Author: Bruce Moreland
Date: 11:27:58 03/22/01
Go up one level in this thread
On March 21, 2001 at 21:56:13, John Merlino wrote: >Chessmaster 8000 plays Kxf3 immediately. What I can't understand is why ANY >program would play Kg3 (but, then again, I'm not a very good chess player).... Black is ahead materially if you just count those two pawns as being +2 and the rook as being +5. If you play 1. ... Kg3 you threaten mate. 2. Kf1 is forced, and black plays 2. ... Kxf3. The winning line for white is 3. Ke1 Ke3 4. Kd1, and now 4. ... Kd3 does not threaten mate, so black can play ... e7 soon after white moves. Black can check a bunch and the white king has to go around the black king and chase the black rook away. That takes a long time from a program's perspective. The while time the program thinks it is winning. So of course it doesn't think that white is going to play the "crazy" Kd1 move. It thinks that white is trying to force a draw, so it probably sees white staying in the e1..h1 box and probably drawing. Which is what really has to happen after 1. ... Kxf3. It would be interesting to see how long it takes a program to get a huge positive score in this position: 2r5/5P2/4P2p/6p1/8/5PkP/8/6K1 w - - 0 1 Some might have a positive score to start, since they might regard the e6/f7 pawns as worth a rook or better. I would also wonder if programs that would play 1. ... Kxf3 in the initial position are seeing a draw from that position, or whether they are simply trying to win a pawn. So there are two issues here, I think: 1) Heuristic knowledge about the pawns on e6/f7. This knowledge applies in this case, but this knowledge is double-edged. So if a program has it, maybe it's a good thing, and maybe not. I think it depends upon how the knowledge is implemented. I think it would be safe to statically evaluate this position as strong for black, but clearly if you put the black king on f8 you could no longer do this. The position of the king matters greatly. 2) Programs may select 1. ... Kxf3 for the wrong reason. If programs think that 1. ... Kf3 and 1. ... Kg3 are both drawn, it's hardly fair to criticize one of them for selecting one of them while another selects the other. The above was written without computer analysis, so please take that into account when firing up Fritz in order to rebut me. 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.