Author: Michael de la Maza
Date: 16:44:01 05/01/99
I was playing white against my portable Novag Opal and arrived at this position: White: Qa7, Re1, Kg1, and pawns on h2, g2, f2, and e3 Black: Qb5, Nc4, Rd3, Kg8, and pawns on e4, f5, g5, and h7 White to move This is clearly a won game for black, but to my amazement I was a able to draw by three-fold repetition (1. Qe7 h6 2. Qe6+ Kg7 3. Qe7+ Kg6 4. Qe6+ Kg7 etc.) I figured that Opal was just such a weak computer that it didn't see the many black moves that preserve the win and instead settled for a draw because it didn't want to lose a pawn or two. So I checked with ChessLab (www.chesslab.com) and, amazingly, the best variation that it gives is actually a win for WHITE! Here it is: 1. Qe7 h6 2. Qe6+ Kg7 3. Qe7+ Kg6 4. Qe6+ Kh5 5. Qf7+ Kh4 6. g3+ Kg4 7. h3+ Kf3 8. Qh5+ g4 9. hxg4 and mate in two for WHITE Two things stand out about this line: 1) There are several ways for black to win but neither program found any good line for black and 2) The variation that leads to mate for white seems so trivial that any program should be able to avoid it. I'm curious to know how easy other programs find this position. The settings for the two programs were as follows: Novag Opal: "Average of 120 seconds per move" Chesslab: "Long Analysis" Michael
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.