This is my main question. I've never seen anything to imply that multi-day workshops are effective methods of learning. Going further, I'm not sure how Less Wrong supports Spaced Repetition and Distributed Practice on one hand, while also supporting an organization that's primary outreach seems to be crash courses. It's like Less Wrong is showing a forum wide cognitive dissonance that nobody notices.

That leaves a few options:

• I'm wrong (though I consider it highly unlikely)
• CFAR never bothered to look it up or uses self selection to convince themselves it's effective
• CFAR is trying to optimize for something aside from spreading rationality, but they aren't actually saying what.

Why not weekly classes of 2 to 3 hours?

Those would be very impractical for people not within a reasonable commuting distance of where the classes are held.

The short answer: because we're trying to teach a kind of thinking rather than a pile of information, and this kind of thinking seems to be vary more easily acquired in an immersive multi-day context -- especially a context in which participants have set aside their ordinary commitments, and are free to question their normal modes of working/socializing/etc. without needing to answer their emails meanwhile.

Why I think this: CFAR experimented quite a bit with short classes (1 hour, 3 hours, etc.), daylong commuter events, multi-day commuter events, and workshops of varying numbers of days. We ran our first immersive workshop 6 months into our existence, after much experimentation with short formats; and we continued to experiment extensively with varied formats thereafter.

We found that participants were far more likely to fill in high scores to "0 to 10, are you glad you came?" at multi-day residential events. We found also that they seemed to us to engage with the material more fully and acquire the "mindset" of applied rationality more easily and more deeply, and that conversations relaxed, opened up, and became more honest/engaged as each workshop progressed, with participants feeling free to e.g. question whether their apparently insoluble problems were in fact insoluble, whether they in fact wanted to stay in the careers they felt "already stuck" in, whether they could "become a math person after all" or "learn social skills after all" or come to care about the world even if they hadn't been born that way, etc.

We also find we learn more from participants with whom we have more extensive contact, and the residential setting provides that well per unit staff time -- we can really get in the mode of hanging out with a given set of participants, trying to understand where they're at, forming hypotheses that might help, trying those hypotheses real-time in a really data-rich setting, seeing why that didn't quite work, and trying again... And developing better curricula is perhaps CFAR's main focus.

That said, discussed in our year-end review & fundraiser post, we are planning to attempt more writing, both for the sake of scalable reading and for the sake of more explicitly formulating some of what we think we know. It'll be interesting to see how that goes.

(You might also check our Critch's recent post on why CFAR has focused so much on residential workshops.)