Would be cool to wrap up this section with a question like: “but could we know the probability of the data given the model?”