Author: Robert Hyatt
Date: 11:24:21 11/14/00
Go up one level in this thread
On November 14, 2000 at 13:32:09, Jeremiah Penery wrote: >On November 14, 2000 at 13:20:42, Robert Hyatt wrote: > >>On November 14, 2000 at 02:46:23, Jeremiah Penery wrote: >> >>>Another position like this is LCTFIN08 (Louget II test, Endgame #8): >>> >>>[D]2k5/p7/Pp1p1b2/1P1P1p2/2P2P1p/3K3P/5B2/8 w >>> >>>If c5 is not played, a bunch of piece-shuffling will happen. At least in this >>>one it's possible to play c5 later (in at least one of the other positions, the >>>move had to be played immediately, or it could never be played) and so the >>>50-move rule would help. However, it'd be nice for several obvious reasons to >>>be able to play c5 more quickly. >> >>Remember, the first goal is to win the game. If a pawn push is required to >>win, does it really matter whether the pawn push is provoked by some obscure >>'weariness' evaluation term, or by the 50-move rule? Yes, moving the pawn >>sooner would make the game end sooner. But since the game result doesn't take >>into account how many moves the game lasted, this really doesn't matter when you >>look at the big picture. >> >>Cray Blitz won many games based on its specialized draw scoring algorithm. But >>it took a lot of time to win those games... > >In all of the other positions I posted, the win occurs not by a pawn push but by >a piece sacrifice. That is the point - a program should be able to recognize >when the material is getting you nowhere, but if you sacrifice a piece you can >still have a (more) winning position. We won an important game in the 1986 WCCC because of this. We were a pawn up, but saw that we could either repeat, or give up a pawn to continue. We had chosen to give up the pawn as we were still better without the pawn. It _could_ have done it immediately, like a GM would have. Delaying until after the second repetition didn't hurt a thing, other than to burn some time.
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.