Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Does Rebel 10.5 Represent the Current State of the Art?

Author: Paul Richards

Date: 16:52:28 05/25/99

Go up one level in this thread


On May 25, 1999 at 15:28:46, blass uri wrote:

>I believe that kasparov also could find the draw in tournament time control
>against a human but he believed in your thoery that  deeper blue cannot do
>tactical mistakes so he even did not check it.

That's not my theory, my theory is that tactics and strategic understanding
are aspects of the same thing. This is a grey area, but when I say that
machines are better at tactics I mean that they are faster and more accurate,
not that they will necessarily find the right move in a given position.  I.e.
I mean it in the same sense as saying that computers are better at sorting
huge databases than we are.  That's a hopeless contest. I am not claiming
that speed and accuracy of calculation means finding the best move in a
position, because even perfect vision doesn't help if you aren't looking
for the right things.  This may be an abstract point, so I understand that
it will be easy for someone to not "get" what I mean. In a practical sense
programs still have a hard time seeing past draws, and as Dr. Hyatt pointed
out they are not too good at following forcing lines (a phenomenon I have
observed many times).  Maybe I can put it this way: a silicon brain is
naturally superior at tactics, and it fails only because that ability has
not been harnessed correctly.  If it could be given an equivalent
understanding of the game, its perfect calculation would see it win
every time.  Imagine Kasparov with a computer chip implant to augment his
searching.  The human has the selectivity to know what to look at, but
not the raw calculation ability.  The selectivity and a decent calculation
ability is usually superior to weak selectivity and great calculation
ability, until you crank up the calculation ability so high that the
other term becomes insignificant. ;) The fact that Kasparov can beat
a machine now does not mean his brain is faster or more accurate in terms
of calculation.  I may be splitting hairs semantically, but that's how
I am defining tactical ability.



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.