After watching an Inconvenient Truth and reading about the IPCC consensus, I assumed that global warming was an established reality.  But a little while ago I stumbled across a couple of curious blog comments by one Mencius Moldbug.  Mencius is a software engineer out in San Francisco.  After scoring in the dot com boom, he went on sabbatical in order to read, think, and write.  He studied the whole global warming debate and came away with some interesting findings.  Below I've tried to compile the long comment thread into one coherent post.  Mencius also writes a blog here.  Unfortunately, he has not blogged about global warming yet.  Since this is the single most compelling case for global warming skepticism that I have ever read, I wanted to re-post his comments in a place where everyone could read them.

---------------------------------------------------------------

Mencius:

The set of people who support or oppose a proposition is quite unrelated to its validity. So, even, are their motivations.

I'm sure there are some businessmen and politicians who oppose the global warming hypothesis for venal reasons. I'm sure there are some who oppose it for sincere reasons. And I'm sure both categories also apply to the large number of businessmen and politicians who support it. I don't see how any information can be deduced from these incontrovertible facts.

Do you have a minute to follow a link? You might find this page interesting. It is rather free of context, but perhaps its tone might spark your curiosity.

Now, what is your first question on reading this? Is it, by any chance, "Is Steve McIntyre a scientist?"

You see, there are two possible definitions of the word here.

One is that a "scientist" is anyone who follows a certain process which, in the past, seems to have been very effective at producing reliable information. This process is called the "scientific method," and while it is certainly not infallible - nothing is - it seems pretty good at eradicating error. And it works for anyone, black or white, Gentile or Jew.

Another is that a "scientist" is someone who has been awarded a certain credential by a certain institution, which entitles him to occupy one of a small number of prestigious positions. In this sense, calling someone a "scientist" is like calling him a "duke." The word "duke," from the Latin dux, originally meant a leader of fighting men, a warlord. There are still dukes today, and there are still warlords, but the dukes are not warlords and the warlords are not dukes.

As it happens, McIntyre is a scientist (I would say) in the first sense of the word, but not in the second.

His background is actually in hard-rock mineral exploration, which is one of the shadiest industries in the world. As a mining consultant with a strong background in statistics, he essentially made his living investigating bad numbers, and he developed an excellent practical sense of the ways data can be fudged.

One day out of pure personal curiosity he decided to take a look at the use of statistics in paleoclimatology. Paleoclimatology is a field which tries to estimate temperature trends from the time before we had scientists running around with thermometers everywhere. It does so by measuring "temperature proxies" which naturally record temperature effects - tree rings, for example, As you can probably imagine, before the rise of global warming, this was a rather obscure discipline, but it has since risen to great global importance.

The first result McIntyre looked at was the famous "hockey stick" curve associated with several notable paleoclimatologists, notably Michael Mann. If you have ever seen a headline saying "200x is the warmest year in XXX years," you have probably experienced the "hockey stick." On the basis of this result, Mann (no relation to the Miami Vice director) had become a star in his field, was named one of Scientific American's top 100 young researchers, etc, etc.

What McIntyre found was that this work was based on a pattern of bad statistics that came very close to simply being fraud. Mann had chosen nonstandard statistical procedures which amplified a single sample, from a set of trees (bristlecone pines) well-known to respond directly to CO2 rather than temperature, into a pattern that looked like it covered the entire world. Moreover, Mann's FTP site had a directory called "CENSORED" on it in which the same calculation was repeated without the bristlecones, showing no "hockey stick" at all.

Scientific misconduct happens. Just because a physicist, like Jan Hendrik Schön, pulls some stunt, doesn't mean Einstein was wrong.

But Schön was rapidly drummed out of his profession. Events in the Mann scandal have gone very differently.

The entire field of paleoclimatology has stonewalled on the "hockey stick." Mann is still very much an honored member of the field. And this despite the fact that two external reports, a National Academy of Sciences panel, and an independent report prepared by Edward Wegman, one of the US's leading statisticians, confirmed all of McIntyre's results.

What was the consequence of all this in the press? Funny you should ask.

