Author: Robert Hyatt
Date: 07:04:50 11/27/00
Go up one level in this thread
On November 27, 2000 at 04:24:46, Uri Blass wrote:
>On November 27, 2000 at 02:01:37, Mike S. wrote:
>
>>I have found another example. I don't think that many programs will play Rxf5+
>>here, although they cannot be called really stupid in this case... but the
>>solution looks easy - if you know it, as usual:
>>
>>[D]8/2p5/p1P1p3/P3kb2/1P4pp/8/5R2/7K w - -
>>
>>[Event "CSS 4/99 (Bednorz/Thiele)"]
>>[Site "?"]
>>[Date "1879.01.01"]
>>[Round "?"]
>>[White "Solovtsov"]
>>[Black "(study)"]
>>[Result "1-0"]
>>[SetUp "1"]
>>[FEN "8/2p5/p1P1p3/P3kb2/1P4pp/8/5R2/7K w - -"]
>>
>>1. Rxf5+ exf5 2. b5 Kd6 3. bxa6 Kxc6 4. Kh2 f4 5. Kg1
>>{A human player may see in the diagram position already,
>>that the 3 black pawns can't harm white's king now, and are
>>going to be captured due to zugzwang. After that, black's
>>king has to move and the a-pawn can promote.} 1-0
>>
>>Regards,
>>M.Scheidl
>
>Most programs will not find Rxf5+ but Crafty17.13 plays Rxf5+ in 0 seconds with
>0.00 evaluation.
>
>It seems that crafty has a wrong evaluation and is simply luvky.
>
>[D]8/2p5/P1k5/P4p2/6pp/8/8/7K w - - 0 1
>
>Crafty evaluates this position as 0.00 at depth 1 and it does not see like other
>programs that black has more pawns.
>
>Testing Crafty shows that Crafty is simply lucky because it also evaluate the
>position without the c7 pawn at depth 1 as a draw
>
>[D]8/8/P1k5/P4p2/6pp/8/7K/8 b - - 0 1
>
>
>This position is clearly a win for black but Crafty says 0.00 at small depthes.
>
>Uri
This is the _same_ bug I mentioned a week or so ago. NOte that white has only
rook pawns in this position, after a couple of moves. The modification of the
"rook pawn" evaluation sent to me had a hole that overlooked the fact that
the _opponent" could still win...
The new version doesn't fall into this...
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.