Appendix: Conformal Bayesian Computation

To us, in the hands of an expert analyst with careful prior elicitation, the Bayesian conditional argument is the more persuasive for posterior and predictive uncertainty. However, to make such strong statements, the Bayesian must usually make the strict assumption of the model being well-specified. At the end of the day, the Bayes and frequentist answer different questions, and the common confusion arises when treating them as answering the same. As long as we are aware they are addressing different needs, we believe both solutions are informative and useful, and indeed that is our recommendation in this paper. In practice, it may be helpful to compute both the Bayesian and CB intervals and compare.