Author: John Merlino
Date: 15:39:25 02/04/01
Go up one level in this thread
On February 04, 2001 at 18:18:25, James T. Walker wrote: >On February 04, 2001 at 00:27:29, kenton Payne wrote: > >>Has anyone any opinions about the personality ratings on CM8000? >>I am running cm8 on W2k 533MHZ 128MB RAM. >>Does anyone know if they (the personality ratings) seem to run true. >>I did one small test of chessgenius (PalmPilot), which was said to be at about >>2000 by the manufacturer, vs. Roxy (1940 rating) and they got a draw. >>Has anyone else any ideas or observations? >> >>Thanx >>kedpayne2@hotmail.com > >I believe those ratings were calculated by playing games at Fischer time control >of 10 minutes + 15 seconds / move vs USCF rated humans. So at that time control >those ratings should be close to USCF ratings. I'm not sure if USCF blitz >ratings or USCF regular ratings were used in the calculation. Maybe John >Merlino can answer that. In either case I also believe you need to play a range >of players around your rating and not concentrate on just one personality to get >a good feel for strength. >Jim You're close, but not quite right. Fischer 5/10 was used. But, yes, where the testers has "quick" ratings, we used those. If they only had standard ratings, we had to use that. However, only five personalities were actually played against humans, and the ratings for these personalities were used as "real-world" anchor points for calculating all of the other personalities' ratings. About 2700 human vs. comp games were played, but another 11,000 comp vs. comp games were played to finish the job. jm
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.