This trail of thought is not logically sound. The whole argument is built on the notion of what modeling is and isn't, but the original quote doesn't make the assumptions you imply. Sure, they might have a model that assumes cows are spherical and etc, but it might also very well be that they have hard constraints that make it work only for spherical cows and nothing else. There isn't enough information in the quote to make the argument you're trying to make.