Author: KarinsDad
Date: 10:39:07 01/31/99
Go up one level in this thread
On January 30, 1999 at 21:03:38, Reynolds Takata wrote: [snip] >> >>Sokolov recently blew away Kasparov at Wijk aan Zee, and he is 150 points lower. >>Freak results happen all the time, like J. Polgar's recent loss to a 2350. >> According to the ELO formula, Sokolov should win 3 times in 10 against Kasparov. And Polgar's opponent should have won 3 times in 20 against her. If you believe in the ELO formula (which I do not, both at extremes and at the high end of GM playing level), then you must concede that these are not freak results at all. They were bound to happen sooner or later. As for computers not knowing who their opponents are, that is just a matter of practicality. I was considering adding a CB type database to my program in an attempt to give it more knowledge, especially against other computers. But it seemed too daunting a task with too many problems, so I canned the idea. It does not mean that someone could not do it, they just haven't. As it is, computers have an edge over humans anyway. It does not matter if computers consistently beat GMs today, tomorrow, or 5 years from now. Eventually they will and whether they have knowledge of their opponent at that point will be moot. For example, with the C.A.P. project, there is a good chance that computers will one day have a deep opening database which will put them at an advantage over any GM that does not play an opening he knows cold. It does not matter what openings a given GM plays (i.e. who the opponent is), the computer will play him move for move and not use up any time at all. KarinsDad
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.