Author: Albert Silver
Date: 07:56:06 01/15/06
Go up one level in this thread
On January 15, 2006 at 10:43:59, Andreas Guettinger wrote:
>On January 15, 2006 at 09:41:35, Torstein Hall wrote:
>
>>I whish I really new, but I have made one observation that I am quite sure is
>>correct.
>>Rybka evaluate potentional free pawns a long time before the other program see
>>them comming. Sometimes you can see Rybka for several moves evaluate the
>>position as +2 or more, while the oponent do not get it before the free pawn has
>>started the running.
>>
>>Torstein
>
>Corrrct. Rybka gives advanced free pawns a very high value. Other programmers
>are much more conservative about this. Allmost every program evaluates free
>pawns, the question is how much. That to high scores can easily backfire shows
>this game I posted some days ago:
>
>http://www.talkchess.com/forums/1/message.html?479282
>
>Rybka advanced the a-pawn with a score of +4.5 pawns to the second row, and
>only realized when the fifty move rule was approaching that the endgame is a
>dead draw. This was positionally dumb play. Not pushing the pawn and exchanging
>all the remaining pieces would probably have scored the full point.
I looked at the game, and all the remaining pieces, when the a-pawn is first
advanced, are a rook each. It isn't so simple to propose to just exchange the
rooks. As to misplaying this particualr endgame, I'd say one could attribute
this more to lack of endgame knowledge than a mistaken overvaluation of
advancing pawns.
Albert
>
>This is only an example, I bet that there are lots of other examples that work
>the other way round were Rybka wins the game because of it's free pawns. The
>quastion is whether this high evaluation of potential free pawns makes Rybka
>positionally superior to other engines.
>
>regards
>Andy
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.