Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: A Good Test Position

Author: Serge Desmarais

Date: 22:45:55 08/27/98

Go up one level in this thread


On August 22, 1998 at 22:02:57, Robert Henry Durrett wrote:

>On August 21, 1998 at 08:55:50, Dan Homan wrote:
>
><snip>
>>I think that playing a good opening is a very hard problem with a
>>very easy solution... just build a quality opening book.  I this is
>>why some programs haven't been "tweaked" to play good openings without
>>a book...  (I think that my program will default to something like
>>the '4-knights defense' without it's opening book - solid development,
>>but not exactly exciting chess :)
><snip>
>
>If there is one position which all chess engines do poorly at, does this not
>speak poorly for all of them?  After all, the "initial position" is just another
>chess position.  True, it occurs more often than any other, and is a good
>candidate for using a "book."  However, if all chess engines do poorly with that
>particular position, should there not be a lot of concern?  Suppose there was a
>position several moves into the game where all chess engines did poorly. Would
>the chess engine designers not be concerned about that and try to make their
>programs do a better job with that position?  True, one could have a "book" for
>every position which gave a particular chess engine a problem, but this seems to
>me to be just "running away from the problem"!
>
>Even if a "book" is to be used for each position that gives the chess engine a
>problem, it would seem to me that the engine designer would STILL wish to remove
>the need for a book [for that position].
>
>True, the "initial position" is the most-recurring position in chess, so it
>diserves more attention than the rest, in terms of making a "book" to MASK the
>inherent deficiencies of the chess engine.
>
>Am I missing something here?


   Well, without a book I saw Fritz 5 play (after 1.e4 e5 2.Nf3 Nc6) 3.Bb5 at
ply 11. The Ruy Lopez. If I remember, Genius 3.0 would prefer to start a game
with 1.Nf3, which is not a bad move. They, by themselves, create fianchetti. The
main problem, according to me, is the lack of variety : without book, they
always play the same moves in the opening (1 line) if the same program has the
same colour and that the opponent aswere the same way. So we lose all the other
lines and defenses. Of course, you see programs blocking their C-pawn with a
knight in Queen-pawn games, which is not quite in the "spirit" of the opening.
But at least it puts a piece in play and toward the center!

Serge Desmarais



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.