Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: About diminishing returns (Uri)

Author: Robert Hyatt

Date: 06:46:11 11/22/01

Go up one level in this thread


On November 22, 2001 at 04:11:41, Jouni Uski wrote:

>Uri what's your goal in experiment? There has been extensive tests with a lot
>of games (many 1000s) in CSS magazine with clear result: in autoplay against
>itself both Fritz6 and Hiarcs7.32 showed same kind of diminishing returns. If
 ^^^^^^

Diminishing returns against "yourself" does not necessarily imply diminishing
returns against "others".




>I remember right test was from 8 to 12 ply (for Hiarcs 1-2 ply less of course).
>And Heintz's tests have given same result as well.
>
>Also common sense says, that benefit from 5->6 ply is more, than from 40->41
>ply, no doubt at all!



Correct.  But does common sense say that 15-16 plies gives _nothing_?  And
does common sense say that using the _same_ engine at two different depths
might not be the optimal way to play chess?  IE I have found that as I go
deeper, I need to do different things with respect to search extensions and
evaluation terms.

All these diminishing return tests are simply anecdotal in nature



>
>Jouni



This page took 0.02 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.