Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: evaluation of rybka is not very close to fruit.

Author: John Sidles

Date: 06:25:36 02/19/06

Go up one level in this thread


On February 18, 2006 at 03:35:23, Uri Blass wrote:
>It is important to choose the better move when the evaluation of many moves is
>almost the same. Uri

Perhaps Uri's comment is the key to Rybka's performance?

Suppose with white to move, moves "A" and "B" have the same minimax score,
but the search tree of move "A" has many branches for white, while move "B"
has many branches for black.  Then move "A" is greatly to be preferred, because
it is more likely to create future strategic options for white.  Given that no
search
is perfect, in the long run white's extra strategic options are sure to pay off.

In other words, perhaps when Rybka picks its moves, it analyzes both the move
score and the search tree topology



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.