Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Fruit 2.2.1 plays differently than the others and that makes it worth it

Author: Dagh Nielsen

Date: 01:07:38 11/15/05

Go up one level in this thread


On November 15, 2005 at 03:27:41, Uri Blass wrote:

>On November 15, 2005 at 02:56:54, Timothy J. Frohlick wrote:
>
>>Jorge,
>>
>>You are correct about the rating differences. I find that Fruit plays stronger
>>at short blitz games because it searches faster.
>
>or maybe because it evaluates better
>How do you know if it is because of search or evaluation?
>
>Note that it is possible that better evaluation may be effective in blitz when
>at longer time control it may be less effective because search may find the
>right moves.
>
>Uri

Also, good search remains important at long time controls for another reason:
the branching factor. If a program has a low branching factor, it may
"outcalculate" an opponent by going deeper as time goes on. I remember the Zappa
author write something like this, that he had spent much effort on getting the
branching factor low, and so he guessed that Zappa advantage would be higher at
longer time controls. Anthony can of course clarify if he likes, I'm just
writing from memory.

In my own experience, relative program strengths simply do not depend much on
time control. I would be surpriced whenever I saw a relative difference higher
than 20-30 ELO points. The CEGT rating lists for blitz and 40/40 seem to confirm
this.



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.