Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: A theory of ratings drift for the SSDF

Author: Russell Reagan

Date: 19:10:39 04/09/02

Go up one level in this thread


I tend to agree with the theory that better programs have been added to the
talent pool. For example, if you look at the different versions of Fritz, there
were several versions out at the same time that were tested.

Fritz 1.0  486/33 MHz                  (1996) 2041   48   -47   215   55%  2008
Fritz 2.0  486/33 MHz                  (1996) 2133   30   -31   527   45%  2167
Fritz 3.0 486/50-66 MHz                (1996) 2258   22   -22   975   52%  2246
Fritz 4.0 Pentium 90 MHz               (1996) 2316   46   -44   243   60%  2243

The rating of Fritz got better with each new version. Granted the hardware was a
litte better with each new version, but all of Fritz's opponents had the same
hardware increase that Fritz had with each new version. So when other programs
were able to take advantage of better hardware, Fritz still continued to rise
with each new version.

As Newton said, there must be an equal and opposite reaction (although he wasn't
talking about chess, or was he?). So since some programs rose to the top, others
had to fall. That's what I noticed from looking at the data.

Russell



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.