Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Noomen matches comparing Rybka Beta 9 Neutral and Optimistic

Author: Vasik Rajlich

Date: 02:44:17 01/14/06

Go up one level in this thread


On January 13, 2006 at 12:31:12, Fernando Villegas wrote:

>I have the idea that the core of this business is to see which sub-set of
>heuristic are relevant to such or such position. To know what knowledege is
>irrelevant is the key of real knowledge.
>It is clear that when, as a human player, you feel that attacking the king is
>the thing to do, due to the overall position, you make like Alexander Tolush did
>and any master does, that is to say, you look for tactical knock out. Tolush
>used a witty sentence: "I look for a way to sacrifice the queen; if I do not
>find one, then and only then I look for more quiescent moves...."
>If you smell a mate attack against black it would be silly to expend time
>considering the positional outcome of playing a3.
>But on the contrary, if you does not see nothing in the realm of king side
>attacks, the thing to do is becoming positional and not calculate sacrifices and
>wild attacks.
>Every seasoned chess player do more or less that.
>I suppose Rybka do that.
>I believe it has some kind of monitoring set of codes lines to decide if going
>for tactics of keeping positional.
>If not, perhaps it should has  :-)
>
>My best
>fernando

Yes, I believe in this, but you need a very good static evaluator if you want to
act with confidence. In the Beta 10 there is a parameter called "Search Focus"
which controls Rybka's behavior along these lines, but it is all extremely
conservative compared to how it should be.

Vas



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.