I don't want to force one notion upon another, but it seems one could make a case for this view. At maximum potential the molecular vibrations of the pendulum material would be statistically more random, "incoherent"; at kinetic maximum they would share a single vector of motion, statistically...