Author: Dann Corbit
Date: 16:44:25 07/25/00
Go up one level in this thread
On July 25, 2000 at 19:42:25, Tom Kerrigan wrote: >On July 25, 2000 at 19:34:44, Dann Corbit wrote: > >>On July 25, 2000 at 19:30:28, Tom Kerrigan wrote: >>[snip] >>>If my evaluation function gave positive scores to winning positions, it would >>>win all of its games, period. If the positive scores were random, it would >>>probably win in stupid ways, but it would still win. >> >>I think this is back to square one. >>This assumes that we have perfect knowlege so we *really* know that a winning >>move is a winning move. We don't really know that. Also, from any given > >Right. I'm just talking about making an evaluation function that correlates well >with GM analysis/play. Surely there's no harm in that? I really doubt that you >would end up with a function that says +5 when Rebel says -300 or something >drastic like that. And if you do, who knows, maybe Rebel is way wrong. It seems like a good idea, but I don't know how you can model it. I think there is likely to be ebb and flow between any two excellent players that are near equals. You will probably have to use some sort of aggregation.
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.