Author: Robert Hyatt
Date: 06:17:21 12/10/03
Go up one level in this thread
On December 10, 2003 at 03:50:05, martin fierz wrote: >On December 09, 2003 at 23:22:12, Robert Hyatt wrote: > >[snip] > >>This nonsense about playing a 3-repeat move and hoping the opponent won't >>see it is totally ridiculous in the context of alpha/beta searching that we >>are all using. > >you are the one talking nonsense here. if you don't have to claim a draw, why do >it? you have obviously never played real chess yourself, or you would understand >this very well. I have played real chess for 40+ years. But it has _nothing_ to do with my playing chess. It has _everything_ to do with how alpha/beta works. And alpha/beta simply does _not_ work in the way you are suggesting. If my program repeats for the 3rd time, it is saying "the best I can do is a draw". It won't play the move, avoid the claim and hope the opponent will screw up. After all, we've done this once previously and the opponent didn't screw up then. Ergo, this is all nonsense with respect to a program using alpha/beta. Which is every program in the event that I know about. >even in the context of alpha-beta search this *can* be useful. your opponent >might have a bug, or he might have less time than before and play a different >move. you can't know with 100% certainty that he will repeat, so why not make >him show? granted in 99 of 100 cases, you will gain nothing. but you will not >lose anything by trying... How do you know whether the position _after_ you repeat but avoid the claim is better for you or your opponent? You never see it with alpha/beta. You just get draw score. I'll take a solid .5 every time rather than flipping a coin and getting some 1's and 0's, without knowing which before I flip. > >cheers > martin
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.