Author: Robert Hyatt
Date: 12:22:17 07/27/99
Go up one level in this thread
On July 27, 1999 at 07:05:19, Ray MacFadyen wrote: >On July 27, 1999 at 00:17:44, Robert Hyatt wrote: > >>On July 26, 1999 at 20:03:50, Mark Young wrote: >> >>>On July 26, 1999 at 19:23:01, John Wentworth wrote: >>> >>>>After Cray Blitz won the 1983 World Computer Chess championship it's rating was >>>>2153. The Cray supercomputer from that time is still much faster than today's >>>>PC's ever dreamed of being yet Cray Blitz's rating was much lower than todays PC >>>>programs. In another post here, Robert Hyatt mentioned that Cray Blitz would >>>>beat Crafty based on hardware. So I guess my conclusion here is that either the >>>>rating of 2153 back in 1983 for Cray Blitz was way off or the estimated rating >>>>of Crafty and other current PC's are way off. In my opinion, Cray Blitz running >>>>on a Cray Supercomputer would still blow away any of the current programs >>>>running on any PC today. I guess I have to think that the rating of 2153 was WAY >>>>off. Any other opinions? >>> >>>Bob can correct me if I am wrong, but was not this rating based on games played >>>against humans, or computers that had ratings playing humans. If I remember >>>right, I think there was a prize for the first computer to achive and hold a >>>master rating playing humans of 2200 which I think Belle won. So I think the >>>rating may be pretty close to accurate because of this fact. >> >>Actually it was based on neither. Ken had a rating program... He computed the >>rating of Belle at 2200 based on it's USCF rating at the time of 2208. Most >>of the rest were assumed to be 2000, for no particular reason. The numbers in >>the crosstable are just TPR numbers... >> >>another case of taking a "rating estimate" as "the absolute rating"... >If crafty had a FIDE rating of 2450,for comparison purposes what rating could >cray blitz achieve running on the latest cray avalable ? >Yours sincerely >Ray Macfadyen This is impossible to predict. If I were to play Crafty vs Cray Blitz, with CB on a T932, I would expect it to win 9 of every 10 games or some such... And if you extrapolate a rating from that, it would be in the 2700+ range. But I don't believe that this really works for human-computer chess... CB would present some difficult tactical challenges for humans, but then it would also exhibit the same positional shortcomings that all of the current programs have as well... So projecting a rating would be really difficult, without enough games. It did beat a couple of GM players in 40/2 games however, so it wasn't 'weak' as some might have you believe. :)
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.