Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Kasparov v. Deep Blue revisited -- Byte's old article on the match

Author: Howard Exner

Date: 13:19:05 05/13/98

Go up one level in this thread


On May 13, 1998 at 14:06:46, Kai Middleton wrote:


>     The prevailing view is that Kasparov was beaten by a
>     sophisticated chess program running on a 1.4-ton IBM
>     supercomputer. Even Kasparov and his adviser apparently
>     think so. However, another view is that Kasparov was
>     beaten by a team of engineers, programmers, and grand
>     masters who used a supercomputer to dodge the game clock
>     in tournament chess.
>
>     Playing alone, one on one, it's highly unlikely that any of the
>     human members of IBM's Deep Blue team could defeat
>     Kasparov. But playing together, pooling their talent, IBM's
>     players probably could defeat Kasparov -- if they had
>     almost unlimited time to ponder their moves, while still
>     holding Kasparov to the game clock.
>
>     It's possible to calculate how much time IBM's team needed
>     to win. A tournament chess player has an average of 3
>     minutes to make each move. IBM estimates that a player of
>     Kasparov's skill can evaluate about three moves per second,
>     or roughly 540 moves in 3 minutes. Based on past
>     experience -- Kasparov's victory over Deep Blue in 1996 --
>     IBM's team was fairly certain it needed to consider 36 billion
>     moves in 3 minutes. Expressed another way, they needed the
>     equivalent of about 380 years to agree on each move.
>     Anything less wasn't enough. Everybody knows how tedious
>     committees are, but this is ridiculous. It's doubtful the World
>     Chess Federation would sanction such a protracted
>     tournament, especially since it would have to be completed
>     by Kasparov's descendants. So IBM found a work-around:
>     It built a specialized supercomputer that could compress
>     those 380 years into 3 minutes.

I've always subscribed to this definition of computer chess. In previous
posts here and on rgcc I've used the comparison that computer chess
is essentially correspondence chess in hyper-drive. This interpretation
I believe also enhances the appreciation of the programmers. It is their
endless tweaking and hours of labor that really are the substance of
the chess programs we use. The computer is just a fast, nifty tool that
the programmers are using to make chess moves.

Thanks for posting this article from BYTE.



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.