If the evil demon theory made specific predictions that turned out to be right and weren't predicted by other theories, then it would be a useful theory. It's possible for a theory of a super-universe to do that.
Imagine if we came up with a super-universe theory that said that the super-universe could only spawn sub-universes which followed conservation of energy plus several other laws no one had ever thought to test before (and weren't implied by any other theories), and as we started testing for those other laws, every single one we tested turned out to hold in our universe.
(If that happened, it would be reasonable to look for simpler theories that also predicted those other laws too, but it's possible that the super-universe theory would turn out to be the simplest possible theory that fits. Theories should be judged by the complexity of their rules, not by the number or complexity of things they predict; a simple theory that implies a large ensemble of universes can be better than a more complex theory that implies only our world or what we can see is real.)
Imagine if we came up with a super-universe theory that said that the super-universe could only spawn sub-universes which followed conservation of energy plus several other laws no one had ever thought to test before (and weren't implied by any other theories), and as we started testing for those other laws, every single one we tested turned out to hold in our universe.
(If that happened, it would be reasonable to look for simpler theories that also predicted those other laws too, but it's possible that the super-universe theory would turn out to be the simplest possible theory that fits. Theories should be judged by the complexity of their rules, not by the number or complexity of things they predict; a simple theory that implies a large ensemble of universes can be better than a more complex theory that implies only our world or what we can see is real.)