Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Hiarcs7.32 vrs Hiarcs6 at Game/ 5 min.> Score 29.5 to 20.5, Only 59% Win

Author: Timothy J. Frohlick

Date: 03:04:17 06/28/99

Go up one level in this thread


Terry,

Like Dr. Hyatt, I take the rating of chess programs with a grain of salt.  The
important thing for now is how well they do against one another.  The next
phase, which is rapidly approaching, will be to see how the monsters of 2000 and
beyond do against the grandmasters on a real tournament basis.

I would say that Hiarcs 7.32 is probably 70 to 90 ELO points stronger than
Hiarcs 6.  When you consider the massive improvements in the endgame databases
and the enormous opening book plus the improved learning algorithm and the speed
up with code optimized for 32bit systems then you can see that Hiarcs 7.32 is
a much more interesting program to play than Hiarcs 6.  H732 is 2,213KB and
H6 is only 268KB in length.  This increase in "knowledge" makes for a much more
human-like chess game. If H732 were doing the same type of search as H6 it
should be about 1/9th as fast.  It is obviously doing some pretty fancy $#%^.

As I have said before, these programs/machines will get so good that only
Kasparov and maybe ten other top players will beat them.  Playing against them
will be even more fun as we see the damn things start giving us strategic plans
on how they will bury us.  "Dear Terry we are playing a B b5 Moscow Sicilian
opening.  I will attempt to cremate you by the following plans: x,y,z .  If you
think that by me telling you this that you can counter my plans then I will
modify those plans at a juncture unknown to you and you will lose!"

You ain't seen nothin yet!!!!


Tim Frohlick



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.