Computer Chess Club Archives


Search

Terms

Messages

Subject: A Better Use for Books

Author: Robert Henry Durrett

Date: 15:28:07 08/21/98


As discussed briefly here at CCC some time ago, a large database such as that
used by ChessBase can [at least in theory] be used to produce a list of chess
positions which occur often in the games of chessmasters and GMs.  These
positions can be listed in order with the position which has occurred most often
being listed first, the position which has occurred in the database next most
often be listed as second, the third most often occurring position being listed
third, etceteras, until all of the positions which have occurred in the database
in more than one game have been listed in order.

I do not know if anybody has ever actually done that, but it should be clear
that it is possible to make such a list.  If several different positions have
occurred the same number of times, some other criteria could be used to make the
list unique if there were any need to do that.

Each entry in the list would have to include essential information such as
whether it is White to move or Black, whether the kings have moved, and other
such information.  Positions appearing identical in the sense that the same
squares have the same pieces, but different in one of these other ways would
have to be regarded as different positions in the list.

Evidentally, the initial position of the game, where it is white to move, would
appear at the top of the list.  The position after 1.e4 where it is Black to
move would be listed second.  Etceteras.

All of the above was just to set the scene for the following observation:

If the programmer who creates a chess engine wishes to utilize "books" to make
his software stronger [essentially by incorporating the knowledge of known
"chess theory"], then it really seems advisable and quite logical to provide a
"book" for each and every position which occurs near the top of the list of
"most often recurring positions."  This would be especially true for
often-recurring positions which have received considerable attention and
analysis in the open chess literature.

One obvious question is: "Is this not being done, in effect, already?"  I
believe the answer is a resounding NO!

Now, there is another generalization of the above, as follows:  Suppose someone
had collected a fairly complete set of "test positions" which are known to
create problems for a particular chess engine.  Wouldn't is also be "advisable
and quite logical" to include in that particular chess engine's "book" a move or
line for each such position?

At what point do the moralists and purists start to holler "that's not fair"?



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.