Author: Mike S.
Date: 19:48:14 06/12/03
Go up one level in this thread
On June 12, 2003 at 17:45:04, eric guttenberg wrote: >Currently a match is ongoing at www.mpillen.de between Shredder 7.04 >and King 3.23 SS=12 at g/60. At the end of 20 games the score was 10-10. > >The next 24 games were 16-8 in favor of S7.04 and Shredder is in the process >of winning game 45. > >Co-incidence or learning? I suppose you would have to look at all the >games to find out but the results so far suggest learning. Theoretically, in this case position learning could come into effect during the match *only* for the 2nd game each, with reversed colours - which is somewhat unlikely - as this is played from special starting positions (from the WM-Test). Therefore, book learning couldn't have an effect at all in that match. - But the testing conditions for that experiment include to switch off all learning functions. The topic of the first 10 positions/20 games was "Attack and Defence", and for the next part it's "Midgame and Endgame". There have matches of that type been played already, between *Shredder 7.04* and other engines. See "Gesamttabelle", http://www.mpillen.de/Mapi/WM-Test-Praxis/Turnier_WM_Test_Praxis/Tabellen/body_tabellen.html It's interesting to see, that while Shredder 7.04 scored a total of 55,6% in the previous matches, the detail results range from 37,5% (positional maneuvres and pos. sacrifices against Junior 7) to 77,5% (open and closed positions against Hiarcs 8). (The total results against the engines were all within 52,5% to 60,5%.) Regards, M.Scheidl
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.