playBunny: I believe that on this site the rating is calculated after the first game. Since it is only one game it is prone to move violently up or down depending on the result.

If they were calculated on the average of the first 4 it would be a slight improvement, but still too few to really mean anything.

In the real world of chess your first 4 games are probably against very established ratings so the results are probably much more accurate than here, particularly in games where luck plays a role (such as backgammon), I believe that is why the call for using a different (standard) system for bg.