Author: Uri Blass
Date: 06:31:20 06/13/03
Go up one level in this thread
On June 13, 2003 at 07:01:08, Walter Faxon wrote: <snipped> >Gentlemen: > >I guess my point is being lost so I have to reiterate it. > >First of all, I have nothing against further research on refinements to the >brute-force approach to computer chess. Good methods combined with ever-faster >hardware are producing grandmaster-level results; this is chess-specific >"intelligence" by any reasonable definition. > >The problem is that as far as chess serving as a general testbed for AI >research, the fact that fast searchers do so well is actually a drawback. Any >experimental method that takes more time -- mimicking human thought processes, >or that of monkeys or martians or whatever -- will almost certainly lose vs. the >now well-tuned searchers. But that is not true of Go, where effective search >without a specific goal and likely path is simply impossible. Go requires other >methods, other sources of "intelligence". That's why Go threatens to usurp >chess as the "drosophila" of AI. > >Like most westerners, I know and like chess a lot more than Go. I would like to >be able to study AI and chess together, not Go. But I already understand the >basics of superfast search and huge endgame tables. I know there's more to >chess than that. I know there is more to learn, there is more to be done. > >My proposal for Limited Search Computer Chess (LSCC) is that we deliberately tie >one hand behind our backs, so to speak, by limiting the search aspect of the >computer's intelligence. This forces us to adopt other methods, perhaps those a >lot closer to that of the best human players, perhaps not. The point is that nodes are not the same. The point is that programs may do some kind of search in their evaluation function and it is hard to say what is a node. An example: A program may detect a trapped piece by evaluation and the way may be simply to see that it has no good square to escape. You can describe the actions also as search by making all the legal moves of the piece that is under threat to see if there is a safe square. 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.