playBunny: First, let me say I have insisted before that there was a bug in the ratings calculator, and I'm glad to see it being addressed.

Prior to today, if I played anyone whose rating was within 400 points of mine, the result was either +8 or -8 -- nothing in between. As a result, assuming that I won more than half my games and only played people whose ratings are within 400 points of mine, I was pretty much guaranteed an upward trend (with random dips, naturally, but the overall trend should be upward). If you will look at your game pages now, you will find that your expected BKR gain from winning against anyone lower rated than yourself is less than the expected BKR loss from losing. This is the way the ratings system should work, and it should have the effect of squeezing the ratings closer together. It should also get harder to improve your rating the higher your rating goes, which is now the case. I suspect the bug that previously existed was the reason that Fencer felt the need to remove the 2700 ceiling on ratings.

Looking at my games, I stand to lose many more points than I stand to gain in each one. Since backgammon is a game of both skill and luck, I will certainly lose a significant number of games -- probably at least 35% against all opponents, very likely more. This is why I expect my rating to go down.

Moreover, as I interpret Fencer's announcement, he is going to recompute all ratings from the beginning, using the historical database of results. Thus, current ratings (like mine) that have been inflated by the bug should decline. It remains to be seen what the relative changes will be.

From what I remember about FIBS, it was said that your rating reflected primarily your last several hundred games -- i.e. that your rating wouldn't be too different if you applied the FIBS formula to the last 400 or so games than it would be if you applied it to your entire history. If that is the case, I may come out of the process in good shape, since I have been rather fortunate lately.

It's been a fun ride at the top, but I won't be upset if the recomputation leaves me somewhere in the middle of the pack.