Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Piece Values in Chess Programs (Larry Kaufman)

Author: Vincent Diepeveen

Date: 10:10:19 05/07/01

Go up one level in this thread


On May 07, 2001 at 06:46:42, David Blackman wrote:

>On May 06, 2001 at 21:43:37, Robert Hyatt wrote:
>
>>On May 06, 2001 at 08:35:07, Vincent Diepeveen wrote:
>>
>>>On May 05, 2001 at 12:01:08, Dana Turnmire wrote:
>>>
>>>>  Here is an interesting article found in a 1989 CCR article.
>>>
>>>Exactly the biggest misunderstanding in chessbooks is that 2 rooks
>>>are stronger as a queen. Even in most endgames a queen wins easily
>>>against 2 rooks.
>>
>>I wouldn't say in "most".  I would say in "some".  Two rooks can box up
>>a queen and make it impossible to win since the queen can't overpower the
>>two rooks without the help of the king...
>
>If it's just 2 rooks and some pawns against 1 queen and some pawns, i'd back the
>rooks usually to win. As long as there are not too many loose pawns, and the
>king safety is reasonable, the rooks can often win even when one pawn down. I
>used to think that the queen would beat or draw with the 2 rooks in an endgame,
>but now i've played a few of those endgames and seen a small amount of GM
>analysis, i've changed my mind.
>
>If there are more pieces on the board as well, it really depends on exactly what
>pieces, and on the kind of position. Knights seem to work very well with queens,
>and sometimes very well against rooks. Queens have much better mobility than
>rooks in many common positions, and can use this if there are several targets to
>attack and defend. But if the whole position is about controlling an open or
>half-open file and then using it, the rooks should win.
>
>I quite like the 1,4,4,6,12 basic piece values and i find they work well. They
>seem to get the right answer slightly more often than 1,3,3,5,9. In the case of
>queen vs 2 rooks, i think 1,3,3,5,9 is closer to the truth, but in most other
>cases 1,4,4,6,12 gets the same answer or is better. But either way can be made
>to work. You just need to get the right positional factors in there, and make
>sure they can add up to more than a pawn when they have to.

And as you see the 1 4 4 6 12 is much closer to my piece values
as 1 3 3 5 9 is to my piece values...

In 99.9% of all positions the computer sees 2 rooks for a queen the
queen side wins.

Just verbose print the positions!

The GM positions are really exceptions!

Harder to evaluate it is when Rook + light piece beats a queen.

There is a position in bs2830 where this happens. The qxd3 position.
Some versions of DIEP find Qxd3 at 12 ply some at 13 ply. But it's
a very hard to find position!




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.