Author: Jorge Pichard
Date: 06:03:31 02/29/04
Programs are more dangerous at endgames than top GMs, they will never blunder a simply position like this :-) http://www.chessbase.com/newsdetail.asp?newsid=1504 [D]8/8/5k2/2K4p/R7/8/1r4p1/6N1 b - - 0 1 Would you find the clear-cut win here? Kasparov played 55...Rf2? and after 56.Ra6+ Kf5 57.Kd4 Rf1 58.Ke3 he had to agree to a draw. But Fritz tells us the 55...Rb1! is the move to play in the above diagram. It leaves Black, in computer-speak, the equivalent of ten pawns up. A possible continuation is 56.Rf4+ Ke5! 57.Ne2 g1Q+ 58.Nxg1 Kxf4 and White is completely lost. It is interesting to note (and was pointed out to us by FM Dr Eric Peterson) that after 55...Rf1 the defence 56.Nf3 is more difficult for human beings. Fritz discards this line because it leads to an ending which the computer can look up in its "tablebases" of every R+P vs R ending that is possible on the chessboard. The program discovers that after 56...g1Q+ 57.Nxg1 Rxg1 the position is mate in 37 moves, and so prefers the above line, where we end up with rook and pawn vs knight, for which most versions of Fritz do not have tablebases. That line only achieves a +18 pawn score
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.