would be to start things that are as beneficial as possible to the world (people at hackathons usually just want to build something "cool").
Thoughts?
The more complex a project is, the less likely are people to complete it successfully. The more meta a project is, the less likely is the model of its usefulness correct.
I expect a risk that people will vote on something very nebulous (more meta! more meta!) as the best idea, and at the end pretty much nothing measurable happens.
To avoid going too much meta, I would recommend adding an artificial constraint, such as "we must be able to complete the whole thing in one week". Sure, this drastically limits how useful things you can do. On the other hand, it allows multiple iterations and feedback.
You know the drill - If it's worth saying, but not worth its own post (even in Discussion), then it goes here.