Author: David Mitchell
Date: 22:13:43 04/28/05
Go up one level in this thread
On April 28, 2005 at 13:33:58, Kevin K wrote: >Is there any chess program which doesn't use minimax algorithm? >And is there any chess program which use minimax algorithm but not using >alpha beta cutoff? Very early in computer chess history, it was believed/hoped, that chess programs would be able to play strongly, using sophisticated AI with plans and goals/subgoals, for each stage of the game. That dream was never achieved, and with the advent of Northwestern's championship CHESS 4.0 program, it was discarded by most chess programmers. Clearly, what is a winning chess plan depends completely on the exact position, so using a more general plan, is risky. But can you make code that will develop a good plan for every position you find during your games?? That's exactly why I anxiously look forward to Steven's efforts with Symbolic. I'm sure it will take him much longer to complete Symbolic, than a traditional A/B searcher, but I believe he's using the right tools (the toolkit and LISP), and he is certainly a high-powered light bulb, and has prior A/B chess programming experience, as well. When I'm playing chess, I see positions that cry out for this or that plan, on my part. But I also see positions that cry out for me to go get another cup of coffee. :) If Symbolic is indeed successful, and using something other than a variation of A/B with a deep search, it will certainly be a milestone in comp. chess. I get the feeling that lots of AI programmers will also be looking at Symbolic's success or failure, for a long time to come. Dave P.S. A/B will ABSOLUTELY give you the same result as a full mini-max search, but considerably faster. I can't see any reason to use a full mini-max search in a chess program, can you?
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.