Author: Uri Blass
Date: 05:11:22 03/05/01
Go up one level in this thread
On March 05, 2001 at 02:49:38, David Blackman wrote: >On March 05, 2001 at 00:05:28, Wayne Lowrance wrote: > >>I am curious as to the strongest program used in Correspondence chess, where >>speed is not an issue, but finding the best moves. > >There was something called the Korresponence Kup held by people here on CCC a >couple of years ago. Turns out that programs that are good at normal time >controls are also good at correspondence. > >Perhaps the biggest surprise was that Crafty did somewhat better against the top >commercial programs than it usually does. (In fact, i think Crafty won the >tournament outright.) I do not remember that Crafty won a tournament in correspondence games. The number of games in the correspondence games was also very small so we cannot get conclusions from it. I have no reason to believe that crafty is better than other programs and my guess is that the best programs at tournament time control are also best at correspondence games. I do not know about program that do special extensions for long time control and crafty is no exception. > >Of course, most of the programs in that event would have been modified quite a >bit since then, so the outright results might not be the same today. But i think >there are 4 main conclusions that are still probably true: > >1. Crafty is very strong at correspondence, probably a bit stronger than it's >normal time control strength would suggest. I do not jump to this conclusion. > >2. Good programs at faster time controls are also good at correspondence. I believe that it is usually the case. I believe that it is not the case for hiarcs that has an horrible branching factor of 6-7 at long time control. > >3. All the top programs are extremely strong at correspondence (ie. most human >players without computer assistence would lose most of the time. However human >correspondance master players are at least in with a chance.) I believe based on my experience in correspondencegames that human master are also going to lose most of the time and the level of top programs at coerrespondence games is strong IM level. > >4. There is not much difference between the strength of the top 10 or so >progams. People here on CCC like to try to argue which is strongest, but in fact >the difference is so small it doesn't really matter. I also guess that the difference is small. Uri
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.