Iteration (aka trial-and-error) occurs on three levels, with feedback from users; to pivoting; and he's elsewhere said YC was initially a way to learn about angel investing, by having many investments, turning over quickly. Like geneticists studying fruit-flies - or Christensen studying the disk-drive market.

One of the disappointing things he's elsewhere said they've learnt through this is that smartness is less important than they'd thought. I wonder why he's fallen back to "smart people" in this interview - maybe it's code for "good people". #5: http://www.paulgraham.com/really.html http://www.paulgraham.com/founders.html