A good discussion ongoing at Scott Aaronson's blog, of interest to many here. What's missing, in my opinion, is epistemological awareness of the role of the observer in any model (at any level.) You can never *really* account for your own prior. <http://scottaaronson.com/blog/?p=326>