Re: [pbmserv-dev] Re: Backgammon Standings

["David J Bush" <twixt@cstone.net>]

| I don't like the idea of players sitting at a high ranking without 
| playing.  However I don't like the idea of pulling someone out of the 
| rankings suddenly, nor do I like the idea of someone returning a year or 
| more later and suddenly being at the top.  I prefer a system where the 
| rankings of inactive players slowly degrade.  ...

This reminds me of the Glicko rating system:

http://www.glicko.com/glicko.doc/glicko.html

Two values are computed, a rating and a "ratings deviation" or RD which
increases over time if the player is inactive. Using this system, there
would be no distinction between provisional and established ratings.
Ratings would be computed the same for all players. New players would
start with a high RD value, which decreases as the player plays more
games.

Perhaps the current "established ratings" list could be replaced by those
players whose RD values are below some threshold value. If you become
inactive, and your RD goes over the limit, you will be taken off the
main list, and returned to the list of "inaccurately rated players" or
whatever you want to call it. If you start playing again you can return
to the "accurate" list.

This system was developed by a statistician, so presumably it would
result in ratings which correspond well to the players' strengths.
It also addresses the issue of inactive players. It was developed for
chess, but hopefully would apply to most other two-player games here.
After all, the same rating system is in place for all 2 player games here
now. (Isn't it?)

David