Computer Chess Club Archives


Search

Terms

Messages

Subject: a standard for evaluation score.

Author: Masciulli Gianluigi

Date: 02:57:05 05/18/00


All the evaluation score, more or less, are based on the pawn value.
that's fine but suppose is the black to move:
some program (type A) say's +1.00 when  white is better a pawn
and some other program (type B) says the position is -1.00 seeing form
side to move perpective.
the type b approch is quite disturbing while you analize a game (exemple giving)
with a program like crafty that will do something like this




+1 34)Ne4
-1 34) ... Nf3
+1 35)Ne2
+4 35) ... Bh2
-1 36)Qf2
-1 36) .. Rh7
-2 37)Ka1

considering the side to move look a bit confusing for me
WHY not settle (TYPE A) as a standard so an analisys like

+1 34)Ne4      :white ok
+1 34) ... Nf3 :nothing new
+1 35)Ne2      :nothing new
-4 35) ... Bh2 :bad surprise for the white black: can kill him
-1 36)Qf2      :missed to do so but...
+1 36) .. Rh7  :white ok again
-2 37)Ka1      :black is really better now

..and so on


teh situation is horrible when i dont know the type of the program
sometime I find on CCC posts that say:


"my program say that position is a -2.45"

so i have to go to the post with the position anddiscover
 1) what the side to move
 2) what type is "his" program (some time is not so easy to discover this :(

Setting (TYPE A) as standard may be much better for lazy people like me.
In any case, even if most of the program like (TYPE B) more, A standard is
better than NO STANDARD!!


ciao

glm




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.