Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Why does Fritz (unlike Crafty) NOT understand the draw.

Author: Roy Brunjes

Date: 06:48:08 03/02/04

Go up one level in this thread


On March 02, 2004 at 09:06:01, Geert van der Wulp wrote:

>Hi to all,
>
>Yesterday eve I held a match between Fritz 7 and Crafty 19.10. It came to a
>(completely drawn endgame) with Bishops of opposite colors, and 4 pawns each.
>Both also had 1 doubled pawn.
>
>If any of the two Kings abandonned his pawns, then at least one of them would be
>grabbed. The pawns were all blocking each other. (I deleted the game
>unfortunately). None of the two programs had a real plan, but Fritz did not see
>the equality of the position.
>
>Crafty's evaluation changed from +0.15 (advantage to Fritz) to 0.00 in a few
>moves. Fritz however continued to believe that it was better with +0.50 (for no
>particular reason as far as I could see) and wanted to continue the play, even
>though he had no plan at all. He just wanted to walk around in circles with his
>King, and move his Bishop around the board aimlessly.
>
>Why was Crafty able to see that there was no way to win, whereas Fritz did not
>seem to have a clue?
>
>Regards,
>
>Geert

Without a specific position to see, of course it is difficult to say much
specifically about either Fritz 7 or Crafty 19.10.  I can say however that I
have seen similar situations in my analyses of various endgames.  I speculate
that Bob Hyatt has spent significant effort in improving Crafty's play in at
least some endgames.  It seems to have resulted in a program that is better than
some commercial engines at spotting drawn endings rather quickly.  For this
reason, Crafty 19.11 is a constant analysis partner of mine in endgame
situations.

Do I believe Crafty 19.xx is stronger than any of the commercial engines?  Not
overall, no.  Is it strong enough that it can show me things I have overlooked
in my analysis?  Absolutely, though to be fair, I am not a strong player.  Is it
strong enough to spot things that GMs have overlooked in their analysis?  Many
times, yes.  Not bad at all for free!

Roy




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.