Author: Mig Greengard
Date: 23:05:19 08/24/01
Go up one level in this thread
On August 25, 2001 at 00:55:56, Christophe Theron wrote: >But as you tell me there is something to learn, I believe you and I think this >game deserves a longer analysis. You can't learn anything from 100 games unless you study each one. That's Botvinnik, and it shows how hard it is for a human or program to improve. Many games might not mean anything, but you have to look at all of them to find the one that holds a golden nugget. DJ-Tiger was not interesting because of how easily Junior won, although that draws attention, but when such a strong program plays into such a bad position there must be something wrong somewhere. (Junior's b4 against Shredder was such a move and will be giving Amir and Shay headaches. 17...f5 is yours.) 17...f5?? (17...h5!) could reflect various misconceptions by Tiger. (below) As always in computer chess, anything can be blamed on horizon, but that is wrong, as you said. Obviously, if it had seen ahead another 10 moves it wouldn't have played it... 1) B versus N. Assuming white would play 18.fxe4??, "winning" bishop for knight. 2) King safety. Not valuing enough the fact that with the knight on h6 the black king would be doomed in the long run. A human Master plays 18.Nh6 instinctively. Just a glance shows that the black king is in serious trouble, sooner or later. 3) Piece mobility. Believing that the knight would go to e5 and not the edge of the board. 4) Various possibilities about pawn play, not pushing an h-pawn, or wanting to play in the center or block it. Most programs evaluate Ne5 and Nh6 as roughly equal, a few have fxe4 (terrible) as first or second option. Junior instantly (<5 sec.) values Nh6 as much better than any other move, showing it's not a horizon issue, as expected. Saludos, Mig http://www.kasparovchess.com
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.