Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: To Bob, probably Crafty has an interesting problem

Author: Robert Hyatt

Date: 08:26:27 06/09/01

Go up one level in this thread


On June 09, 2001 at 05:09:57, Ferdinand S. Mosca wrote:

>On June 08, 2001 at 11:16:42, Robert Hyatt wrote:
>
>>On June 08, 2001 at 01:53:58, Ferdinand S. Mosca wrote:
>>
>>>
>>>Hello Bob,
>>>
>>>I have been examining games on my tournament among winboard engines
>>>and found this position which Crafty should have not played.
>>>
>>>Comet B.33 - Crafty 18.9
>>>[D]8/8/1n6/2b2B2/5B1k/3p4/5P2/5K2 b - - 0 62
>>>Here Crafty played 62...Bxf2?? with an evaluation of 0.0
>>>on P3 668mhz.
>>>
>>>My Nalimov EGTB which is from Hiarcs 7.32 CD
>>>does not include KBBKN, which I think not even related
>>>here because it is clear that the two bishops
>>>will win against a knight.
>>
>>Sorry, but 2B's don't always win against a knight.  If you take the
>>position after Bxf2 Kxf2, and remove the final pawn, this is a dead
>>drawn tablebase ending.  The KBBKN tablebase will show this pretty
>>easily.
>
>Hi Bob,
>
>Normally 2B's will often win against a knight and 2B's normally will not win
>against 1B. BTW according to Jeremiah, from other post, Bxf2 is mate in 44
>moves.
>
>I guess the current version of Crafty running without complete 5-man EGTB is
>weak at minor piece ending. Not enough, when both sides have rooks and Crafty is
>slightly inferior, it will exhange rooks and goes for an inferior minor piece
>ending which it thought a complete drawish. Interesting to look at the databases
>now. Testing with position set up is also very nice.
>


I did some random probes.  In reasonable positions, KBB vs KN is not won
very often.  And overall, K**K* where * is any minor piece is almost always
drawn when yuou consider all the possible pieces that * could be.  Crafty
uses this basic piece of "knowledge" to go for what should be the easy "draw"
here (it enters a K** vs K*P ending which is drawn by its simple rule.  I
might fix this one day, but 99.999% of the time it is correct.  IE I have
seen many positions where it was KRNP vs KRBP where the opponent's pawn was way
advanced.  It will take the pawn in an instant, leaving it in a totally drawn
KRN vs KRP (it has the R and P).






>Maybe it's time to increase the disk space too!!
>
>Regards,
>Dinan
>
>>
>>
>>
>>>
>>>I disabled the EGTB and still Crafty likes to play
>>>Bxf2. I have tried the Chessbase Crafty 18.08 with the same result.
>>>
>>>Maybe someone will verify this on their unit.
>>>
>>>
>>>Regards,
>>>Dinan



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.