Author: KarinsDad
Date: 16:37:51 06/09/99
Go up one level in this thread
On June 09, 1999 at 19:17:42, José de Jesús García Ruvalcaba wrote: [snip] > > Could you please elaborate your definition of «perfect move» in a drawn >position? >José. How do we know that drawn positions are even possible with perfect play? If they are not, then there is no need for such a definition. If drawn positions are possible with perfect play (or if you just wanted to check a random drawn position), then I guess that the perfect draw move for any given drawn position is one which still leads to a draw (obviously, if a given move leads to a loss in a drawn position, it has to be less than perfect). So, there may be multiple perfect moves. Of course, it might be better to get to a draw with a minimal number of moves, so one draw move may be more "perfect" than another. For example, if the tablebase had 40^60 positions in it (obviously, this could never occur in the physical world), it could take a REAL long time to look up a given position on the hard disk. So, it would be beneficial to look up the "best" move that leads quickest to a win or a draw (since there would presumably still be a time control on the game). The rest of the draws for a given position would not be needed in the tablebase. Also, some people may argue that the perfect move in a drawn position is one that maintains the draw, but gives your opponent the greatest possibility of making a mistake and losing. I think this would be hard to evaluate accurately. Just counting up the nodes probably would not be best. Instead, it may be best to have a system of the "most complicated draw", but that is beyond my ability to rationally determine what that might be right now. KarinsDad :)
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.