Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Difficult endgame test position (but solvable for humans!)

Author: Robin Smith

Date: 17:28:31 06/11/00

Go up one level in this thread


On June 09, 2000 at 20:12:54, Bruce Moreland wrote:

>On June 09, 2000 at 19:33:08, Robin Smith wrote:
>
>>This game and position reminds me very much of one I just finished in the US
>>correspondence chess championship.  In the following position:
>>
>>[D]8/4kpp1/1pQ4p/p7/P2q4/3P2P1/2P2P1P/6K1 b
>>
>>black played 34. ... Qd6? and lost after 35.Qxd6+ Kxd6 36.Kf1 b5 37.Ke2 bxa4
>>38.Kd2 and white easily stops the a pawn and wins.  After the game I was
>>surprised to find that even after very long "thinks" programs seem to
>>want to avoid playing 35.Qxd6, even though it is not at all hard (for a human)to
>>see that white can stop the a-pawns, and after that white's connected c & d
>>pawns are unstopable.  The only exception I found was Fritz6a, which finds Qxd6
>>fairly quickly.
>>
>>Maybe this position would make a good test position?
>>
>>Robin Smith
>
>I think the position is too hard to really test anything.  I bet that programs
>want to play c4 right away as white in order to stop from "losing" the a-pawn.
>
>A good to test might be where white chooses to "lose" the pawn.
>
>bruce

For whatever reason, Fritz6a DOES want to play 35.Qxd6, along with the CORRECT
PV (as played in the game).  Fritz6a seems to understand that the connected c&d
pawns are stonger than blacks outside passed a pawn(s).  So this position is NOT
too hard for at least ONE program.

Robin Smith



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.