I'm at the part where he suggests that the addition of information from a good prediction market can actually change the probability of the thing being predicted. Realizing this is gonna be too good to listen to groggy, I shall quit and resume tomorrow.