Mike Lynch and the tragedy of the Bayesian