Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: wrong question!

Author: martin fierz

Date: 08:53:03 03/23/04

Go up one level in this thread


On March 23, 2004 at 11:49:15, Richard Pijl wrote:

>On March 23, 2004 at 09:11:14, martin fierz wrote:
>
>>On March 23, 2004 at 04:35:49, Tord Romstad wrote:
>>
>>>[D]k1r5/p5n1/1prp3p/5p2/P1PPp1pP/2P1P1P1/3KBP2/1R4B1 w - -
>>>
>>>This position occured in a blitz game on the ICC with Gothmog (white) against
>>>Arasan.  Of course, as is immediately obvious to a human observer, white is
>>>dead lost.  It's impossible to activate the bishop on g1, and white is
>>>effectively a rook down.
>>>
>>>To my disgust, Gothmog was quite happy about its position, and showed a small
>>>plus score.  And because Arasan appeared to be equally clueless about the
>>>position, Gothmog even went on to win after a really ugly endgame.
>>>
>>>After the game, I decided to check Gothmog's static eval for the position.
>>>It thinks that white has an advantage(!) of 0.24 pawns.  Of course it
>>>notices the bad mobility for the bishop on g1, but it doesn't understand
>>>that it will never be possible to activate the bishop without loss of
>>>material.
>>>
>>>How do other engines evaluate this position?
>>
>>IMO this question is not the right question to ask. i think gothmog is rather
>>good at giving up the exchange compared to other programs. it's static eval for
>>this position would be quite ok if the white bishop was on c1 for example, where
>>it's mobility is apparantly only very little bigger (one more square to go to).
>>therefore you have to ask not only what the static eval for the position is that
>>you gave, but also for the one with the bishop on c1. many engines will give
>>black a clear edge here because they are (too) materialistic. they will do this
>>in both positions. the really interesting question is whether any engine can
>>detect the HUGE difference between having the bishop on c1 or g1...
>
>The Baron doesn't see a big difference in static eval, the value is even
>slightly worse for the bishop on c1. But after a short search it makes quite a
>difference.
>Current Development version
>
>Original position static: -0.77
>After short (10 ply search): -1.5
>
>Bishop on c1 static: -0.87
>After short (10 ply search): -0.9
>
>The difference in score is purely because of mobility of the bishop (and rook,
>when looking at the static eval).
>
>Richard.

as expected :-)
(my engine of course behaves exactly the same and even gives the Bc1 position a
lower score...).

tord, are you feeling better now?

cheers
  martin



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.