My favorite quote from all of the talks was something Reid Hoffman said. I don't remember the exact quote, but he said something like "A startup is a set of hypotheses that must be true in order for it to meet success."
I thought there was a lot of insight in that line. A big part of what we entrepreneurs do is make predictions about the market and, more broadly, about people in general. If our analysis is correct, and our prediction on the best way to exploit that analysis is correct, we have a very high probability of success. If we're living in a fantasy world (for various values of fantasy) then our analysis is flawed and it's much harder for our predictions to be correct. In order to be successful, an entrepreneur must see the world as it is. (Yet, as Brian Chesky illustrated, they must also possess tenacious endurance.)
Dalton Caldwell gave me a peek into a market that I've only heard vague stories about, which I thought was very interesting. As I've never sought funding, Paul Graham's talk did the same for a very different marketplace. I really liked those.
However, the most practical talk for me, personally, was probably Andrew Mason's talk or Adam D'Angelo's talk (despite its lackluster delivery). Those had the most takeaway for me.