Saturday, September 08, 2007

The Irrational Discipline

The Social Atom blog has been pleasant because of the wide-eyed amazement of a very modern physicist with a solid rep in networking and complex systems happening upon the silly, overrated discipline that is economics. Especially that discipline's insistence that humans are rational. Since so much of what we do in corr sent is also based as on the same assumed rationality that is never as prevalent as theory would have it. Here are some of the better excerpts of his most recent post:

But I'm still not convinced that anyone can really predict and manage an economy on the basis of anything more scientific than the reading of tea leaves. As far as I can tell, economics, especially the "macroeconomics" of entire economies, is still very far from being a real science. (I wonder if any readers out there -- perhaps, some economist -- can point me to any data showing that economists can predict the movements of economies with anything like the accuracy one routinely finds in the rest of science? I'd be very interested to see such evidence.) This post from the Leiter Reports, although three years old, gives an interesting (legal philosopher's) perspective on economics as a science, and it's one I find myself largely agreeing with. Economics has been so committed to rational behavior, to proving mathematical theorems (which makes it look scientific), and even worse, perhaps, to the notion that all systems can be understood by supposing they are in some equilibrium, that it's tied one or both arms behind its back in its efforts really to understand economic phenomena.

Remarkably, these theoretical addictions still persist. An economist told me that it's still very hard to publish in some of the "best" economics journals if one does not assume that economic agents are fully rational. This is a problem in the sociology of the economics discipline. But it's changing, of course, as it would have to if economics is going to grow into something more intellectually sound. One promising development, for example, is a great body of work in
computational economics, in which researchers aim to make plausible assumptions about human behavior -- based hopefully on empirical evidence -- and then use computational models to explore outcomes in large systems of interacting agents.

Which, as I've been whining for months now, is exactly what criminology and sentencing academe should be doing, and would be really cool if we figured out how to do it in Second Life.

No comments: