Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Which Program Plays Closest To a Human Grandmaster?

Author: K. Burcham

Date: 19:10:44 12/27/01

Go up one level in this thread





Does anyone have any opinions on Which programs play the Most like a Human
Grandmaster in terms of strategy and Tactics? I am aware that Computers do not
have the Understanding of Chess that Grandmasters have, but what programs are
able to simulate Grandmaster Play the Most?

Will Bundy

first i would like to quote Chris.
> "Software programs typically use a wide band width brute force search,
>combined with an in depth search for tactically active lines"
Christophe

i wont say that todays programs can win every game with a  GM.
because we know they cant. but i also know that a GM cannot win every game with
any of several of todays top programs.

to answer your question:  "I am aware that GM's do not have the
Understanding of Chess that todays top programs have.
which GM's are able to simulate Program play the Most?   well i do not
think at this point it matters. the leading programs are all running in
a tight pack. it seems this pack is more than capable of holding its own
against any GM.

it also seems that if you are building your first program, and you want it to
be the strongest program of all time, then you cannot model this program
after GM or super GM play. these are the best in our world, but the GM and super
GM blunder, play into positions they do not understand, constantly
lose positional advantage, lose material, miss mates, cannot play some common
endings, and more. so why are you worried about our programs modeling
GM play? i am glad they dont. i do not think a programmer should try to get his
program to play tactics or strategy similiar to a GM. people here are constantly
posting difficult GM positions that our programs sometimes cannot solve. but!!!
we can find thousands of games with thousands of positions that the GM played
the losing move. take the last big GM tournament. play through each game and see
why each game was lost by a GM or super GM. many many postions
are misunderstood by the GM. many positions a GM plays into and does not know
he is playing into a losing position until it is too late.

i do not want any of my programs playing like any human GM or human super GM.

we all know why kramnik wanted to study the program before his upcoming
comp vs super GM match. just this fact in itself tells us that the super GM
knows he can very easily lose to any of todays top programs. so kramnik
will study its play and look for positions the program will not understand. once
found he will use these to an advantage in the match.

there are several here that say todays programs cannot play GM level.
but i have a very strong opinion, that if there is a tournament consisting
of 12 of the top Grandmasters, and including in this tournament is 12 of todays
top programs, there would not be a program in last place. several of these
programs would finish toward the top. and i would not be surprised if it was won
by a program. and i would not be surprised at all if the last five places were
GM.

dont get me wrong, i would "mow the yard" for any GM or super GM.
i would carry his luggage, and polish his shoes. i would make an ice run for his
coke and bourbon.     no  problem.

 but i dont want my program to play like a GM  or super GM. i like my programs
very much. and i want my programs to improve. and my programs will improve.
they always do. but i want my programs to stay programs. i do not want my
programs to play like a GM or a super GM.

when programs start to play like humans, i will update my racing software,
update my chassis library, warm up on my round track chassis technology again,
get on someones payroll and go back to racing.

kburcham











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.