Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: When Too Much Strength Is A Handicap

Author: Gordon Rattray

Date: 06:21:26 01/16/02

Go up one level in this thread


On January 16, 2002 at 07:54:58, Graham Laight wrote:

>There is a special situation where, IMO, weaker chess computers play better than
>stronger ones.
>
>When the opponent is clearly winning, but not yet completely won, very strong
>programs - possibly because they see refutation against all possibilities within
>their horizon - tend to "give up" and play bland moves which offer no fight to
>the enemy.
>
>If I was using a chess computer against a human opponent, and I was behind on
>material, I'd want the feisty old fighters of yesteryear - machines like Chess
>Master 2100, or Kasparov Travel Champion. In those days, they knew that the best
>way to trip up a human opponent was to create tactical mayhem. Many a time I've
>seen these great old machines snatch victory from the jaws of defeat. By
>contrast, newer programs tend to allow you to win just by playing obvious moves.
>
>Under the circumstances, they're trying to be too clever. They have the
>attribute that they can beat me without me knowing what went wrong - which is a
>sign of much higher mastery of the game - which the old fighters can't. But
>neither can they make a fight out of a lost position IMO.
>
>Of course - in normal chess, the stronger program is probably less likely to get
>itself into a mess in the first place...
>
>-g


An interesting point.

I think it's a question of what is the best strategy.  For example, I'd guess
it's possible to change a top chess program to weaken itself if it is losing
badly, and hence it may start playing like some of the weaker programs you
mention.  But is this always best?  It depends on the opponent... weak human?
grandmaster level human? another computer?  Trying a subtle trap which risks a
mate to oneself may sometimes work; other times not.  It's very difficult to
judge what your opponent will find to be straightforward or not.  I think the
approach of playing the position, and not factoring in the opponent, is a fair
approach.  Top programs while sometimes making a win easier, can also produce
stubborn defences as a result of their ability/technique.

I find this situation similar to the case where a chess program tries to get
into a won endgame tablebase position by giving away material.  It can be
comical, and doesn't look like the best play by far.  But it is logical and
"best" from a certain perspective.

Gordon



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.