Author: Poschmann
Date: 06:41:55 11/19/99
Go up one level in this thread
On November 19, 1999 at 06:45:57, blass uri wrote: >On November 19, 1999 at 01:56:01, Poschmann wrote: > >>On November 19, 1999 at 00:15:36, Baldomero Garcia, Jr. wrote: >> >>>I was wondering how strong computer chess programmers are. >>>On average, are they masters, experts, class A players or lower? >>>Baldo. >> >>One year ago I was very angry at my bad chess knowledge. At this time I got a >>nice book in german "Schach am PC" (Chess on PC) from Frederic Friedel and some >>other authors (for example Chrilly Donninger). My idea was to improve my own >>chess playing strength by writing a chess program. Very soon I found out, that >>this idea was completely false. Alpha beta and other well known algorithms have >>nothing to do with the way of thinking by good players. Most of their knowledge >>is "hard wired" in their head, created in their childhood. They cannot explain >>how they find out good moves. In the book mentioned above I found the nice >>sentence: "The only persons, who cannot write a chess program are grand >>masters." Currently I think that is true. If you write a chess program, you have >>not enough time to train yourself. > >I understand that you have less time to train yourself when you write a chess >program but I do not understand the reason grandmaster cannot write a chess >program. > A good chess player _see_ the best moves in a position mostly _without_ calculating all answers. His brain automatically excludes bad moves from further consideration. But very often he can not explain why. If you write a program you must formulate rules for good and bad moves. All players know some basic rules: centralize your figures, attack the opponent king and so on. But good players know the exceptions and useful violations of this rules too. The problem in programming is: How you can formulate the differences between situations fitting the rules and the exceptions. The probable result will be a poorer program if you consider more exceptions. In tournaments I have seen grandmasters, walking from board to board. They stay some seconds at each, grasp the situation and go to next. They can rebuild the figures, they can announce the best move but mostly they can not explain why. This kind of pattern recognition they use is unprogrammable. Ralf >I think that they can stop playing chess if they want and start writing a chess >program. > >Uri
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.