Good point in that the aesthetics of a model are in the eye of the beholder - although:
> that does not make a simple model better than a complex model that matches observation better
Perhaps this is less of an issue in physics than some other fields because the observations are less noisy (?) -- but if getting the best match for existing observations is all you care about from a model, you could end up with something grossly over-fitted. I guess it goes without saying that any proposed physical laws need to pass the equivalent of cross-validation :)
> that does not make a simple model better than a complex model that matches observation better
Perhaps this is less of an issue in physics than some other fields because the observations are less noisy (?) -- but if getting the best match for existing observations is all you care about from a model, you could end up with something grossly over-fitted. I guess it goes without saying that any proposed physical laws need to pass the equivalent of cross-validation :)