Wednesday, July 13, 2005

Can't think of a title

Economist's View has a very interesting post which looks at the problems that string theory has and notes how similar they are to the problems of economics. The comments are definitely worthwhile reading. Of course, like Mark Thoma, I can't get enough of this stuff.

Mark Thoma sums it up:
String theory has a problem. It may be a theory of everything, but it may be vacuous, a theory of nothing. This article describes the various efforts to test of string theory in some detail, but I am interested in a different aspect the article. Physicists are grappling with what to do if the theory is untestable with today’s technology. If that turns out to be the case, is pure mathematics enough?
It's easy (easy may be the wrong word) to build a model to explain wht we already know. What you'd like is for a theory to deliver testable implications for which the answer is not known in advance (prediction out of sample for example).
Commenter Lee Arnold:
If you settled for mere consistency with what you know about the world, you might end up with Ptolemaic economics. Reasonable and perfectly matching, but incorrect.
At some point you have to stop looking at the world. We don't want something that just describes the set of observations about the world that we posess, we want something that goes beyond that, describes observations we don't posess. And to do that, we need a model that is "right," not just representative of the data. So, I suppose, you have to invest your own intuition in the hope that nature and society end up making sense.

Commenter Lee Arnold again:
I think economists have another big problem in addition, and it is one they do not like to talk about. The economy is a complex system, and no matter how good the theory is, the economy may remain unpredictable.
And on Cafe Hayek, a similar but unrelated line of thought:
The economy is not an engine where more gasoline or more oxygen have predictable consequences for the speed of the engine and little else.

The lesson here is to avoid metaphors taken from physics and engineering that are inevitably cause and effect metaphors and think instead of metaphors from biology where results emerge from the actions of multiple interactions in a complex system. Think rain forest not engine.
Maybe I'll have something to add later.


Post a Comment

<< Home