Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Just learning capability?

Author: Hans Gerber

Date: 18:43:40 06/14/00

Go up one level in this thread


On June 14, 2000 at 20:17:36, Dann Corbit wrote:

>On June 14, 2000 at 18:51:10, Hans Gerber wrote:
>[snip]
>>Yes, in general. But in certain special positions programs make big mistakes.
>
>[snip again]
>>Now we have the problem of sense. In my opinion it does not make sense if we
>>compared human beginners with good computer programs. Let us compare masters and
>>machines since the claim is that machines could at least play like a master. In
>>my opinion this is not true. For the reason that machines still have too many
>>weaknesses in certain positions.
>
>For positional moves, gambit sort of moves, sacrifices, and things of that
>nature, computers often play like a 1400 player.  For tactical moves, computers
>play like a 2800 player.
>
>I have seen computers do horrible, ridiculous gaffes in the early endgame that
>anyone who has played one hundred games would not make.
>
>But in a 50 move game the odds that somewhere, sometime you won't see a deep
>snare are very high.  That's why the computers play so well.  They are the best
>"piece snackers" around.  If there is a way to spear something hidden deeply
>away somewhere, they will unerringly find it.
>
>I think if the GM's learned how to play anti-computer it would shave two or
>three hundred points off the computer ratings.


Thank you, that was a good clarification. And I take for sure that yourself are
a computerchess expert?!

Let me point out that the reason for the situation we have (that
anti-computerchess is not so much liked) is mainly a more or less aristocratic
self-image of chessplayers in general. Independent of the rank chessplayers have
 a very special understanding of pride. So a real chessplayer would not accept
any pawn or pieces advantages even if it goes against the World Champion. The
same with playing a computer. Pride tells them to play normal chess and not
trying second-best moves just to bust the machine. BTW that is the content of
the history of such computergames against prominent GM's. Humans tend to allow a
draw much more frequently than against GM opponents of the same category. The
other day Mr. Villegas reported what Karpov had told him after their game.
Karpov confessed that he even had started to play real chess. All what he had to
do was making his choice in the dictionary of his collected patterns. In other
words you can't expect the master to examin the concrete position. He will rely
on a more general view. Or differently put: he won't start to calculate deep
lines but will rely on his feeling based on his experience.

Computers do have so many weaknesses that it would be easy for such chessplayers
to exploit them. But you know chessplayers don't the the reason why they should
play such dull chess. Chessplayers always want to create some good games.

The situation is different if you are a weaker player and you might succeed
against very strong machines if you use certain "anti" strategies. For such
players this can be very satifying, also intellectually.

GMs however are used to play against creative opponents. That is fun for them.
Not proving that they are able to beat a machine that has no own imagination.


Hans Gerber



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.