Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Strength of the engine in chess programs

Author: Robert Hyatt

Date: 08:42:10 05/26/02

Go up one level in this thread


On May 26, 2002 at 05:24:57, Uri Blass wrote:

>
>I think that players that lose exactly the same game again and again should be
>banned from ICC.
>
>Programmers of programs with not enough learning to avoid the problem should not
>let their program to play too much.


Even with position learning a program can play the same _mistake_ 100 times
before the learning backs up into the tree far enough to force the game to
follow a different path.

It is non-trivial.



>
>They can let their program play only until they lose and after the first loss
>that they cannot learn to avoid again the program should automatically refuse to
>play.
>
>I see no reason to ban the winner and not the loser in double games.
>It seems more logical for me to ban the loser.


What if the "winner" finds a way to win against program X, then it plays each
of the "clones" of program X and beats them in _exactly_ the same game?  How
do you solve that?  The fritz operators pounded on the tiger clones with an
oddball opening.  They beat _all_ the tiger clones in exactly the same way.
How to stop that?


>
>If sonmeone repeats draws to inflate his rating then the player with the better
>rating should be banned.
>

That has happened.  A higher-rated player _has_ been banned for intentionally
inflating the rating of a lower-rated player.  Computer operators have done
this to their programs on occasion...



>Uri



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.