We are pleased that our work onward inference in mixture models is drawing attention.

We are pleased that our work onward inference in mixture models is drawing attention, although businessed that Webber finds it misleading. Webber takes issue with the Bayesian approach in general, and believes that our succes was driven by the agency of overly informative priors and constraints that "improperly distort estimates."

First of all, Webber states that a posterior mean is incomparable to ML because it is not invariant to changes in scale. Many estimation principles (eg unbiased estimation) are not scale invariant. Are we to avoid using unbiased estimates too?

By itself, the plat of the gamma and inverse-gamma densities says little