Author: Dann Corbit
Date: 15:31:03 03/08/02
Go up one level in this thread
On March 08, 2002 at 17:59:44, Mike S. wrote: >On March 08, 2002 at 15:46:50, Dann Corbit wrote: > >>>(...) >>>Are only games between grandmasters and top programs considered? Or also >>>verifiable tournament condition games between club players and computer >>>programs? > >>The games in the list are computer verses computer. No new human data is added. > >On Tony Hedlund's pages, there is a "Man versus Machine" list, or results >collection. Unfortunately these (few, compared to SSDF) games have been played >at various different CPUs and clock rates, so I think thay cannot be used to >calibrate (?). > >http://home.interact.se/~w100107/manmachine.htm I think we need to consider it as a separate effort. It is useful (certainly) as a ballpark estimator. And I think it is reasonable to expect that a highly ranked SSDF program will do well in a real-world event. In fact, that is what we tend to see (to nobody's surprise). A large number of games would be very nice, but isn't terribly practical. On the other hand, if we had enough low ELO players try, we could eventually get an accurate rating that way (as long as they have many FIDE games under their belt and their ELO isn't so low that we get a 100% failure rate).
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.