"Evidence in economics is messy---and political. Trust logic instead."

The problem is, Rothbard's logic is also every bit as political as, say, Marx's logic.

You're talking to someone who was a confirmed libertarian, Austrian sympathizer, and Ayn Rand enthusiast from roughly the age of 16 to maybe 20 or so. I know the arguments and I know inside and outside the kind of thinking they come from, which is simply put, a tendentious effort to justify one particular political ideology over another. That is the exact opposite of what any science should aspire to, even economics. And that is the sin Rothbard is guilty of, a sin he relentlessly accuses others of in order to distract from his own guilt.

This is not to say he in specific, or the Austrians in general, were totally bereft of good ideas--just that their ideas should not be taken as proven unless tested. You're suggesting the scientific equivalent of "the code looks reasonable so there's no need to test it".

I certainly don't agree with everything Rothbard or other Austrians say, and indeed I agree that their greatest errors are driven by ideology. (And despite my general sympathy with Ayn Rand, this goes triple for her and her followers.)

That being said, a substantial subset of economics is analogous not, as you say, to a programmer saying

  the code looks reasonable, so there's no need to test it

but rather to a mathematician saying

  the proof is correct, so there's no need to test it

That is, what is the point of testing the Pythagorean Theorem? Perhaps we should measure a bunch of triangles? Why would you do that, when you have a proof? And yet, many people seem to believe that, e.g., raising the minimum wage might not (ceteris paribus) raise unemployment. They even sometimes collect "evidence" that it doesn't! (See, e.g., the paper by Card & Krueger showing a rise in employment after a minimum-wage hike, which Bill Clinton used as political cover to get Congress raise the minimum wage.)

Of course, you'd better be double-sure your proof is right, and I agree that the Austrians sometimes fall down on this point (Rand, triply so). But their approach is right: economics, correctly done, is much closer to mathematics than it is to physics.

N.B. I too am a reformed libertarian-Objectivist, and I too know the arguments forward and backward; I could crush a traditional libertarian in a debate, because I know the weaknesses all too well. But it seems that my ultimate parting of ways with them has taken me in the opposite direction from yours. You see, I'm even more radical (or, rather, reactionary) than they are. :-)

Mathematics doesn't deal with the real world, nor does it use induction (in the normal sense of the term--mathematicans use the word "induction" to refer to a type of recursive proof method, but inductive logic is the method of reasoning general rules from particular pieces of evidence). Triangles are an abstract concept that abide to rules that we make up, and if we make up a different set of rules, triangles no longer obey them. That's why simply measuring triangles doesn't work.

Economics does deal with the real world, and like any other science it's subject to the demands of empiricism rather than merely the demands of logic. You can use logical and mathematical reasoning to come to a lot of useful conclusions in physics, for instance, but we're constantly substantiating those by experiment as well, and occasionally we are wrong. (Use Newtonian mechanics to calculate the orbit of Mercury and it won't match.) Do you really think that human behavior is so much more simple than physics that we can model it without evidence at all?

Ah, so we don't really disagree at all---or, at least not much. It's my fault for using a bad analogy, and you hit on a better one. You're right, the axioms, whether in economics or physics, must come from the empirical world. Given the axioms, you can then prove theorems, but of course the axioms might need tweaking---which can dramatically change the theorems. What I see in economics are cases where people implicitly accept particular axioms, and yet reject many of the consequences (i.e., theorems). It's as if they agree that Newton's law of gravitation is correct, but don't believe that two-body orbits are closed.

Though complex in general, there are aspects of human behavior that are consistent and quite simple, and serve as a solid foundation for a study of human action. (You probably recognize this as the Austrian approach of praxeology.) When I read the Austrian argument that any fixed amount of money-stuff (e.g., gold) is sufficient for exchange, I understand the axioms, follows the steps, and accept the conclusion. When I read the Keynesian arguments for the benefits of monetary expansion and "fiscal stimulus", I see---great clouds of fog. Their arguments are often too vague to untangle, but I can see that their conclusions (a) appear to be based on the same axioms I accept and (b) disagree with a theorem. Something doesn't add up.

In sum, the value of empiricism lies in finding the right axioms. If you discover "empirical evidence" that a theorem is wrong, you must either propose a change in the axioms or find a flaw in the proof. Contemporary economists often appear to do neither.

(The Card & Krueger example is again instructive. If you raise the price of labor, people buy less of it, so minimum wage hikes must increase unemployment. And yet, from a contemporary NY Times article (http://bit.ly/5afSyk), we find this:

    They [Card & Krueger] surveyed 400 fast food restaurants and found that
    those in New Jersey actually added 2.5 workers after the minimum wage went
    up. In Pennsylvania restaurants, meanwhile, payrolls shrank.

    Why should bosses hire more workers if it's more expensive to do so? The
    two economists speculate that any fast-food restaurant typically operates
    with a couple of vacancies it can't fill because it doesn't pay enough.

    "If you raise the minimum a little," said Professor Card, "teens who are
    sitting [at] home go out looking for jobs." 

Apparently, these restaurant managers are too stupid to realize they're allowed to pay above the current minimum wage, thereby getting those lazy teens off their couches to fill the vacancies without a minimum wage hike. Truly, it boggles the mind.)

"In sum, the value of empiricism lies in finding the right axioms. If you discover "empirical evidence" that a theorem is wrong, you must either propose a change in the axioms or find a flaw in the proof. Contemporary economists often appear to do neither."

That's not how the real sciences work, though--physics doesn't operate through axioms and theorems, it establishes a few theories and as long as they hold within experimental error, they don't complain too much. When they stop holding, we either postulate extra "stuff" we don't directly observe (most of the outlying planets as well as the Kuiper belt were discovered by measuring their gravitational effects on planets we could directly observe) or use a more refined theory on the edge cases (Mercury's orbit). But all the while physics gathers more and more empirical data, and spends a lot of time trying to make sense of it.

"When I read the Austrian argument that any fixed amount of money-stuff (e.g., gold) is sufficient for exchange, I understand the axioms, follows the steps, and accept the conclusion. When I read the Keynesian arguments for the benefits of monetary expansion and "fiscal stimulus", I see---great clouds of fog."

The other problem I find is that most armchair Austrians take their entire economic understanding from an 80 year old dispute and have no conception of what's happened in the field ever since. Keynes is a dead old man and the field of economics has moved on from his work.

Incidentally, even if you just take the approach of pure reason, it's still pretty fucking easy to debunk the idea of the gold standard.

"Apparently, these restaurant managers are too stupid to realize they're allowed to pay above the current minimum wage, thereby getting those lazy teens off their couches to fill the vacancies without a minimum wage hike. Truly, it boggles the mind.)"

And here you see one of the biggest problems with the field of economics in general, a problem that's only been tackled in recent decades--modeling how actual human beings behave in the real world rather than supposing that each economic actor is fully informed, perfectly rational, and perfectly self-interested. How many 16 year old kids would know whether or not Burger King paid above minimum wage? Of course Burger King only pays minimum wage. It's a cultural expectation. The same thinking applies to the manager--is a fast food manager really going to be fully informed, perfectly rational, and perfectly self-interested? For the most part, his job consists of executing a manual written at the corporate headquarters in a different state which dictates that crew are paid local minimum wage, and quite frankly, the corporate overhead of rewriting the corporate manual to allow wage adjustments to recruit people to work at a fucking Jack in the Box isn't justified.