Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: can GM's learn from programs

Author: Mogens Larsen

Date: 04:23:51 05/22/02

Go up one level in this thread


On May 22, 2002 at 06:53:36, Uri Blass wrote:

>I think that it is dependent on the position.

Could be.

>In some openings GM may let a chess program to analyze for hours when he sleeps
>and he may discover some novelties.
>
>I know that there were novelties that were discovered by a computer.
>
>Computers may find ideas that GM's did not think about and I see no reason for
>GM's not to use computers to find novelties espacially in positions that they
>believe that computers are stronger.

The question is if checking your own analysis is faster or slower than verifying
and expanding computer analysis. I tend to think that checking your own analysis
is faster. With the obvious exceptions, chances are that a GM is better at
"reducing the tree". Guiding the computer analysis from scratch is clearly less
efficient IMO. Thorough analysis is time consuming, even with a fast computer,
so letting it probe a line unfamiliar to your style (or human style of play if
there is such a thing) or against your percieved strengths would be a waste of
time.

>There is no reason not to let the program to analyze the position at the same
>time that you analyze the position independently.

I agree.

>The program may find moves that you did not think about and you can learn from
>them.
>The program may find a novelty at the same time that you analyze the position
>and do not find the novelty.

Both correct. That's more or less what I meant by "missing moves". However, both
results are supplementary.

In general I suspect that Grandmasters don't exploit computing power as much as
they could, except for the very best. But that is merely a hunch.

Regards,
Mogens



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.