The NAS panel, which included many of Mann's colleagues, came up with a very nice dodge. They admitted that Mann's results were useless, but claimed that, since other studies - many using the same flawed methodologies - reported the same results, the entire concept was vindicated. You may have heard the phrase "fake, but accurate."

As a result, the release of this report was actually an occasion for a new wave of "Earth is Warmest in 2,000 Years" stories. You had to be very, very savvy to understand the actual content of the report, which included a large dose of spin.

Wegman's report contained no spin at all. However, it also happened to be commissioned (pro bono, with no payment at all, etc, etc) by Joe Barton of the House Energy Committee. Therefore, it received no press coverage at all - despite the fact that it exposed one of the major scientific frauds of the last century, one which was lavishly promoted by the press.

In fact, not even Republican news outlets - like Fox News or the Washington Times - would touch the Wegman report. Why would they? Why go through another round of being attacked as shills for the oil companies? What's in it for them? Because they are, of course, shills - but for the Republicans, not for the oil companies.

The problem is very simple. It has nothing to do with Republicans or Democrats or whatever. McIntyre is a liberal. Wegman voted for Gore. He is not some kind of crazy AEI activist, he is the former chair of the NAS panel on statistics, who had never heard of any of this crap before Barton called him. Look at his CV - this is normal diligence for a statistician. Checking that people are using stats correctly is what a guy like Wegman does.

No. The problem is money and power and fraud, pure and simple.  Trillions of dollars are being reallocated on the basis of statistical work that would flunk a sophomore. And you're not interested? What kind of skeptical antennae do you have?

MBH99 is like a bad audit finding. When you find that Bear Stearns or Enron or Citicorp or whoever has fabricated some nonexistent asset on its books, you don't give it a slap on the wrist and tell it to cross the line out. You unleash the forensic guys. You don't find just one cockroach in a kitchen. And you don't tell the health inspector, "sorry, we'll kill it."

MBH99 is not an ancient, irrelevant result. It is the single most publicized piece of research in the history of IPCC climatology. It singlehandedly put Mann in Scientific American's list of America's 100 top young scientists. And it really is not a stretch to compare it to an ape jaw glued to a human skull. The thing reeks. It is data laundering of the worst kind.

Sure, the "hockey team" has "moved on." Despite the fact that the NAS panel chaired by North and the Wegman report, both of which concurred in all substantive opinions, both stated that further paleoclimate proxy studies should not use bristlecones or foxtails, we continue to see paleoclimate reconstructions full of them.

And when IPCC climatologists are actually willing to use phrases like "if you want to make cherry pies, you have to be willing to pick cherries" (d'Arrigo) to committees investigating their work, we really have no reason to believe that their results are generated from their data and not the other way around. Actually, we have considerable reason to believe the opposite. "Why should I make the data available to you, when your aim is to try and find something wrong with it?" (Phil Jones.)

The trouble is that the entire evidentiary case for AGW is nonreplicable and unfalsifiable. If you do not personally trust the researchers, you have no reason to trust the "science."

The first arm of the AGW case is paleoclimatology, which is trivially distorted by selection bias, ie, "good" results are published and not "bad" ones. The second is general circulation models or GCMs, which simulate the Earth's atmosphere. GCMs are the product of tuning - they are certainly not derived directly from the laws of physics. See Nir Shaviv on the fine art of fitting elephants.

Because of this episode I judge the IPCC "consensus" the way I might judge the collective views of, say, the IMF. As a bunch of very smart people who nonetheless have a very clear institutional agenda. If you insist on taking them at face value, perhaps I could interest you in an Argentine bond or two.

It would be wonderful if we lived in a world where there were these institutions, like "Science," or "The Church," or "The New York Times," which you could just trust. But such a world has never existed in the past. Why should it exist now?

The logic I've laid out is my motivation for treating the Scientific Consensus as epistemologically irrelevant to my own estimation of future climate trends. I believe I've described a social process that could produce the Consensus whether or not the AGW proposition is true.

