gjm comments on Kalman Filter for Bayesians