Reminds me of Jaynes point that IIRC the beliefs of two perfectly rational agents with very different Bayesian priors can diverge even if they are presented with exactly the same facts/information/data...