2011/7/3 Andrés Domínguez <andres...@gmail.com> > 2011/7/3 Jean-loup Gailly <jl...@gailly.net>: > > Leon, > >> One of problems (which I tested with gogui, thankyou very much) > >> was losing points in endgame when program is winning. > > This is by design. Pachi maximises the chance of winning, not the number > > of points. But if you want Pachi to win by more points while increasing > > the risk of losing, > > I totally disagree. Playing stupid moves in the endgame doesn't > increase the probability of winning, probably the other way around. >
The problem is that your definition of "stupid moves" is out of context from the point of view of a strong go program. To a go program a move is stupid only if it decreases it's chance of winning the game. That seems like a good practical definition to me. It's not natural to us, but it's more logical. If the program could speak for itself, I think it might argue that what WE do is stupid, Maybe they would laugh at us for aimlessly trying to win every point we possibly can far beyond when the real game is over. In chess that is how two beginners play. You will see them getting several queens as if that is the point of the game. > The MCTS probability is not real, but estimanted. I think the problem > here is that the noise is bigger than the useful data. Maybe adding or > extracting komi when the score is nearly 0 or 1 can help. > You are covering very old ground here. I suggest that before you try to contribute code, you should read the archives to this list going back several years - because you are only now just discovering what has been known and discussed for several years now. It seems to you like you just discovered something new here but that's not the case. > > 0.001 vs 0.000 or 0.999 vs 0.998 probability is not statistically > significant when you only play a few thousand playouts. > Read the archives. This is not a new intuition but has been discussed at length and there has been a lot of research on this. > > Andrés > _______________________________________________ > Computer-go mailing list > Computer-go@dvandva.org > http://dvandva.org/cgi-bin/mailman/listinfo/computer-go >
_______________________________________________ Computer-go mailing list Computer-go@dvandva.org http://dvandva.org/cgi-bin/mailman/listinfo/computer-go