The absolute fibs rating is a function of the GNU checker and cube error rates, and as you might expect, the higher the rate the lower the fibs rating. So any regularity you see is expected. But it might not be the right number.
The GNU cube error rate had to be revised because the isCloseCube function was broken. How that influences the fibs ratings I do not know, but I would be interested in hearing some statistics from your data set (per match length) 1) actual wins/losses 2) wins/losses predicted by absolute fibs rating 3) wins/losses predicted by luck adjusted result. Christian. _______________________________________________ Bug-gnubg mailing list [email protected] http://lists.gnu.org/mailman/listinfo/bug-gnubg
