Author: Bruce Moreland
Date: 09:49:26 08/26/98
Go up one level in this thread
On August 25, 1998 at 07:28:29, fca wrote: >But the question is: WHY does it reject Qxg7. > >i.e. given say 1 hr/move, playing for both sides, what line and evaluations (for >each move) does it produce after Qxg7 is played ? > >>>Is Gandalf the best PC Chess Program ? :) > >>please no smiley ! Don't underestimate Gandalf. > >To estimate Gandalf we need the answer to the question. > >If the answer suggests it is simply because open lines to the castled king are >given a big no-no in the static evaluation, then I may be able to produce >positions where Gandalf **should** grab a pawn (or similar) but does not do so. Mine rejects it at 45 seconds on a P2/300 (it prefers Qe2) for the simple reason that it doesn't want to get its head ripped off on the g-file. Assuming the move is forced, mine doesn't think that white is that much better, something like +0.6. It would probably play the white side of the attack much like whatever it was that played white in the real game, and would get killed, maybe. It might get killed without takng the g-pawn. Assuming that the pawn is taken, white is up two pawns and black has an attack. Every black piece looks good, but the question is whether the attack is worth two pawns. I don't think any program can "solve" this. They are all going to apply some heuristics to this, and come up with a score, and if the score is better for black, great, but there are probably plenty of similar positions where the attack isn't worth two pawns, and as a consequence a positive evaluation would be erroneous. One way to make a program reject Qxg7 is to apply a huge asymmetric penalty for an open g-file. I don't do this. Remember, that there are positions that have the best chance of being solved by a computer that picks a move at random. bruce
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.