“Yet, if all models are wrong, there is certainly something to be said for sticking with a simple one that is known to be wrong rather than a complex one that could be wrong in more obscure ways.”