Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Which program plays

Author: Serge Desmarais

Date: 11:55:28 09/12/98

Go up one level in this thread


On September 12, 1998 at 02:23:55, Wayne Mendryk wrote:

>I don't know if any of you have seen the special "Nova" program dealing with
>computer chess. If you haven't try to find someone who has a copy or contact the
>PBS station near you, as this program discusses the question posed very well.
>I remember that Kasparov said the following in that show, Computer programmers
>should not attempt to duplicate how a chessmaster thinks, but rather should make
>use of the characteristics of the computer (hardware) and the minds of the
>programmers who have knowledge of which tasks within the program are necessary
>to make it play chess well. (This is not the exact statement which Kasparov
>made, I can't remember it) The fact that chess playing programs can learn from
>their mistakes, is in my opinion a very significant step in the continued
>improvement of programs (a feature which is now integrated into chess softward
>which is affordable for most every chessplayer). It is interesting to remember
>what Kasparov said after losing his match to Deep Blue, implying that the
>programmers were cheating because "the program had played like a human in
>several games. However, in my opinion, Kasparov was trying to make an excuse or
>is not willing to admit that the programmers of Deep Blue did a masterful job,
>when Kasparov thought the programmers were using illegal help to win. This must
>have been the ultimate compliment to the Deep Blue Team....Dr. Murray Campbell
>a member of the DB team is from my home town, of Edmonton,Alberta,Canada, and he
>did most of the programming for the chess algorithm DB used. Well done Murray
>!


   Kasparov said it because, if you remember, in the first game DB was ready to
do about anything just to prevent its opponent to force it to trade a bishop for
a knight. Deep Blue was a kind of computerized Janowsky, then ; it was its
"personnality". Then, for the second game, that wasn't true anymore... DB had
suffered a brainwash and remembered nothing about the bishop being better than
the knight, previous state which was the result of a full year of testing and
fine tuning. Kasparov has a talent in identifying the strenghts and weaknesses
of his opponents. It was a mistake to allow the IBM's team to tweak the program
between the games (the match was already so short). It resulted in DB being a
different machine/program every game. If they wanted to test how good their full
year of improvement, it was not a good idea to do changes after every game, or
so (as for every change, you normally have to test it for hundreds of games to
be sure it was an improvement).


Serge Desmarais



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.