Simon, the point of discussing it on the blog is to help people who were confused by the similarity of names (not a hypothetical scenario, it did happen). And yes, if you are in a many-worlds situation of any type then the optimal prior is a distribution, albeit one that you will never realistically be able to compute.