Author: Stan Arts
Date: 13:01:51 06/23/02
Go up one level in this thread
On June 23, 2002 at 14:32:35, Kurt Utzinger wrote: >It is known that good opening books [may] have a good influence on the strenght >of chessprograms. Based on running tests, I however dare say that the opening >books do influence the results of computer matches to a much greater extend I >have ever imagined. Results to follow in some weeks. > >Has somebody made similar tests to report? > >Kind regards >Kurt Yes, it seems that openingbooks take care of the fact that programs can´t calculate some very long term positional things, and so create a good position for a program to start doing it´s tactical calculating. I saw that some program have huge openingsbooks and sometimes know the first 10-15 moves "out of book", that way you are well into the game and the type of game its going to be is pretty much decided. You can that way "take away" the type of games your program seems to be bad at I think. Sometimes. I added a small openingsbook to my program, to stop making itself look silly at the beginning of a game, and it makes MUCH difference to not using it. The rest of the game it plays much better, because it had a good start, causing better development etc. I just put in about 50 moves in book.. but it makes a big difference. Plus you can more easelly add a random factor so it´ll play different openings like I did. PS.: I like the idea that programs have and use openingsbooks (and endgame bases) against human players, since strong human players have this type of knowledge aswell, but when one program plays another, I think it´s silly to do that, because it´s not the actuall "program" doing all the playing. My thought would be something like..: Max first 5 moves out of book or so, for computer-computer matches :) i think it would be more fun. Ok that was just another thought. Stan
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.