Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: How do you effectively dumb down an engine? (some suggestions)

Author: Bob Durrett

Date: 20:23:10 04/19/04

Go up one level in this thread


On April 19, 2004 at 19:15:37, Mike Byrne wrote:

>On April 19, 2004 at 15:41:45, Robert Hyatt wrote:
>
>>On April 19, 2004 at 13:49:15, Gian-Carlo Pascutto wrote:
>>
>>>On April 19, 2004 at 13:33:01, Robert Hyatt wrote:
>>>
>>>>It's losses are tactical losses only, not
>>>>"positional losses" which is not how a 1300 player (or 1500 player) actually
>>>>loses games...
>>>
>>>I'd think any game between <1900 ELO players is decided almost exclusively
>>>on tactics.
>>>
>>>Of course positional playing strength does matter, but it's not what
>>>decides games, usually.
>>>
>>>--
>>>GCP
>>
>>That wasn't the point.  You don't have a 1300 player that understands everything
>>about pawn structure but is tactically horrible.  IE with a chess program, with
>>a reasonable evaluation, tactics have to be way beyond bad to offset the
>>positional evaluation.  Then it just doesn't feel like playing a weak human, it
>>feels like playing something else entirely.
>
>"Human like" is tough - although many players like to play the weakened Crafty -
>not one of then has ever said it was like playing a human.  Even weak humans may
>plan long term strategies and just come up short on tactics, weak computers are
>for the most part less tactictal wilth zilch long term planning - but it's hard
>to program to just the let queen (pick your piece) "hang".  Weak human will
>generally not play knowingly playthe QxR and lose the exchange, but they will
>overlook the ocassional piece  hanging.

One GM said, in one of his books, that he selects examples from the games of the
top GMs because the games of lesser [human] chess players are filled with
serious mistakes mandating many changes of plans for both sides. On the other
hand, the really great players display a consistency and clear implementation of
correct plans.  This gives some insight into the play of intermediate players,
say from ELO 2200 to ELO 2600.

Amateurs considerably weaker than ELO 2200 have MANY huge mistakes in all of
their games.  Each time a major mistake is made, a new plan likely will need to
be formulated because of the drastically changed nature of the position.
Unfortunately, the very weak chessplayer will be unable to correctly assess the
position to determine the available achievable worthwhile plans in the
positions.  All too often, very weak players play from move to move and are soon
lost in the tactical [and sometimes positional/strategical] complexities of a
position.

It is a mistake to equate an amateur to a beginner, however.  Beginners make
many beginner mistakes even in the simplest of positions.  Beginners have a lot
to learn about basics.  On the other hand, what intermediate amateur has not
heard of a minority attack, a pawn storm, and the like?

Before one can emulate the play of an amateur, one must first formulate a
reasonably accurate model of the play of the amateur.  [The play is being
emulated and not the individual.]

Bob D.



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.