| Ange1o DePa1ma 2006-11-19, 9:05 pm |
| I thought I had escaped this sort of nitpicking by deleting
rec.games.chess.politics from my newsgroup list.
I can't speak for the 0000 rating because I don't know how USCF treats a
0/18 result. I believe they give 400 points below the average opponents'
ratings but I'm not sure. Mathematically, someone who loses all their games
should have a rating of zero, but there may be sound reasons for awarding
some other rating.
How can you say that a player who knows only how the pieces moves is of 1000
strength? When I was rated 969 back in the late sixties I knew far more than
how to move the pieces. Over the years I've played dozens of opponents,
adults and children, in the 900-1300 range. I assure you they are much more
savvy than you say they are.
"Kenneth Sloan" <KennethRSloan@gmail.com> wrote in message
news:ehoeo6$3dc$1@SonOfMaze.dpo.uab.edu...
quote:
> Taylor Kingston wrote:
>
> Complete beginners do not have Elo ratings of "effectively zero".
> Different implementations of Elo-based rating systems make different
> assumptions - but I would say that, on balance, it is fair to say that
> an adult beginner (knows only how the pieces move) of average
> intelligence can be said to have and Elo rating of approximately 1000.
>
> This is very different from 0000.
>
> In fact, 0000 is not a "special" number on the Elo scale. The only
> number which is at all special is 2000 - the dividing line between
> club players and "national class" players. Elo-based systems which treat
> 0000 as "special" are arguably broken.
>
> The USCF system currently has no such rating as 0000, and there is
> no sense in which a beginner can be said to have a USCF rating of
> 0000.
>
> FIDE's lowest rating is currently a moving target - but it's not
> 0000 (to the best of my knowledge), and 0000 has no special meaning.
>
> You certainly can't say "this player is a beginner, so his rating is
> 'effectively zero', and therefore we can compute his expected
> score against a 1000 player by considering them to be 1000 points
> different". If you want to do such a calculation (which is the
> only truly appropriate computation to perform on Elo ratings, then
> 1000 is a far better estimate than 0000.
>
>
>
> --
> Kenneth Sloan KennethRSloan@gmail.com
> Computer and Information Sciences +1-205-932-2213
> university of Alabama at Birmingham FAX +1-205-934-5473
> Birmingham, AL 35294-1170 http://www.cis.uab.edu/sloan/
|