If anything makes science Science, it's that the "scientific method" is a social process which, in the past, has shown a tendency to eradicate error from its own conclusions.

I believe McIntyre's work has shown quite conclusively that, at least in the field of paleoclimatology, this process is not now operating, at least not in the form that produced its reputation.

I also believe that, since the culture of climate modelers (such as James Hansen) does not appear to differ from that of paleoclimatologists, and since climate modelers have at least as much, if not more, of an opportunity to tune their models in the same way that paleoclimatologists tuned their statistics, I should treat the information generated by both with the same distrust.

Since AGW research consists (since it can only consist) of modeling and paleoclimatology (the former is much more important - the paleoclimate follies are very much a canary in the coal-mine kind of thing), I choose to ignore it.

For example, if climate modelers wanted me to change my decision on this, they would have to build a new climate model in a clean-room process, which (a) correctly simulated Earth's climate in the past and (b) was not made to do so after a very extensive process of tuning, with abundant opportunities for the introduction of unconscious bias.

To be more specific, the largest source of uncertainty in climate models (GCMs, general circulation models) is their handling of water. ("Besides that, Mrs. Lincoln...")

Water in the vapor phase is a greenhouse gas. Water in the solid phase, and aerosolized water in the liquid phase, reflect radiation back out into space and cool the earth.

The interaction between these phases, and the more common surface liquid phase (like WC Fields, I never touch the stuff) is chaotic, which essentially means unsimulatable.

All GCMs in operation today predict a powerful positive feedback effect from (clear) water vapor, which strikes me as a pretty convenient truth if ya know what I mean. As notorious climate-criminal Richard Lindzen points out, the increased temperature caused by CO2 can just as easily produce a negative feedback from increasing cloud cover.

I'll bet something like 1% of the people who read about global warming in the papers know that the effect is logarithmic. Doubling the CO2 concentration produces a global radiation increase that, in the absence of modeling, if you just run the numbers straight, would make Earth about 1C warmer. Double it again - same result.

In the last century, CO2 has gone from about 280ppm to about 380ppm. The level of positive feedback the GCMs need to assume in order to turn this into an epic human catastrophe is, obviously, quite nontrivial. In fact, with a little more tuning, the GCMs would probably be quite happy to turn us into Venus.

And if that result generated more funding for the people writing the GCM, that's probably exactly what it would print out. But there's a balance of credibility that has to be maintained. In fact, GCMs in testing apparently quite often turn the Earth into a snowball or a boiling planet. The fact that they work at all is a genuine accomplishment.

So what is my prediction of the weather in 2057? I have no idea, I am not a climatologist. But Nir Shaviv's view strikes me as pretty sensible.

Of course, Shaviv is just one guy. He's not a Consensus. I'm afraid I'm going to have to disappoint you on that one, although again, if you are interested in the subject, Climate Audit is the place to be.

The following is a highly edited version of a comment thread found on the blog Gene Expression.  The thread discusses the role of empricism verus deduction in the social sciences.  The star commenter PhysicistDave actually studied under Feynman, which gives him some nice cachet.  Unfortunately, Dave does not have his own blog.  I am republishing his comments so that I can send the link to people the next time I get in a debate over economics. The other commenters are Razib, the host of GNXP, and Mencius, a semi-retired (post-IPO) software engineer who spends $500 a month on books.

-------------------------------------------------------------

Razib

There was a time in the past when I was a hard-core libertarian.  Now, I'm not one of the few dozen people in the world who has actually read Ludwig von Mises' Human Action (I'd be willing to bet some gold that half of these individuals who've gotten through von Mises' magnum opus are virgins!), so my libertarian nerdishness only went so far. All that being said, there was a time I would have said I favored the Austrian School of economics.  I was a libertarian, and the Austrian School was congenial to libertarianism, ergo, I supported the Austrian School (I knew I opposed Keynesians as well as the neoclassical models).

But I'd always had issues because I knew that the Austrian school rejected econometrics and positivism; and being steeped in experimental science I'd always viewed positivism as a Good Thing.

