Author: Dann Corbit
Date: 14:57:59 11/29/99
Go up one level in this thread
On November 29, 1999 at 03:04:50, David Blackman wrote: >On November 28, 1999 at 15:31:15, Jonathan Lee wrote: > >> Here are 2 conversions between USCF and FIDE on pages 84 and 85 of the 4th >>edition of the United States Chess Federation rule book. >> USCF = FIDE + 50 >> USCF = (0.895)*FIDE + 367 >> Pages 84 and 85 give other conversions but they don't cover all the >>countries especially the Swedish. >> To make the whole thing more simple, what is the rating of Chess Tiger >>(SSDF=2690+) for USCF or FIDE? >> Jonathan > >SSDF are not the same as Swedish ratings. No one knows how to convert SSDF >ratings to anything else because the programs on the list have not played enough >games against strong human players under tournament conditions. Most people >suspect they are within about 300 points either way of USCF and FIDE ratings, >but it is impossible to say more accurately than that. > >So Chess Tiger should at least give an IM a tough game, and is perhaps a lot >better than that. The Rebel GM experiments may give a standard deviation of less than 300 by now. In any case, enough games and we will get a pretty good feeling. I think most computer programs have a tactical skill of perhaps 2800 and a strategic skill of perhaps 2000 or less. If programs have tablebase files they will usually play like a supergm in the endgame. Of course, they could still be fooled by zugzwang unless they turn off null move at some point. The remarkable results many programs have had at longer time controls recently leads me to believe that computers will give GM's a tough time and they have to be especially careful of tactical mistakes. But given enough data, we can answer the question. Thanks to Mr. Schroder, we are starting to collect some very good information in that department.
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.