If you extend OR to theories you should select the one that has the least ad-hoc hypotheses (or the smallest set of auxiliary hypotheses in the Lakatos sense).

If you extend OR to theories you should select the one that has the least ad-hoc hypotheses (or the smallest set of auxiliary hypotheses in the Lakatos sense). In other words one could argue that theories get "worse" with more auxiliary hypotheses so from a scientific process point of view if you diligently falsify and force the creation of more auxiliary hypotheses you can weaken a theory enough for a (more elegant) alternative to take its place.