So we're using the uniform upper and lower bound to essentially come up with upper and
lower bounds on this term by extracting it out from the tail of some arbitrary policy.
So this is