Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Repitition detection

Author: John Lowe

Date: 15:37:23 12/27/02

Go up one level in this thread


On December 27, 2002 at 18:24:36, Bruce Moreland wrote:

>On December 27, 2002 at 17:41:09, John Lowe wrote:
>
>>I'm going to have to face this soon - I know it's not trivial and my conception,
>>as usual, will be simplistic and naive.
>>
>>A repeat move becomes significant only when it is returned as the players move.
>>If it leads to a draw it is in the interest af a side that is losing otherwise
>>it is to be avoided. At some stage it must acquire a score of "0".
>>
>>Does the inclusion of the "repeat move" information in the search itself affect
>>the move that will be made in avoiding the repetition? Or is the second best
>>move found in the search (assuming not another repeat) either not adequate or
>>uncertain?
>
>If you don't handle repetitions, the humans won't figure this out for a while,
>but once they do, watch out.
>
>And before they figure it out and start exploiting it like crazy, you'll still
>draw games where you are a rook up and winning effortlessly.
>
>Having this stuff in the search allows for some pruning.  I don't know if the
>amount is significant, but as a side-effect, I'll take it.  Detecting reps in
>the search has to be significant tactically, since there are so many positions
>where perpetual check is an issue.
>
>I may not have answered your question.
>
>bruce

Hi Bruce,

You've certainly made a start - food for thought - I've already had computer
programs snatching draws from lost positions and it would be good to keep DIY
ahead of my four-year old grandson for a little while :)

John



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.