Author: Walter Faxon
Date: 16:57:33 10/17/03
Go up one level in this thread
On October 17, 2003 at 07:23:20, Uri Blass wrote: >On October 17, 2003 at 02:50:33, Walter Faxon wrote: [snip] >>The reason is simple: computer chess has long been dominated by >>ever-more-efficient search of huge game trees, a topic which is now understood >>pretty well. Fast searchers have no time for construction of dynamic patterns >>or learning or generalization, or any deeper understanding of causality than >>refutation by alpha-beta. > >Nonsense. > >The reason that people did not teach chess programs to do it is simply not >knowing or not finding effective ways to do it. > >> In practice the best programs on good hardware >>consistently defeat all but the very best humans, so why go off on a tack that >>might take decades to come to fruition? > >because it can help in every thinking game and not only in chess(for example >programs may beat the best go players by the same strategy). > That's the idea. >> Better to find a trick that gives you >>an extra ply in some critical positions -- do that and you'll have a >>world-beater. For a few weeks, anyway. >> >>I've posted on this subject here in CCC before. To bring computer chess as a >>topic back into the mainstream of AI will require more than a lone researcher >>writing a planning or pattern-matching chess program in order to support a >>master's thesis. No, not even a dozen of them. Rather, it will require an >>organized competition that will reward computer chess performance _without_ >>full-width search. I've called it "Limited-Search Computer Chess" (LSCC). >>Limit competitors to 100 nodes/sec or even 10 nodes/sec (nodes not knodes). > > >meaningless. >The meaning of nodes is not the same for different programs. > "Node" is meaningless? I agree that it can be a little fuzzy, e.g., with swap-off tables, etc. But try this: if you symbolically make a _legal_ chess move and do some sort of test on the result -- not necessarily a full evaluation, maybe just to see if the destination square is safe -- the resulting position counts as a "node". That's at least a little like humans do it. So every position that has any sort of evaluation done on it, and every position above it in the tree, is a "node". That ought to be simple enough. At any rate, once a competition is initiated the terms will be firmed up. For now, you are welcome to suggest alternative definitions. >You can only limit the speed of the computer. > >Uri You can limit all sorts of things, computer speed, memory, databases, etc. (We're getting a real test with the PDAs, even multifunction cell phones now.) But since compared to humans computers are still rather dumb, limiting computer power per se is pointless. My idea is simply a competition to encourage programmers to apply more varied, hopefully "intelligent", methods to chess, even if these methods can't yet beat Kasparov, or Uri Blass, or even patzers like me. I do expect that programs specifically built to search 100 nodes/sec will be smart enough to defeat current programs artificially limited to 100 nodes/sec (with 99.9% of their time spent in an idle loop). -- Walter
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.