Thanks to Bruce Rannala for an interesting explanation of the problem.
I have always been interested in a Bayesian approach to phylogeny
estimation. There was a paper in the mid-80s as I remember, that proposed
a bayesian approach, but it didn't get any press (I think some authors have
proposed it since as well).
If I remember correctly from my stats theory classes, it should be possible
to arrive at a best estimate distribution from nearly any starting point.
But is this method hampered by the same dependence on a "particular family
of distributions over the sample space?"
I'll do my best to come up with the original reference, but don't hold your
Dr. Thomas M. Boyce FAX: 61-6-2464173
Entomology, CSIRO Telephone: 61-6-2464159
GPO Box 1700 Internet: thomasb at spider.ento.csiro.au
Canberra, ACT 2601