Sure. Even Zurek has said that MWI may need to postulate that "there are systems". The model of detection begins with a microscopic interaction that is then amplified and decohered. If the universe can't be modelled this way then pretty well all of science falls apart.No. Why should they be? There is only one Hamiltonian, that for the whole universe, and there is no reason at all why this Hamiltonian should have separable eigenstates. Almost no Hamiltonians, except very simple contrived ones, have this property.