Author: Uri Blass
Date: 09:19:22 08/20/03
Go up one level in this thread
On August 19, 2003 at 20:05:54, Mike Byrne wrote: >On August 19, 2003 at 12:00:35, Kerry McDermott wrote: > >>Thanx Mike. That was very helpful information. I am going to post my question >>differently to see what others may think. How did you find out about USCF >>ratings being adjusted down by 150? >> >>Regards, >> >>Kerry > > >Until last October , I had not played USCF for about 5 years. Although not >playing USCF, still keeping my hand in chess with computer and the what not. >When I came back, I felt like I was better player than 5 years ago. I had 3 >draws against players that were juts baove my rating. So much for feeling like >I was better - I was wrong or so I thought. Fast forward to May of this year, I >played in my tournament since coming back. I noticed one of old friends rated >near 1600. Back in 1995, he was rated over 1800. Hmmm, the aging process is >not going for my friend or so I thought. > >http://www.64.com/uscf/ratings/10159130 > >During the tournment I kept an eye on his game. His game did not seem any >weaker. To make a long story short, one of the other locals started telling >me how the USCF systematically too everybody's rating down because of rating >inflation. Of course that only impacted those who played. They set floor that >were genrally 200+ points below their high. In Roy's case above his floor is >1600. One can see he touched the floor several times. The last year, the USCF >decided they over corrected and are now psuhing ratings back up again. You can >see that Roy's rating is going back up. > >Ken Sloan probably understands ratings and ratings systems as well as anyone - >this short piece is great: > >http://groups.google.com/groups?q=g:thl3870583014d&dq=&hl=en&lr=&ie=UTF-8&oe=UTF-8&selm=t766h9mrlv.fsf_-_%40uab.edu > >This has been discussed at length at r.g.c.p - this thread has of the >participants that were actually involved in the USCF rating decisions - there is >a lot of mud slinging. > >http://groups.google.com/groups?q=g:thl286037684d&dq=&hl=en&lr=&ie=UTF-8&oe=UTF-8&selm=8t99ed%24bom%241%40nnrp1.deja.com > > >Marty Glickman and Ken Sloan are, IMO, two people who really understand how the >rating system works. Generally speaking, USCF has never had totally neutral >rating system - there has always been inflation or deflation parameters at work. > Search usenet where Ken Sloan or Marty Glickman are the posters to gain the >most intelligent understanding of ratings. I think that it is possible to prevent inflation and deflation by using a weak non deterministic chess programs. These programs(let say with average rating of 1600) may play in many tournaments when every program get the same number of games and their performance should be 1600. If they perform like 1650 when their should perform as 1600 then it is going to prove that the rating of humans is too high and all humans are going to go down by 50 elo. If they perform like 1550 when they should perform as 1600 then the opposite should happen. In order to prevent big luck factor in the adjusment we can decide that at least 1000 comp-human games need to be played before the adjusment. Uri
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.