Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: how much does eval effect strength?

Author: Uri Blass

Date: 05:44:05 09/18/03

Go up one level in this thread


On September 18, 2003 at 08:37:13, Vincent Diepeveen wrote:

>On September 18, 2003 at 08:25:52, Uri Blass wrote:
>
>>On September 18, 2003 at 08:11:49, Vincent Diepeveen wrote:
>>
>>>On September 18, 2003 at 07:36:25, Tony Werten wrote:
>>>
>>>>On September 18, 2003 at 06:30:27, macaroni wrote:
>>>>
>>>>>thats an example position it got to, it gives about a -0.1 for that position,
>>>>>when clearly it is hopeless, the pawns broken up, and very bad pieces.
>>>>
>>>>The position doesn't seem to be lost ( for a computer that is )
>>>>
>>>>After 30 min XiniX still gives only -1 ( what it also gave after 1 sec.)
>>>>
>>>>Tony
>>>
>>>make that -10
>>
>>-10 pawns?
>>
>>I guess it is easy to change it from -1 pawn to -10 pawns by multiply all
>>evaluations by 10.
>
>Considering your evaluation must be fitted within 2000 lines of C code, that
>sounds an excellent solution to me for movei, assuming this 2000 lines is a hard
>constraint.

The discussion here is not about movei so I see no reason to discuss about it.
I only said that  value of -10 pawns means nothing because it is easy to get it
without big positional evaluation.

-1 of one program may be equivalent to -1.6 of another program in terms of
probability of winning so comparing static evaluation is nonsense.

Uri



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.