Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Crafty bug?

Author: James Robertson

Date: 19:49:52 11/20/98

Go up one level in this thread


On November 20, 1998 at 20:15:08, James B. Shearer wrote:

>On November 20, 1998 at 00:47:46, James Robertson wrote:
>
>>On November 19, 1998 at 22:34:09, James B. Shearer wrote:
>>
>>>      I was playing a Crafty clone on ICC and the following position arose:
>>>
>>>setboard r5k1/ppp1Q2p/1q4p1/8/8/2P1B3/bP4P1/4R1K1 b - - 0 24
>>>
>>>The clone played qb6xb2 which loses quickly after be3-d4.  I checked the
>>>position after the game was over with crafty 15.15 on my k6-233.  Crafty seems
>>>to take an inordinate amount of time to reject qxb2 (over 100 seconds).  Every
>>>other program I have (cm5500, rebel8, rebel decade20, fritz4) either never
>>>considers qxb2 or rejects it quickly (under 1 second).  Is this a bug?
>>>                        James B. Shearer
>>
>>Perhaps a null move problem (unlikely)? The mate is at ply 12 although Black
>>should realise that it loses it's queen (even if it doesn't see the mate) by ply
>>8. Ply 8 is when Crafty finds 1. ... c5 (after 28 seconds on my P233). GnuChess
>>is another program that has problems with this position.
>
>          Actually black gets mated or loses his queen at ply 6.  This may be a
>null move effect since the bad stuff happens at the end so playing a null move
>will push it over the horizon (because of the reduced depth after a null move).
>So perhaps R=2 means this gets pushed to ply 8, which is bad news in a bullet
>game (which this was).  I have seen this sort of thing before with Crafty, it
>seems very dangerous to allow the null move to end the search entirely.
>           Also I suspect the other programs have some sort of extensions which
>allow them to pick this up before ply 6.
>                          James B. Shearer

Sorry, I miscounted the plies. To get a program to analyze this under Winboard,
you have to first make a move with White, and then select Computer Play Black. I
made the first move with white, and then white mates on move 6, which, if
counted incorrectly (doubled) makes me get 12 plies for mate, or 2 extra plies.
:)

James



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.