Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Proof for the CSS Claim that WM-Test Elo Numbers meant Strength!

Author: Steve Glanzfeld

Date: 09:11:46 06/15/04

Go up one level in this thread


On June 15, 2004 at 11:19:31, Bryan Hofmann wrote:

>>The real experts have already stated the same thing. Testing a an engine on a
>position shows how it will make a move in that position only.

So what?  Testing an engine by playing (a) game(s) shows how it will perform in
that game(s), only. When you doubt testsuites to work as good samples, I can
also doubt any set of test games being good samples (compared to other games,
under other conditions, against other opponents...).

There's no alternative to trying to use adequate samples when you want to test.
It makes no sense to say the practise itself is the only valid test. This would
put an end to all testing. You'd never have an (more or less early) estimation
for a chess program's strength...

Seems like some people here have interest in avoiding any testing - I can guess
the reason! :)) But please search for better arguments when you go for that.
What I've read so far is very poor.

Steve



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.