Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Shredder losing in spite of +2.98

Author: Robert Hyatt

Date: 12:47:57 01/18/04

Go up one level in this thread


On January 18, 2004 at 13:55:06, David H. McClain wrote:

>>
>>>Robert when a program has a three move repetition problem in the code, is this
>>>different than when a program that cannot see that +4 should be -2 because
>>>position is too deep to see true score?
>>>
>>>kburcham
>>>
>>
>>No, you are taking my comment too literally.  The repetition bug is one
>>issue.  Evaluation bugs also lose games.  Search bugs lose games.  Poor
>>evaluation loses games.  Poor extensions lose games.  There are thousands
>>of ways to turn +5 into -Mate...
>>
>
>Dr. Hyatt,
>
>some basic questions:
>
>How do you fix this or is it a case of "If I knew I'd have the best program in
>the world?"  Would any attempts at correcting deficiencies entail playing
>hundreds if not thousands of games?

I don't believe it can be fixed perfectly, until a chess program can play
perfect chess.  That isn't going to happen within my lifetime for sure, and
I'd be willing to bet on "never" but I won't be around long enough to collect
on that bet. :)


>
>And another:  Can the user of a prgram find alternative methods by his own input
>regarding better parameters and opening books to overcome program deficiencies
>in certain areas?

I don't know.  Most of the things I fix are things that I watched happen on
the chess servers.  I know my search is not going to see certain types of
very deep perpetual draws, and I just don't worry about them at the moment.





>
>What are "extensions"
>

Search extensions, that attempt to guide the search deeper along those
paths where tactics are important, such as the deep perpetual check I
was mentioning...




>DHM



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.