If you were creating a rationality-increasing org, how would you design it? Focus on whatever properties you care about.

New Answer
New Comment

2 Answers sorted by



FYI, I think this is a lot harder than it sounds. A lot of people take a look at CFAR and think "hmm, they don't seem to be doing the obvious things, what up?"... but actually they're either...

  • already doing the obvious things
  • ...or, the obvious things turned out to be wrong, or a lot harder than you think, or trade off against other things that are more important.

I have come to believe there should be multiple rationality research orgs running in parallel, trying things differently, but the things I'm most excited to see tried differently aren't the usual things people bring up (such as long boot camps, online materials, measurement and RCTs. I think CFAR mostly doesn't do those for good reasons).

Instead, I'm interested more in people pursuing subtly different agendas, following different intuitions and hard-to-convey-models, oriented around somewhat different goals.

My impression is that CFAR ended up orienting a lot around what I'd call "the internal alignment paradigm" (including but not limited to the stuff Kaj Sotala is writing up these days, although Kaj himself is not from CFAR). I think this is a good paradigm, but I'd like to see more diversity of paradigms trying to solve different problems.

I don't think this results in 10x CFAR, just different research paths.

As for CFAR itself and what would 10x it, the answers I think are mostly of the form "have clearer vision and better internal management structures" (which are pretty hard for outsiders to have useful opinions about)

I also think it'd be better if CFAR wrote up more of it's stuff, not from an "online training materials" standpoint, but from a "allowing others to build off their research" standpoint (which isn't precisely a way to 10x CFAR itself, but to maybe improve the overall output of people-who-are-developing-rationality-training)

Matt Goldenberg


1. More scalable testing. Use online courses not because you expect them to raise the sanity waterline, but because you can more rapidly and scalably test different courses, until your courses are so good that you can teach them adequately without being live.

2. More engagement with the existing organizations and processes that have already worked on the problem. Have two people who's primary goal is to go to other self-help workshops/therapy sessions, steal the best stuff, and bring it back.

3. More focus on group rationality and new forms of organizations. Experiment with new ways to work together and new ways to organize.

4. Experiment with longer form 2-3 month bootcamps which involve progress on real world projects. Take a percentage of the profits and therefore align yourself towards actual winning.

The idea of an online course sounds quite interesting. CFAR workshops only run for a certain amount of time, but I'm sure there are extra lessons that they'd love to be able to squeeze in. This could provide a solution.

Also, it's very hard to have a strong, lifelong influence over someone over the course of just a long weekend. I agree that there could be huge value with a longer format.

Some specific ideas, particularly "What is the best way to give at a conference?", need not be optimized so much, but the same idea applies to everything, or at least a clear goal set in the first place.