But of course, it's not the size of the win that counts, it is rather the confidence that it really is a win. In random playouts that continue from a position from a close game, the ones that result in a large victory are generally only ones where the opponent made a severe blunder. (Put another way, the score of the game is affected more by how bad the bad moves are, rather than how good the good ones are, or even how good most of the moves are. Others have commented on this effect in this list, in other contexts.) Since you can't count on that happening in the real game, these simulations have a lower value in the context of ensuring a win.
Even when the program is losing (say by a little bit) it is more important to play moves that it thinks are the most likely to convert the game to a win by confusing the opponent, rather than to play moves that will make the losing outcome more apparent. These tend to be different moves, and Monte Carlo methods are good at uncovering these differences. I agree that this is a bit surprising, but I find it much less so when I think about it in these terms. Given that people have reported such a strong effect, I am actually wondering if these simulations (those that result in a large score difference) should be _penalized_, for not being properly representative of the likely outcome of the game. In other words: value = 1000 * win - score Weston On 2/7/07, Don Dailey <[EMAIL PROTECTED]> wrote:
My intuition is also the same as yours - one would think 100 point win is better. It's not that it's better or it's a better move objectively, but it should be better against a fallible opponent who could easily get distracted by the little battles and forget the real point of the game. At least it's a way to fight when you are losing.
_______________________________________________ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
