Author: James T. Walker
Date: 09:39:42 11/28/00
Go up one level in this thread
On November 28, 2000 at 01:33:05, John Merlino wrote: >On November 27, 2000 at 23:17:00, James T. Walker wrote: > >>On November 27, 2000 at 13:10:54, John Merlino wrote: >> >>>On November 27, 2000 at 11:32:30, James T. Walker wrote: >>> >>>>On November 26, 2000 at 18:19:42, John Merlino wrote: >>>> >>>>>On November 26, 2000 at 16:21:18, James T. Walker wrote: >>>>> >>>>>>At what time control were the games played? >>>>>>Jim >>>>> >>>>>Oops. I forgot that key point. Because of time constraints (from the time the >>>>>test was mailed out to the time when ALL of the testing needed to be finished >>>>>was less than 4 months) the time control for all games was Fischer 5/10. >>>>> >>>>>jm >>>> >>>>Thanks for the info. In my opinion this means that the ratings are only >>>>accurate at that time control. >>>>Jim >>> >>>You are, of course, correct. It is impossible to have a SINGLE rating that is >>>accurate at ALL time controls. >>> >>>jm >> >>The problem is, this is a very unpopular/uncommon time control. I don't know >>what would be most desirable but I'm sure 5/10 is not it. >>Jim > >The reason that a Fischer time control was chosen was so the games could >continue indefinitely, and all personalities (and humans) wouldn't have to worry >about losing on time as much as with a fixed time control (you always have at >least 10 seconds to make your next move). > >The reasons that the time control is relatively short are twofold: >1) The development team wanted to make the games short enough for the human >testers so they would be more willing to play all 30 games required for the >test. If the time control had required a longer time commitment from the >testers, they very likely would have received less data. >2) Over 13,000 further comp vs. comp games needed to be played to complete the >testing and get reasonably accurate ratings for all of the personalities. At >5/10, the testing was done literally a few days before the program was finished. >Obviously, then, a longer time control would have caused the ratings testing to >not be finished within the deadlines of the main program. > >jm Hello John, All good reasons and I agree with them. The problem is your team seems to have ignored the most important ingredient in the formula. A rating (Even if not perfectly accurate) would be more useful to people at a more popular time control. On ICC/FICS some of the popular time controls are 5/3 or 2/12. In any case I'm sure people will appreciate this feature of trying to find their own rating if they don't already belong to some organization that will give them a rating. As usual the new ChessMaster has some bugs and hopefully they will be fixed soon. I have two almost identical computers and CM8K failed to install on one of them. I will write an email to Tech support today. I get an error message: An error occurred during the move data process:-133 Component Data for classic chess games File Group:dll_classic File:F:\Classic8k.pgn I did manage to install CM8K on the second computer by doing a selective install of only the first 4 components on the list and it works (kindof). Jim
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.