Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Computer invented openings?

Author: Jon Dart

Date: 09:10:45 07/13/01

Go up one level in this thread


On July 12, 2001 at 16:59:45, Dann Corbit wrote:

>Ha ha!
>Delete your opening book and then play against GM's on ICC (or other programs
>equipped with a good book).  Computers are weak at opening calculations [which
>is why they *need* an opening book], absurdly strong in the middle game, weak
>again in the early endgame, and suddenly become omniscient in the late endgame.
>There are some programs that can play reasonably well without an opening book.
>But they will all make big mistakes from time to time without one.

On the other hand .. books made from collections of GM games will contain
a lot of bad lines. I have lots of examples where the game result doesn't
indicate how good or bad the opening was. I have been experimenting lately
with a much smaller opening book, tuning it based on ICC and autoplay
results. I don't think it's any worse than the bigger (500k moves or so)
book I have been using and may be better. Playing fewer lines also gives
my program more chances to apply book learning.

Of course the commercials often have hand-tuned books, but this isn't so
common among the amateur programs.

--Jon



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.