This year the Ratings Committee (RC) had a full plate of
tasks that kept us quite busy.  Our main accomplishment was
to work out the final details of the title system, which
will hopefully be rolled out sometime this year.  We also
addressed other issues including testing modifications to
the Quick Chess (QC) system that incorporate game results
from slow time controls, revisiting the FIDE to USCF
conversion for youth players, and performing diagnostic
analyses to monitor the rating pool.  A few other
minor issues were addressed as well.

The bulk of RC work this year was to finalize the details of
the norm/title system.  In June 2008, the decision was made to implement
a version of the norm/title system that was approved by the Delegates
for adoption back in 2003.  The basic premise of this title system is
to award permanent titles based on sustained performances at particular 
rating levels.  As an example, a player who is vying for the 1800-level 
title would need to demonstrate several qualifying tournament performances
in which his/her game results would be considered sufficiently impressive 
for someone rated 1800.  For each qualifying performance, a single norm 
is awarded.  Once five norms are collected, a title for that level 
is issued.  Norms and titles cannot be lost through poor performance or
inactivity.  A player may be working on several titles simultaneously.

The main work on revising the title system was in the technical details
and the formulas for earning a norm in a single event.  The final 
set of formulas are substantively different from the previous versions
of the norm/title system, but have the advantage that they are much
simpler and that they have been validated based on historical USCF
tournament results.  The earlier rule that multiple norms could be
awarded from single events has been eliminated.  The current version now 
uses post-event ratings rather than pre-event ratings in the formulas, 
which permits results against unrated players to be incorporated into the 
norm calculation in a straightforward manner.  This version also
implements a rating requirement for titles at the 2000-level or higher in 
which a player must have or have had an established rating at the 
title level in order to earn the title.  A document containing the 
complete description of the title system can be downloaded from 
http://math.bu.edu/people/mg/ratings/titles-0509.pdf .

In 2008, the RC updated the formula for converting FIDE ratings onto
the USCF scale using a methodology that identifies all players active
in both USCF and FIDE events, and determines a statistically 
accurate relationship to map FIDE ratings to USCF ratings.
In addition to using this conversion formula for rating players
in USCF events who are USCF-unrated but have a FIDE rating,
the conversion is used in updating USCF ratings of players who compete 
in foreign FIDE events that are not normally rated by the USCF.
In January 2009, after correspondence with the parent of a child
who competed in the 2008 World Youth Chess Championship (WYCC), it became
clear that the FIDE-to-USCF conversion formula derived in the prior year
was not appropriate for youth events.  The problem is that 
the US participants were selected for participation in the WYCC
through their very high USCF ratings, so that their USCF-FIDE 
differences can be expected to be larger than average due to the 
selection process.  In other words, there is an inherent selection 
bias that results in larger-than-expected USCF-FIDE differences.
To derive a more appropriate conversion, analyses were run on
a set of players with both USCF and FIDE ratings who were
under 20 years old and who had the highest USCF ratings in their 
age group.  This produced the following simple conversion formula:

USCF = 210 + 0.93*FIDE

The recommendation was that this formula be used for the World Youth
Championship, the Pan-Am Youth, and the World Junior, as these are
all events that select for top USCF-rated youth players.  For all other
FIDE events, the usual formulas derived last year can still be used.

A task that we addressed in mid-2008 was the ongoing problem
of QC ratings being substantially out of alignment with regular 
ratings for a non-trivial number of players.  Starting August 2008, 
the RC considered a proposal in which the QC system would rate all 
games slower than G/60 but using half-K in the established rating 
formula.  This suggestion seemed promising, and both the RC and 
the EB liaison were in agreement that this was a potential direction 
to proceed, but an agreement could not be reached on how to initialize
the change.  Mike Nolan tested several systems, including one that
rerated all events going back to 1/1/2004 using the proposed change,
but the change in the divergence between QC and regular ratings was
minimal compared to doing nothing.  The RC proposed to lower the 
effective number of games on which QC ratings were based starting 
1/1/2004 and then rerate all events, but the EB liaison felt that 
this proposal violated the "once rated, always rated" USCF policy.  
The proposed methodology has therefore been shelved.

The RC addressed several minor issues over the past year.  The committee
was asked to comment in September 2008 on whether time odds games
should be rateable.  The question was motivated by the USCF office
noticing the existence of events that had time controls of G/80 for 
white and G/85 for black.  The office left these events as rated, but 
the RC chair noted that time odds games should ordinarily be associated 
with a rating advantage for the player with the slower time control.  
The relationship between the rating advantage and the time odds
would require further study.

A second minor issue presented to the committee had to do with
the increasing number of players returning to tournaments after a long 
hiatus.  Such players often have stale ratings that are no longer valid 
estimates of playing strength.  The suggestion by the RC chair was
to create a new field in the database indicating the effective 
number of games which would ordinarily contain the results from the usual
rating formulas, but could be set to a low number in case the player
has not competed in tournaments in a long time.  The effect of lowering
the effective number of games is that the player's rating would 
experience greater rating changes after returning to tournament play.
The issue has not been addressed beyond this discussion.

A third minor issue was a discussion about whether blitz time controls
should be part of the QC system, especially in light of the rules for
blitz chess being slightly different from non-blitz (e.g., an illegal
move loses a game).  Blitz chess became rateable in the QC system
in 2003 under "emergency" conditions in which the EB did not consult the RC,
and the committee never had a chance to evaluate the EB decision.  
The opinion of the RC based on discussions from 2003 was to have a 
separate blitz rating system.  We understand that an ADM has been
submitted to stop rating blitz chess in the QC system and to 
create a separate blitz rating system.

Each year the RC performs a set of diagnostic analyses to monitor
trends in the rating pool.  Overall rating levels have deflated 
from the mid-1990s through 2000 when rating floors were decreased 
by 100 points without a counteracting inflationary mechanism.  
With the new rating system implemented in 2001, ratings started
to re-inflate.  The RC has the goal of restoring rating levels
back to where they were at the end of 1997.  The focus of RC work
has been on players with established ratings who have been
active over the current and previous three years and who are aged 35-45 
years old in the current year.  Based on the continued decline in 
ratings for this group, the RC recommended last year lowering the 
bonus point threshold from B=10 to B=6 to accelerate the re-inflation 
of the rating pool.  This change was applied retroactively to the 
beginning of 2008.  The results of this year's analysis have
the median rating for this group increasing by 20 points, and the mean 
increasing by 10 points.  Because the average rating for this group
is still about 60 points lower than the corresponding average rating 
at the end of 1997, the recommendation of the RC is to continue with 
the bonus point threshold at B=6 and to continue monitoring based on 
the annual rating lists.