Author: Dan Andersson
Date: 09:10:48 06/15/04
Go up one level in this thread
> >The idea is to test with an ADEQUATE SAMPLE. This is being done in many areas. >For example, cars are tested on some - preferably "typical" - roads. You can't > It is not only the size, but also the seeming impossibbilty of making a decent and small categorization of the set of legal chess positions than make the task of a functional test seem impossible. This makes the adequate sample a moot point. The only thing close to an adequate sample I can think of when facing such a conundrum is to play actual games. That will settle on a reasonable level of confidence in infinitesemal time to the test approach. > >I'm curious what comes up next :) probably things like "a chess program isn't >stronger just because it searches deeper" or more stupid stuff of that kind. > >Great fun! :) But where are the real experts please? > The second statement isn't implied in Bryans post. You can't make up things and attribute them and be taken seriously. But it actually points to one flaw with positional tests. Faster processors may actually result in a lower score. The third point is easily discarded. An argument stands by its own merit. Calling for an expert is only a kind of handwaving. MvH Dan Andersson
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.