Author: Guido Schimmels
Date: 03:03:16 05/15/98
Go up one level in this thread
On May 14, 1998 at 13:53:56, Alvaro Polo wrote: >Here is an idea, I don't know if it has been proposed before (probably >yes) or if it can be useful. > >The idea is to construct a chess meta-program, using a program to >combine the output of the best chess programs. Lets suppose we let the >best programs think during 1.5 minutes at a given position, and we >collect the results. Perhaps the probability of having any of them >proposing the best move, at this shorter time control, is greater than >the probability of the best of them individually to output this same >move at 3 mins per move. This could be tested empirically. If this was >the case the remaining problem would be how to know which of those moves >is really the best one. We could use the 1.5 minute per move left to >decide this. > >This whole approach is not very different from Althoffer's three-hirn >play, it is just an authomated version of it. Perhaps a new program >could follow the individual PV to decide which one is better, or a >majority votation could be made among the results proposed, etc. Don't be surprised, Stefan Meyer-Kahlen (Shredder) is working on an automatic three-hirn as his doctoral thesis ! I don't know much about it, only he has two chess-programs (Clever & Smart) with different amount of selectivity and a decider-program. Guido
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.