This can be used by code that depends on policy to re-evaluate
choices they have made in the past that could now change based
on the new policy.