Author: Bruce Moreland
Date: 13:59:13 01/24/99
Go up one level in this thread
On January 24, 1999 at 13:49:04, Don Dailey wrote: [snip] >>Vincent Diepeveen sent me the games from the first weekend of play. I picked >>3 as that was all the time I had. I had crafty search each (on my quad P6/200 >>which I figured was slower than the machine they used by a significant amount.) >>I had crafty search for 10 minutes per move, and if it chose the same move >>anywhere between 3 and 8 minutes as they did, I called this a match. I got >>all but one move in those three games. Not a 'perfect' matching scheme, but >>with the parallel search, it is non-trivial even on identical hardware. > >Ok, good this seems like a pretty reasonable way to test. There is >some issue about the book, but I'll do the same test with Cilkchess >just to convince myself one way or the other. I'll take a wild guess >and predict a cilkchess match about 70% using your matching rules. >I would point out that any matching rules by this are by neccessity >fairly liberal (no other reasonable way to do it though.) I think >this kind of test is better for proving innocence than it is proving >guilt because there could be many matches by accident. Yes, let's see the games, and I'll try it too, because I think this is a worthy experiment. If Crafty gets a much higher match rate than the rest of us, it doesn't prove that an unmodified Crafty was used, by the way, it may just indicate that it is harder to change a program's personality than we thought. I would be very interested in seeing the match rates for: 1) The version of Crafty that was originally used by this guy, which is described on his web site. 2) The version of Crafty that was upgraded to, to get the SMP search. 3) A version of Crafty from a year or two ago. 4) Any other programs. To be clear, my goal when I helped start this discussion wasn't to try to nail Bionic Impakt's author for something. I simply want to prevent modified Crafties from driving weaker amateur programs out of sanctioned events. People shouldn't be forced to choose between doing things Bob's way (with his program) and not competing. >An interesting experiment you could also do is try different >versions of Crafty (perhaps you have already tried this.) Does >the match rate change a lot with different versions (with different >evaluation functions?) If they do, this is an argument against >the possiblity that you match a lot because he started with your >search, your extensions and the very same evaluation terms. He >told me that he added a pre-processor (he didn't call it this) >and that this was a significant component. He also claimed >that this made a really large difference in playing strength. How did he determine this? It is very hard to tell, IMHO. bruce
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.