Aligned to Whom?

You are relying very heavily on the priors of the model to do a good job for you. What happens when you cannot judge whether those priors are good?