Author: Uri Blass
Date: 04:30:50 03/08/02
Go up one level in this thread
On March 08, 2002 at 06:55:28, Andrew Williams wrote: >>After 12 games I decided to stop the match since Shredder 6 was winning >>by a big margin. >> >> > >Really, there is no reason to stop here. I have gone to bed with a new >version of PostModernist beating an old version 10-0 and woken up with them >approximately even. If I *really* want to know if a new version is better than >an old, I would normally prefer something like 200 games. I play (say) 100 >openings and make each version play black and white for each opening. Even then, >this isn't a tremendously good way of testing two similar versions. > >Andrew I agree that 6-2 and 4 draws is not a significant result but 10-0 may be a significant result unless you know that the change is really a very small change. If you expect a change of 100 elo from the new version than 10-0 for the new version is a significant result If you expect 1 elo change then even 20-0 for the new version is not a significant result if you still believe that you did not do a more signiicant change. Uri
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.