Author: Peter Berger
Date: 02:57:15 01/16/01
Go up one level in this thread
This test makes no sense . I do understand that SSDF has a scientific, statistical conception . I also understand that it is important to play new programs against established ones to get a good, reliable ranking . But the ELO system only works well within a certain range . When the difference in strength and hardware is too high the results get completely random IMHO . One match point more or less will result in _huge_ rating differences . I think the maximum rating difference for the opponents chosen for a match should be limitted . And for the time when the rating is still unknown you should make an educated guess ( similar to seeding ). This is not meant to critisize your list , only as a suggestion for improval. pete
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.