Author: Robert Hyatt
Date: 19:51:39 05/01/01
Go up one level in this thread
On May 01, 2001 at 14:12:47, Peter Berger wrote: >On May 01, 2001 at 13:52:16, Robert Hyatt wrote: > >>On May 01, 2001 at 11:51:06, Ulrich Tuerke wrote: >>I don't believe this test is worth a can of beans for predicting a program's >>rating. > >Me neither and I don't think this is the main purpose of a test like this one - >the first numbers look like they are in a range that makes _some_ sense , that's >all . But it _does_ say something about the understanding of opening play of the >program IMHO . A few of them are quite tricky ; look at 22 for example or the >one with the minority attack - a theme few programs understand very well . > >The rating part is only for fun - compairing the results of different programs >is interesting though IMHO - especially looking for those a program/all programs >can't understand at all . > >Larry ran this very test through Cray Blitz at one of the last ACM >>events. We solved almost every position in 0 seconds. That blew his formula >>so badly he went back and changed every 0 second time to .5 seconds, since >>CB didn't display fractions of a second, and he reasoned that everything up to >>but not including one second would show up as zero. > >Wow , congrats ; nearly all of them in 0 seconds , that's incredible - I'll >definitely try Crafty next. > >pete > Remember Cray Blitz is not the same as Crafty by any stretch. We were using a horse of a machine when Larry and Don sat down with me to run the test positions. I will have to try to find the old output. I don't remember it failing to find a single answer, although some of them are really questionable as to whether they are the best move or not. Just because Fischer played a move, for example, does _not_ make it the best one possible. >> >>Predicting ratings like this simply _can't_ work...
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.