That's a good question. In the Discussion Thread that accompanies these posts, a couple of other netters have asked about the distribution (variance) of ratings for any one show. There's actually a (small) literature about what happens when people see the distribution of of ratings for a given product. One study of online book sales found that for any two books with the same average rating, the book with the greater distribution of ratings had higher sales. In the context of Phish shows, I think this might mean that, of two shows with the same rating, more people would be driven to listen to the show that had a wider variety of ratings.

I've not done anything with "within show" variance of ratings, but that could be really interesting.