But over the past few years, and especially over the past months, I've been doing some reading on Google Books and elsewhere on the intellectual history of the Austrian School, and especially praxeology. What the hell is praxeology? Well, from praxeology.net:

Praxeology is the study of those aspects of human action that can be grasped a priori; in other words, it is concerned with the conceptual analysis and logical implications of preference, choice, means-end schemes, and so forth.

The "grasped a priori" part has really bothered me.  The more and more I read about psychology the more I think that anyone who believes that they could develop an axiomatic system of human action from insights they grasped a priori is totally retarded (mad props to Aristotle though, he worked before the cognitive revolution).

I have suggested that an attraction to libertarianism is in part a function of your personality. Normal people rarely become libertarians, rather, it's a ideology driven by young non-alpha males with Roark/Galt fantasies.  Any survey of the biographies of von Mises or Murray Rothbard emphasizes their stubborn heterodox tendencies; but at this point I just wonder if they were social retards to whom their a priori logic was plausible because they really weren't as complicated as most humans, who engage in habitual and casual hypocrisy and contradiction.

-------------------------------------------------

PhysicistDave

I read "Human Action" in high school (as well as Keynes' "General Theory").

Mises was a Kantian -- Kantians believe in the "synthetic a priori." Simple as that.

Rothbard was not a Kantian, and, in principle, disagreed with Mises on this – i.e., Rothbard thought that economics was ultimately based on some very broad (and very obvious) empirical facts about human nature. In practice, though, Rothbard did note that essentially all economics of all schools was based on “armchair” reasoning – e.g., Keynes did not go out and carefully measure economic variables before writing the “General Theory.”

The great exponent of positivism in economics was Milton Friedman, but in fact he did not really mean it. If your read “Free to Choose,” for example, or his more scholarly works (as I have), he uses empirical facts to illustrate economic theories but not really to test them.

The “a priorist” issue was largely just a matter of noting what economic theorists actually do and of codifying already established academic boundaries. For example, Rothbard’s Ph.D. thesis, “The Panic of 1819” and his later “America’s Great Depression” both included, of course, empirical data. But Rothbard (rightly, I think, in terms of academic terminology) labelled these as “economic history” rather than as pure economics. (A similar point is true of Friedman’s work.)

In a sense the distinction Rothbard was driving at was the same as the distinction I, as a theoretical physicist, make between math and physics. I use math, but I view it as a priori and do not even consider testing the math experimentally. (That of course does not mean I do not sometimes make errors in the math.)

Roughly speaking, as Rothbard and Mises saw it (and as most academics actually practice economics), economics is to economic history or econometrics as math is to physics. It’s more a matter of terminology than anything else.

The one non-terminological point here is that, periodically, some economists who are in fact engaged in clearly a priori thinking insist on claiming that they are empiricist positivists. Some Keynesians did this until it turned out in the ‘70s (stagflation, the shifting Phillips’ curve, etc.) that Keynesiansim empirically did not work. In fact, this was clear a priori to anyone of any intelligence from the beginning – Keynes’ reasoning was wrong in the “General Theory.”

If I make a math error as a physicist, empirical testing is pointless – we’re not really testing the theory if my math derivations from that theory are wrong. The same point applies in economics: since Keynes’ reasoning was wrong (he made errors in dimensional analysis – economists call what he missed the Pigou or real-balance effect), empirical testing was really pointless. While it is nice (although tough for the country!) that Keynes’ predictions were falsified by empirical experience, those predictions did not follow from his premises anyway.

Incidentally, I can falsify your hypothesis about Rothbard. You wrote:  I just wonder if they [Mises and Rothbard] were social retards to whom their a priori logic was plausible because they really weren't as complicated as most humans, who engage in habitual and casual hypocrisy and contradiction.

I knew Murray personally over several years and also had several mutual acquaintances. From my observations (and others’ reports), Murray was, oddly enough, a convivial, wild-and-crazy party animal. He was a big sports fan, loved soap operas, and was fond of early jazz.

In all of those respects, I hasten to add, I am Murray’s opposite – I’m a nerd and proud of it.

