Author: Mark Young
Date: 17:12:05 06/18/98
Go up one level in this thread
On June 18, 1998 at 16:45:22, Mark Loftus wrote: > >I was reading through some old Web CCR articles earlier today - there aren't too >many new ones - and I noticed one by Fernando Villegas dated 3/14/97. It is >entitled, "A Proposal To Create A Rating List on Games Against Human Beings". >This is a suggestion I reaaly think has some merit and is a more important >statistic than the SSDF or Hallsworth could provide. This would involve more >than just the top pc programs of the past year or so. Many of us in CCC have a >rating and many different chess computers and programs, all we would have to do >is have a series of ten or so "serious games" against these programs and turn >them in to someone to verify and determine the rating. Plus, any book cooking >or designing programs only to beat the existing programs would not guarantee the >programmer a top spot on the list. > >Mark Loftus I know from experience that computer programs rated against human player would be somewhat different, then computers rated on computer verses computer play. I have started to get farther away from computer verses computer testing. I instead find it better to play humans when rating computers. Or at least doing half human and half computer testing. I was shocked how some of the strongest programs can play like absolute idiots from time to time. When playing medium to strong human chess players at game 30 or game 60 time controls. But those same programs playing other computers play like GM’s. I for one would like to see more human testing, but I understand why this is very hard to do.
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.