New people should apply regardless of knowledge level:

• I reliably benefit from doing epistemic rationality exercises of the sort we’ll be doing, despite having done many such exercises, and having spent three years working with highly skilled rationalists (by current human standards). Fundamentals stay interesting, and can be internalized on many levels.
• For the topics where advanced students are most likely to not learn from beginners' questions, such as Bayesian probability and also such as AI risk, we'll be breaking the group in two halves based on how much background folks have with that specific topic (folks can choose which half to join).
• Anyone who is trying to figure out how to reduce existential risk should especially consider coming, since discussing strategy with others who’ve thought carefully about it, and meeting people you can potentially work with, is typically valuable.

I’ll have to discuss the issue of past fellows with others. We’ll select mostly participants who’ve never stayed here before, in order to have made contact with more people; but that isn’t an absolute rule.