Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Are over-optimistically evaluations stronger than realistic evaluati

Author: Robert Hyatt

Date: 08:55:23 04/28/03

Go up one level in this thread


On April 27, 2003 at 18:06:35, Tim Foden wrote:

>On April 27, 2003 at 10:56:45, Robert Hyatt wrote:
>
>>On April 26, 2003 at 19:15:04, Drexel,Michael wrote:
>>
>>>Today I started an interesting experiment.
>>>A match Chessmaster 9000 against Shredder 7.04 in Chessbase GUI with
>>>over-optimistically settings for The King 3.23.
>>>With this settings The King engine evaluates his positions almost always as
>>>better for himself, except it is completely lost.
>>
>>You are overlooking a key point.  The alpha/beta algorithm is relative to
>>the initial score.  If a program starts off at +3.0, and another program
>>starts off at 0.0 in the same position, the two programs can play
>>identically.  All alpha/beta tries to do is maximize the raw score at the
>>root, whether it is +3 or -3 is irrelevant...
>
>This is of course true.  And the effects on a particular engine would depend
>much on the way that the piece scores are used.
>
>However, he isn't just adding +3.0 to the score of one side... he is (in effect)
>adding <some-const>*pieces.  So every time otherwise equal material is swapped,
>the engine will believe it has lost relative to the opponent.  I thus this would
>make the engine less likely to exchange material, and this must modify the way
>it plays... it would try to make more positional compensation before swapping,
>for example.
>
>Also, if it used the piece-material values in some way for king safety, then it
>may make the king safety into a kind of a-symmetrical one.
>
>So, basically, I don't agree that it will have no effect at all.  :)


It depends on how it is done.  For example, in crafty I have the usual
"trade pieces if ahead in material, trade pawns if behind in material."

That inflates the score (or deflates the score) in endings because if it is a
piece ahead, it will have an even bigger + score as pieces are removed.  And
these scores are generally _correct_ in that a bigger score is better, but the
absolute value of the score is not particularly significant, since I have never
seen anyone say "if you are ahead a piece and you have traded all pieces away
except for your extra piece, give yourself another pawn advantage."  Yet all
programs do this.

So, again, ignore the _value_ and just take the best move as the "best move
according to the program."  If you believe the "white is +.5 pawns better" then
you are in a grey area, at best...

If you believe just "white is better" you might be safer, but even then it is
not so clear.




>
>Cheers, Tim.



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.