Author: William H Rogers
Date: 16:11:32 01/28/02
Go up one level in this thread
This is a quote from Chess Skill in Man and Machine.
The authors of BELLE tried some experients with different versions of their own
program. Although this is directly related to the main topic, it does refine
some of the points pointed out by the above post.
They played many different games between the same versions of their program only
with different dept of searches and found the following:
Perf P4 P5 P6 P7 P8 P9
----------------------------------
P4 1235 x 5 .5 0 0 0
P5 1570 15 x 3.5 3 .5 0
P6 1826 19.5 16.5 x 4 1.5 1.5
P7 2031 20 17 16 x 5 4
P8 2208 20 19.5 18.5 15 x 5.5
P9 2328 20 20 18.5 16 4.5 x
Each side played 10 games with each other for over 300 games in all.
Most interesting is that the rating increases seemed to dimish with the upper
plys of the game going from 370 points from ply 4 to ply 5 down to 120 points
between ply 8 and ply 9. Maybe someone can expolate these numbers and give us
an idea of what the ply ratings will be at even greater depts.
Bill
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.