Author: José de Jesús García Ruvalcaba
Date: 13:56:25 07/27/99
Go up one level in this thread
On July 27, 1999 at 15:06:41, Robert Hyatt wrote: >On July 27, 1999 at 14:07:22, Dave Gomboc wrote: > >>Last night I reached the following position as White in the blitz portion of a >>game/30: >> >>White: Kg2, Qd5, pawns on b3, f2, g3, h4. >>Black: Kf6, Qe5, pawns on b4, g7, h7. >>White to move. >> >>I played 43. Qc6+, and my opponent blundered with 43...Qe6?? I went on to win >>the pawn ending after 44. Qxe6+ Kxe6 45. Kf3. >> >>Flipping through the game with Crafty, I noticed that it also liked 43...Qe6. >>In fact, it holds this move until at least depth=16. (I didn't let it analyze >>further, but really, this should be seen at depth far less than 16.) It clearly >>doesn't realize just how winning White's position is after the queen trade. >> >>Another test position for Bob's collection? >> >>Dave > > >It will understand this before long. It already understands 'outside passed >pawns' quite well... but at present, it doesn't understand that pawn majority >turns into outside passer later on. Crafty is not the only program with this weakness. And it can be exploited by humans who are aware of it. I once swindled a draw against crafty from a very inferior ending, using this weakness. Crafty preferred to have a central passed pawn which was easy to block, instead of an outside pawn majority which eventually would lead to an outside passed pawn (but it would require a very deep search to see that). I have seen similar mistakes a lot of times in comp-comp games. José.
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.