Author: blass uri
Date: 03:43:24 05/02/00
Go up one level in this thread
On May 02, 2000 at 05:19:04, Mogens Larsen wrote: >On May 02, 2000 at 04:30:05, blass uri wrote: > >>How do you know that the conclusion is wrong? > >You can't test the result 11-9 out statistically by only playing another 20 >games. If you play another 80 games besides the ones you've already got, then >the average might converge on 15-5. But since noone tried, we don't know. I guess that chessfun looked also in the games and not only in the result and found that there is something wrong in the games that crafty won 11-9. I found something wrong in the games that Fritz won 9-0 by discovering that my crafty17.10 need less time to get the same depthes with slower hardware. Chessfun also found that I am right. > >>Did you ask for the games to check them with crafty and Fritz? > >No. > >>If you prove that the moves in the 11-9 match can be reproduced with the same >>times and that the moves in the 15-5 matches cannot be reproduced then it is a >>proof that the problem was in the 15-5 matches but I understand that you did not >>ask for the games. > >If I can reproduce the results, it means nothing. If I cannot, it means nothing. >This is due to the fact that I cannot emulate the hardware/software >configuration of Chessfun. This should be obvious in my opinion. I guess that the software/hardware is similiar and it is probably possible to find evidence that something is wrong in the games in the same way that I found that something was wrong in the games that fritz was leading 9-0. I suspect that in the 11-9 result something is wrong with fritz6a's moves. I do not have fritz6a so I cannot check if something is wrong with the moves of fritz6a even if I get the games. 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.