Author: Robert Hyatt
Date: 06:58:42 01/09/00
Go up one level in this thread
On January 09, 2000 at 07:23:46, Graham Laight wrote: >On January 08, 2000 at 22:37:02, Robert Hyatt wrote: > >>>May I suggest you setup a filter on Crafty as follows; >>> >>>5 Minute Time Control >>>--------------------- >>> Win Loss Draw >>>Opponent ELO >>> >>>Once you have this filter for each time control you should be able to come up >>>with a close estimate of Crafty at FIDE ELO's at various time controls. In >>>other words, you may be able to determine whether Crafty is IM at 2/40 or >>>whatever !? >> >> >>Excepting 40/2, I can give that kind of data, because I have games vs humans, >>while the SSDF doesn't. But a program doesn't have to be a lot stronger than >>another to 'pull away from it'. It has been noticed many times that small >>changes in a program produdce significant improvement over older versions, >>even though that difference isn't seen in real human vs computer events... > >From past postings by various people (including programmers), the impression I >get is that it's easy to make a program beat an older version of itself >consistently, but not so easy to make a program able to beat a wide pool of >other programs. > >To get a good SSDF rating, you have to beat a wide variety of programs, not just >one. > >-g But it is the same idea. One program, armed with some new (and maybe even simple) idea that the others don't currently have... Will not only beat older versions of itself, but also old versions of other programs. Look at the stratification of the SSDF list to see what I mean... you can see when new versions came out clearly...
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.