Tim Mann 	  Mar 22 1994, 2:37 pm     show options
Newsgroups: rec.games.chess
From: m...@src.dec.com (Tim Mann) - Find messages by this author
Date: 22 Mar 1994 22:37:28 GMT
Local: Tues, Mar 22 1994 2:37 pm
Subject: Re: How do ICS ratings relate to the world?
Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse

Herb Kanner writes:
> ICS (in Help Ratings) indicates that they are striving for an average
> rating of about 1700. Empirically, the average rating in the U.S.
> (USCF rating) is about 1500. ...
> I don't know why ICS chose to inflate their ratings.

ICS didn't choose to inflate its ratings.  The goal is for the ratings
to be similar to USCF ratings.

Because the set of ICS users is not the same as the set of USCF
members, there is no reason to expect that the average strength of an
ICS player should be the same as the average strength of a USCF
member.  And if the average strength of the two pools differs, the
average rating will differ even if every player who is in both pools
has exactly the same rating in both!

So in spite of the difference in pool-wide average ratings, it's still
quite possible that the ICS system gives people who are also USCF
members ratings that are approximately the same as their USCF ratings.
It's also quite possible that it doesn't---i.e., that 1720 is the
wrong value to keep the two systems in line.  If 1720 is the right
value, this would say that the average active ICS player is stronger
than the average USCF player.  I don't know if that's true or false,
but it seems possible, as ICS attracts strong players from all over
the world.

The best value to use may change with time---as ICS becomes better
known, the composition of the pool of players who use it may change.
So 1720 may have been right at one time, but too high or too low now.
(It was initially chosen after the ICS ratings had been adjusted to be
closer to USCF ratings by adding a constant to everyone's rating; the
value chosen was based on a survey asking ICS users for their USCF
ratings.)

If someone cares enough to spend the time on it, I suppose we could
have another ratings survey to find out how much people's USCF and ICS
ratings differ, and use the results to adjust the ICS ratings.  (I
personally don't have time for this.)

        --Tim (Milwaukee on ICS)

	
David Gomboc 	  Mar 22 1994, 11:29 pm     show options
Newsgroups: rec.games.chess
From: gom...@cpsc.ucalgary.ca (David Gomboc) - Find messages by this author
Date: Wed, 23 Mar 1994 07:29:39 GMT
Local: Tues, Mar 22 1994 11:29 pm
Subject: Re: How do ICS ratings relate to the world?
Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse

In article <2mngii$...@apple.com> kan...@apple.com (Herbert Kanner) writes:
>ICS (in Help Ratings) indicates that they are striving for an average
>rating of about 1700. Empirically, the average rating in the U.S.
>(USCF rating) is about 1500. At the higher levels (>2000), USCF
>ratings are about 50 higher than FIDE ratings. ICS uses the same
>formula for computing ratings as USCF, but the placement of the scale
>is arbitrary. Hence, with enough games to go by, the difference
>between the ratings of two players should be the same in ICS and USCF,
>but the absolute values will be different. I don't know why ICS chose
>to inflate their ratings.

What makes you think the ICS ratings have been inflated artificially?
If you compare the USCF ratings of ICS chess players who have them to the
distribution of USCF OTB chess players I think you'll notice that the ICS
mean is significantly higher.  Alternatively you could compare the ICS
rating and the USCF OTB rating for the group on ICS; I think the mean in
this case would not differ by much.  Of course, I have no collected
statistics to validate or invalidate my opinions.

-- 
Dave Gomboc
drgom...@acs.ucalgary.ca (is forwarded to..)
gom...@cpsc.ucalgary.ca

	
Steve Zimmerman 	  Mar 23 1994, 1:24 pm     show options
Newsgroups: rec.games.chess
From: ste...@code3.code3.com (Steve Zimmerman) - Find messages by this author
Date: Wed, 23 Mar 1994 21:24:21 GMT
Local: Wed, Mar 23 1994 1:24 pm
Subject: Re: How do ICS ratings relate to the world?
Reply to Author | Forward | Print | Individual Message | Show original | Report Abuse

In article <2mnrv8$...@src-news.pa.dec.com>, m...@src.dec.com (Tim Mann) writes:

|>
|>
If someone cares enough to spend the time on it, I suppose we could
|> have another ratings survey to find out how much people's USCF and ICS
|> ratings differ, and use the results to adjust the ICS ratings.  (I
|> personally don't have time for this.)
|>
|>   --Tim (Milwaukee on ICS

)

I'll do it!  Ok. If you are both an ICS- and USCF-rated player, send me a
short letter with the following information:

1. Current ICS blitz rating and number of games played
2. Current ICS standard rating and number of games played
3. Latest USCF rating and approximate number of games played
4. Your own opinion about how accurate your USCF rating is. For example, if
        your latest USCF rating is 954 because your last rated tournament
        was in 1973 but you've been studying chess every day since then, let
        me know about the discrepency.

If I get enough responses, I'll post a followup article that shows the
corrolation (or lack thereof) between ICS an USCF ratings.

--ste...@code3.com