Author: Bruce Moreland
Date: 22:19:37 01/30/01
Go up one level in this thread
On January 30, 2001 at 23:24:08, Dann Corbit wrote: >On January 30, 2001 at 23:12:30, Robert Hyatt wrote: >[snip] >>I agree that if I play games against old GNU chess versions, I don't see that >>kind of nonsense very often, and that most of this is highlighted by the fact >>that the two versions I am testing are _very_ close to each other, making the >>test very difficult to interpret. > >Which brings up an interesting point -- with close engines we *do* have a sort >of unstable equilibrium. It could take ten thousand matches before we find out >if some small change produces an incremental value. > >So how *do* we check to see if an idea is a good one? You use statistics. You figure out what the odds are that your result is due to chance. If the odds are high enough that you are scared that you are making a mistake, you throw out the result or do more tests. bruce
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.