The Ratings Committee (RC) this year tackled a fairly short
list of tasks.  The main tasks involved addressing a question 
concerning ratings near the absolute floor of 100, and
how to ensure greater correspondence between Quick Chess
(QC) ratings and regular ratings.  We also carried out a 
detailed examination of rating changes in the rating pool.

Correspondence with Mike Nolan in November indicated that a
much greater number of players were at the absolute
rating floor of 100 this year compared to previous years.
The RC was asked to consider possible responses to this
issue.  After some discussion, the RC offered the proposal
that if an unrated player earns a rating less than 100
(before the player ends up with a floor at 100), then the
player should remain unrated as though he/she had not yet
played any games.  This proposal borrows from the FIDE model
of requiring a minimum performance to receive a first
rating.  In response, the Executive Board (EB) decided that it was
preferable not to take any action, arguing that having an
increased frequency of 100 ratings may be indicative of
an increase in very weak players.  They also preferred to
continue the practice of issuing ratings once a player has
played at least four games.

An ongoing task from the last couple years has been how to 
address concerns that QC and regular ratings are out of
alignment for players having both.  The solution we proposed
was to replace the QC system by one that rates all events, while
keeping the regular system which rates events with time
controls G/30 or slower.  The revised QC system would likely be
called something other than "Quick chess" - an alternative that
was discussed was the "universal" system, but other names are
possible.  In implementing such a system, players would be
considered unrated under the new system going back to the
start of 2004, and then the "universal" system would be
applied prospectively to all events from that point.
There are several positives to this proposal.  First,
one's "universal" rating is equal to the regular rating until a
player demonstrates evidence otherwise through a good amount of fast
time-control events.  Second, results from regular events will
affect "universal" ratings, but not vice versa (for events with
time controls quicker than G/30).  Finally, any differences that
persist in "universal" and regular ratings are arguably a demonstration 
that the player's quick and regular strengths differ.  The EB
had a positive reaction to the proposal.  Currently, Mike Nolan
is testing different strategies for initializing the "universal"
ratings in addition to the method mentioned above.

The ratings committee was also asked to address a few smaller
concerns.  We were alerted by the USCF office that the USCF was 
considering becoming involved formally in bughouse and Fischer
random chess.  The RC was asked about whether these variants
could be rated by the USCF.  The RC chair responded by
mentioning that the likelihood was that separate rating systems
would need to be constructed, but doing so was not problematic.
When the USCF is closer to overseeing bughouse and Fischer
random chess tournaments, the RC will become more involved in
rating discussion and development.  A second issue that arose
was whether the RC should be involved in updating the USCF correspondence 
rating system.  This was motivated by some lack of clarity in
the online explanation, especially in the description of how
the system handles provisionally rated players.  The RC chair 
is currently in discussions with Mike Nolan about making small 
changes in the correspondence rating formulas, incorporating some 
ideas from the current over-the-board system to improve rating behavior. 

Every year the RC performs a set of diagnostic analyses to monitor
trends in the rating pool.  As is well known, 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 in place, ratings have begun to reinflate.  As a rough
goal, the RC has been intending to restore rating levels
back to where they were in 1997.  Typically, the RC has focused 
attention on players with established ratings who have been
active in the past three years, and who are aged 35-45 years old.
Our analysis this year was more expansive.  To summarize, we
examined the difference in average ratings (for players
active over the current and previous 2 years) between years 1997
and 2000, and between years 1997 and 2006.  The results are
shown in the accompanying figure.  The dashed line, which 
displays the average rating difference between 1997 and 2000
as a function of age (the curve is called a "locally weighted
scatterplot smoother", or "loess"), indicates that ratings in 1997 were
higher than ratings in 2000 by about 60-70 rating points for 
players under 30 years old, and about 40 rating points
for players older than 35 years old.  The dotted line, which
displays the average rating difference between 1997 and 2006,
shows a much more optimistic picture.  On average, ratings
as a function of age are at most 20 rating points different
in 2006 compared to 1997.  This result indicates some degree
of success of the new rating system with its bonus and
feedback mechanisms, as well as higher K for lower-rated
players.  The RC will continue to monitor the rating pool
for anomalous changes over time.