Author: Chessfun
Date: 00:02:19 05/02/00
Go up one level in this thread
On May 02, 2000 at 02:31:37, Mogens Larsen wrote: >On May 01, 2000 at 23:43:14, Chessfun wrote: > >>If Crafty had won the first set 11 - 9 I wouldn't have questioned >>it. But Fritz had won it 15-5 therefore it was natural to conclude >>there was a problem when you have another identical set with a totally >>different set of results. > >Now, it's no longer possible to confirm the hypothesis that you would have >accepted the result 11-9, but since you revealed the problem I think you would >have. Of course I would have, otherwise why bother playing and offering the games to anyone or posting them. >My objection is that you conclude there is a problem on the basis of two >testsets. Even 20-0 for Crafty is possible, even though less probable. The conclusion that a problem existed after 15-5 then 9-11 to me is a natural one and one I would always make. As both can then be checked to confirm validity I don't see the issue. >>I had not intended to run it twice, it had looped through the night. >>If there was only one set of results I would have looked at those games >>on alternate machines to confirm some of the moves. This I do regardless >>of winner. > >It is still unclear to me, how you checked the results. If it isn't to much a >bother please explain how you checked the results. It can't just be the >comparison of game score, or? After each set of twenty games. I loaded the other engine in the opposite computer. Looked at the times and evals and checked critical moves, with both computers now running the other engine. >>Did I expect Fritz to win. Yes sure, I have said that since my very >>first post after Jouni's. I expect it to win with ponder=off or ponder=on. > >You inadvertently proved that the probablity of a 12-8 win for Crafty isn't >zero. Really.....since Jouni's were on a single machine ponder=off and that 11-9 we are talking about was two machines ponder=on, how do you conclude that?. If it is on the basis of my statement above relevant to ponder=on "I expect it to win with ponder=off or ponder=on." then the 12-8 isn't valid as that was Jouni's score and in fact the games were bogus as playing them a third time proved. Thanks.
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.