I am very pro-empiricism in economics. I see an important difference between a field like math and fields like econ, physics, chemistry, etc. Math is a game played with symbols, and deduction from axioms is often how you play the game. But the other fields are meant to describe the real world. Hence what matters is how well our models work, not just how elegant they are. Like many smart people, I am very attracted to elegant proof-building (my B.S. is in math), but I think its misguided to confuse that clean game with the dirty, messy task of trying to figure out the world.
Now its true that the universe sometimes works on simple rules, (ie the 4 forces in physics), and so axiom-play can be an effective tool. But I think this is much less true in economics than in physics or chemistry, and we should avoid the temptation of thinking the dismal science is as abstract as the hard ones.
I guess I'm skeptical of how much you can "find out more and more about reality" without constantly checking your assumptions and results against reality. ie how do you pick the assumptions to be made about humans? If not empirically, why will the results have anything to do with reality?
- Music:Astral Projection-Faithless - God Is A DJ (Astral Projection Remix)
Profile
![[info]](http://l-stat.livejournal.com/img/userinfo.gif)
patrissimo- Patri Friedman
- Patri's World
Comments
The real issue is that it's not clear where the content comes from, or what the best way is to articulate it. Ludwig von Mises seems to have been influenced by the tradition of German transcendental idealism, which posited that there are some things that are given automatically when some other thing is given. For example, if you are given "every line has exactly one parallel through any given point", you also are given "the angles of a triangle add up to 180 degrees". The only point at which observation enters into the question is to determine what, in fact, you are given. The transcendental idealists were also keenly interested in empiricism, and saw this kind of knowledge as the natural complement of empirical experience. All experience is structured according to these "transcendental" principles.
So, when it comes to economics, Mises said that we are given a certain manner of experiencing, valuation. And there are certain things that follow from that, as surely as that the angles of a triangle on a flat plane add up to 180 degrees.
The transcendental idealists criticized the conventional way of presenting geometry, however; they believed that it cultivated the false impression that the axioms were arbitrary and that the deductions were purely negative (ie., "this is so because denying it leads to a contradiction"). Schopenhauer, especially, insisted that a better way of presenting mathematics was possible, one that made mathematics immediately relevant to lived empirical experience. But he didn't actually produce an example of how that might look. He just cited Euclid (and Spinoza) as negative examples. I guess we'll still have to wait to see how a speculative science can be as organized as Euclid's geometry while nevertheless not looking like it's all based on arbitrary assumptions that are pulled from the ether. I believe it's possible.