Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: strange fritz evaluations

Author: Uri Blass

Date: 01:39:44 04/16/02

Go up one level in this thread


On April 16, 2002 at 03:58:08, Steve Maughan wrote:

>Martin,
>
>>aloha!
>
>The warmer climate is starting to show ;-)
>
>>i just received my fritz 7 copy, and played a couple of blitz (4'+2'') against
>>it. i tried playing anti-computer, in the sense that i just tried to hold a draw
>>as white. of course, i didn't succeed :-)
>>in two games, i took back a move, and managed to draw, but what disturbs me is
>>the evaluation of fritz in the 2 final positions (which are dead draws):
>>
>>[D] 8/4k3/8/p1p3p1/1pP1p1p1/1P2PpP1/P2K1P2/8 w
>
>Just tried the easier position with Shredder 6.02 which gives Black +2.2.  These
>type of positions, with a pawn chain, are notoriously tough for computers.  The
>computer is seeing that Black has an extra pawn and they are advanced - assuming
>that Black can break through the computer is correct.  In this situation the
>assumption is wrong!
>
>Regards,
>
>Steve

Yes

I know that a friend of me is disappointed from shredder and based on his
analysis of his correspondence games say that shredder is only another program
that knows nothing about endgames and the evaluation of Fritz or Junior are
closer to the truth.

I wonder what is the knowledge of shredder in the evaluation function that other
programs do not have.

Does shredder know to evaluate the last endgame hiarcs8-smirin correctly?
My impression based on analysis with shredder5.32 is that it only after
understanding that the pawn d6 is trapped by search it understands that black
has better chances but I cannot be sure that the latest shredder is not better.

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.