Finally, Bayes has no good way of dealing with […] with cases where the data was generated in a complicated way that could make it highly biased […].
Wait, what? If we don’t know about the possibility of bias, we’re doomed anyway, are we not? If we do know about it, then we just have to adjust our prior, right? Or is this again about the intractability of true Bayesian computation in complicated cases?
Wait, what? If we don’t know about the possibility of bias, we’re doomed anyway, are we not? If we do know about it, then we just have to adjust our prior, right? Or is this again about the intractability of true Bayesian computation in complicated cases?