Shit, actually, one other thing on that ordinal utility theory. As I said, it is an advance, insofar as it injects economics with a healthy dose of relativism (which is essential for democracy...) but I would argue that it doesn't really go far enough. There's a great passage in one of his seminars where Jacques Lacan is discussing the behaviorist observing the rat in the maze. Lacan jokes that the behaviorist himself - that distant scientific observer - is already deeply caught up in the experiment. We can be sure, he points out, that the behaviorist, if he didn't design the maze himself, he certainly chose the design. Lacan jokes that the poor little rat is running around in a material embodiment of the behaviorists desire - what the behaviorist finds in his experiment is his own desire staring back at him. His own desire to watch a rat run around a specifically shaped maze - its quite funny really.
This all sounds a little dumb, but it has serious practical consequences. Once the economist/social scientist is given direct subjective input into his theory - for example, when he determines what/how he will survey the consumer's behavior in the marketplace - you can be sure you'll find traces of his own desire in it. From Lacan's bad rat-maze jokes, I'll let Wiki have the last word:
Another entirely different problem is whether ordinal utility is indeed an observable variable in real world. For example, in closed hypothetical system like above, there are only orange and apple to choose from. Therefore, when an individual choose orange, one can definitely say that an orange is preferred over an apple. However, in real world, when a consumer purchases an orange, it is often impossible to say what good or set of goods or collection of behavioural choice (including not purchasing anything at all or doing something else) were discarded as options.