Author: Russell Reagan
Date: 20:50:31 11/26/03
Go up one level in this thread
On November 26, 2003 at 23:31:36, Chessfun wrote: >On November 26, 2003 at 23:20:45, Russell Reagan wrote: > >>It looks like the strongest engines at longer time controls are the strongest >>engines at fast time controls, more or less. > >Haven't checked how it compares to my blitz list but I did compare a while back. >http://www.chess-archive.com/ccc.php?art_id=189592 > >One of the purposes was to show that with enough games most if not all engines >will fit within a few points of their normal rating. That's good to hear. We discussed this not too long ago, about whether tournaments or matches needed to be at the super long time controls that some people use when trying to determine which engine is better. My thoughts were that there is a good medium somewhere in between really long and super fast. At really long time controls, it seems like a weaker engine will be able to cover up its weaknesses more, given the extra search depth. I think that means it takes more games to get a clear picture about which engine is stronger, plus the games are longer. I'd rather have 2x or 4x the quantity of games in that case. At faster time controls, the smallest weakness can be fatal (ex. sub-par time management). So I like medium time controls. G/30 with or without a small increment is probably the one I use the most. Of course, I'm looking at this through the eyes of a programmer, and I'm more concerned with things like determining if a change made to a program results in an improvement or not, so I care more that the tournament or match shows me that, rather than producing better quality games, or other important goals.
This page took 2.28 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.