Author: Joseph Ciarrochi
Date: 13:35:26 12/13/05
Go up one level in this thread
But this looks like it could fall into the margin of error catagory, e.g., rybka still won two out of three with the SEE openings. I think you'll need a couple hundered games per opening type (nunn versus SEE), and then a moderate "real" difference, to detect a significant observed difference. Is it possible that programers could actually tune to the nunn openings? how much is the cegt based on nunn versus other books? And finally, where can I get these SEE openings? On December 13, 2005 at 11:03:32, Jouni Uski wrote: >Nunn openings > >1 Rybka 1.0 Beta 32-bi 11001111111½1½10010½ 13.5/20 >2 Fritz 9 00110000000½0½01101½ 6.5/20 > >1 Rybka 1.0 Beta 32-bit ½1111½1½1½½½11½10½1½ 14.5/20 >2 Fruit 2.2.1 ½0000½0½0½½½00½01½0½ 5.5/20 > >1 Rybka 1.0 Beta 32-bit 110010½0111½½1111110 13.5/20 >2 Shredder 9.1 UCI 001101½1000½½0000001 6.5/20 > >SEE openings > >1 Fritz 9 ½1100½01011011½101½½ 11.5/20 (!) >2 Rybka 1.0 Beta 32-bit ½0011½10100100½010½½ 8.5/20 > >1 Rybka 1.0 Beta 32-bit 010½11½½1010½0111111 13.0/20 >2 Fruit 2.2.1 101½00½½0101½1000000 7.0/20 > >1 Rybka 1.0 Beta 32-bit 10011½1011½1001½11½½ 12.5/20 >2 Shredder 9.1 UCI 01100½0100½0110½00½½ 7.5/20 > >One quess: Franss has tested a lot with SEE positions (like CSS ratinglist). > >Jouni
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.