He did have an omnivorous curiosity (rather like you) and was, in my observation, fastidiously honest. In those respects, I suppose he could be called nerdy.

Incidentally, his biggest fights within the libertarian movement came over his insistence that empirical reality had to rule. He insisted for example that one had to look at the actual historical reality of American foreign policy, which of course turns out, historically, not to be exactly the model of high-minded altruism that most Americans (and a large number of libertarians) conceive it as being. This, and his empiricist insistence on looking at the concrete details of American historical icons – everyone from Washington to Lincoln to Reagan, as well as at the reality of the political power and shenanigans of Big Business (many libertarians – especially Objectivists – worshipped Big Business; Murray didn’t) antagonized a large number of libertarians.

He was actually the most empirically-oriented economist I’ve ever known – but he did indeed insist to this dying days that all of his empirical stuff was economic history, poli sci, sociology, etc., not real economics. As I said, largely a terminological matter.

Incidentally, I seriously considered majoring in economics rather than physics and actually had an offer to do a post-doc in econ after my Ph.D. in physics, which is why I am acquainted with all this. I can actually get up on the spur of the moment and give a one-hour lecture comparing and contrasting Keynesian, monetarist, and Austrian theories of the business cycle, one of my many useless talents, I fear. (I am not up-to-date on rational-expectations business-cycle theories, however, since I was already out of school when they were developed.)

----------------------------------------

Razib, it won’t surprise you that I’m curious about behavioral economics, game theory, etc. However, I have a “proof-of-the-pudding” attitude here: I’ve seen very little from those areas that were interesting, significant, surprising results (the one exception that comes to mind, which includes both behavioral economics and game theory, is Axelrod’s classic “Tit-for-Tat” study). On the other hand, old-fashioned “armchair” economics has come up with results that are extremely surprising to most people until they learn economics: an increase in the money supply does not increase wealth but only produces price inflation, minimum wage laws destroy jobs, rent controls cause housing shortages, etc. (I know those points may seem obvious to everyone here, but go out on the street and talk to ordinary Americans and you'll find lots of people blissfully ignorant of those basic facts of economics!)

I think the reason for that is that traditional economics does not base itself on psychology at all but rather on extremely “robust” assumptions that are almost certain to hold true independent of details of psychology.

Consider for example the “profit-maximizing” assumption. Taken in a strict monetary sense (it can be expanded into a broader sense of psychological “utility” of course), no one really thinks it is true. Rothbard and Mises were well-aware of this, and, indeed, Rothbard was openly contemptuous of people who thought that “utility” was some measurable quantity that could actually be observed: he stated repeatedly that “utility” was simply short-hand for the fact that people do order their choices, make decisions, etc. (He was also contemptuous of “indifference curves,” etc. – obviously, indifference curves are a fiction useful for giving final exams to undergraduates, and not much else! The “results” you get from indifference curves were already well-known long before anyone heard of indifference curves.)

However, even though many firms do not actually systematically, single-mindedly try to maximize profits, a firm that pays no attention to its profits will go under and will, metaphorically speaking, be removed from the gene pool. In the end, the firms that survive will behave in a way that is pretty close to traditional profit maximization.

Do you know of Ollie Williamson’s work in so-called neo-institutional economics? Fascinating stuff, lots of empirical examples, but, in the end, he applies the same-old traditional armchair logic-of-action analysis used by Rothbard and all the traditional economists. The same can be said of Gary Becker et al.’s extension of economics to everyday life (“Freakonomics” etc.) – one can argue about whether their empirical data has been properly interpreted but to argue about whether or not people make choices, whether those choices are influenced by incentives, etc. would be rather silly.

Incidentally, natural science often works just like Austrian economics (really, all economics). You know the Maxwell velocity distribution for molecules in a gas – Maxwell derived it completely a prior; I worked out a novel (completely a priori of course) derivation myself a few weeks ago. Of course, it has been empirically tested many times, but if a test claimed to disprove it, we physicists would disbelieve the experiment, not Maxwell. The assumptions behind the derivation are so simple, and the math so straightforward and solid, that for a “classical” gas, it is hard for any physicist to see how it could possibly fail to be true. Again, let me emphasize that this was true long before it was tested empirically.

