Author: Robert Hyatt
Date: 19:25:33 06/30/99
Go up one level in this thread
On June 30, 1999 at 15:19:54, Christopher R. Dorr wrote: >On June 30, 1999 at 12:16:44, KarinsDad wrote: > >>On June 30, 1999 at 11:20:54, Pete Galati wrote: >> >>[snip] >>> >>>Would somebody pleas explain to me why these ratings are needed? I'm quite sure >>>that I have a very narrow focus on the whole thing, but I really don't see what >>>they are being used for. If it's a marketing thing, then it does not seem to >>>apply well to most programs?-most of which are not comercial as I see it. >>> >>>Pete >> >>They are not needed per se. >> >>However, the idea was mentioned due to the normal rating problems that are >>exacerbated on ICC. So, if you had a more tournament-like environment and hence, >>tournament-only ratings, it would be more difficult to inflate your tournament >>rating, especially for a computer program. It would still be easy to deflate >>your tournament rating, just by cheating and playing poorly. And a person could >>inflate their tournament rating by cheating and using a computer. But, the >>mechanism of using a computer and only playing certain opponents could not be >>used to inflate a computer tournament rating. > >Interesting point, but I doubt that this would help much. The only people on ICC >capable of beating the better programs are the *strong* IM/GM types (ICC rated >over 2700). They play on ICC, in part, precisely *because* they get to choose >who they play. Why would a ICC 2850 want to play in some kind of random >tournament, playing 1282s (or even 2282s for that matter)? What would he/she >gain? Nothing...so they wouldn't play...so nobody who *would* play would have >any realistic chance of challenging a good computer. Hence, the 'tournament >blitz' rating of the computer would be meaningless. How much information do we >get from the fact that Crafty went 70-0-0 against players rated on average 1950? >We already know this. This system would give us no additional information, in my >opinion. The players who could give us more information wouldn't play, and those >who did play would not allow us to generate meaningful conclusions. > >I just feel that this kind of rating system is a bad idea from the word go. > >Chris > > > > Another important point is that GM players will _not_ play in a tournament on a server unless there are significant prizes to be won. And without the GM players, the rating pool gets distorted just as badly... >> >>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.