How rollouts work II

It will come as no surprise that a rollout with a limited number of trialsfollows exactly the same procedure. It's sufficient to say that theproportion of wins/gammons etc. that come up when Jellyfish plays againstitself (say) 1296 times, aren't likely to vary all that much from theproportion we would get if we measured the proportion of results in everygame we could possibly get of Jellyfish playing against itself. (Ofcourse, there may still be some doubt whether the results of JF vs. JF arerepresentative of the results of a perfect player vs. a perfect player, orof you vs. Joe Average, but that's another story.)