Perhaps even a closer case is natural selection – once you know some very simple facts about Mendelian genetics and sexual reproduction, you just have to have natural selection. The application of natural selection to, say, industrial melanism can be debated in terms of empirical details, but the broad analytic points that we all know about changes in gene frequency vs. selection pressure are essentially a priori.

It is in exactly the same sense that Rothbard thought of economics as “a priori”: to him, the word “economics” referred to the analytical apparatus analogous to the analytical apparatus we all know in modern genetics. In both cases, to refer to that analytic apparatus as “a priori” seems not unreasonable. This of course is the reason Popper once claimed that evolution was not really science, since the underlying principles (once you have Mendelian genetics) are not subject to empirical refutation.

I’m afraid that a lot of people who criticize the Austrians (and in fact all competent economists) for their a priorism are making the same mistake Popper made: they’re failing to look at the full structure of the whole discipline. Indeed, as I said, the same criticism could also be made of huge chunks of physics, such as statistical mechanics.

Empiricism is a false description of science – whether physics, biology, or economics.

-----------------------------------------------

I’ve already mentioned Keynes’ error, usually referred to as ignoring the Pigou/real-balance effect. To technical people, a clearer way of putting it is that he wrongly changed his units without seeing what he was doing – in the middle of his analysis, he switched from real units to monetary units without recognizing this. Had he stuck with one or the other, it would have worked (and he would have gotten results that were quite traditional). It’s the same error that kids make when they switch from feet to inches without multiplying by 12. The error is buried in a complicated discussion in a section of the “General Theory” that almost no one pays much attention to because it seems a pointless sideshow. Unfortunately, it’s not pointless, and it wrecks his whole system (and explains why Keynesianism failed disastrously in the US back in the ‘70s).

Keynes was actually a sort of pomo literary intellectual – a fascinating guy who had some interesting things to say about culture, etc. But his analytical skills were very weak. His literary skills bamboozled many of his fellow economists for quite a while, although some eventually started to escape from the spell. For example, John Hicks, one of the most prominent of the early Keynesians, actually wrote an interesting book in his later years explicitly extolling the Austrian theory of capital, investment, etc.!

The monetarist (basically Friedmanite) theory of the trade cycle is too a priori. It basically studies a make-believe economy in which monetary inflation is distributed uniformly throughout the economy, as via a helicopter that drops money across the country – hence, it is disparaged as the “helicopter model.” Incidentally, the current Fed Chairman is sometimes referred to as “Helicopter Ben” because he has publicly threatened to do just this to stimulate the economy!

Friedman applied this “toy model” to the experience of the US in the Great Depression, and it does in fact explain much of what happened after the ’29 Crash, though not the crash itself.

The Austrian theory, curiously, despite all the brouhaha over “a priorism,” is the most empirical of the three. The Austrians insist on rejecting the naïve helicopter model and following through the actual process by which the Fed injects new money into economy: new money first goes into the capital markets and from there diffuses through the economy at large. Initially, it therefore inflates prices in the capital markets (causing a capital boom) and only later causes price inflation in the economy as a whole.

You can show that the boom in the capital markets will subside once the money has diffused throughout the economy as a whole. The subsiding of the capital-market boom is what initiates the crisis.

This theory, incidentally, was worked out by Mises early in the twentieth century, well before the ’29 Crash, which it so nicely explained. I’ll leave it as an exercise for the student to look back at Fed monetary statistics and see how the theory explains the dot-com bust and the current subprime crisis.

The Fed can prolong the boom (and ultimately the crisis) by increasing its expansionary monetary policies – this seems to be what dear old Helicopter Ben plans to do.

One of the main points that Rothbard made is that the empirical details of how this works out are always different each time because of institutional changes, popular psychology, etc. He was fascinated by those historical, empirical differences, as shown by the fact that he published book-length studies of both the ’29 Crash and the 1819 Panic. Of course, all of us have followed this pop psychology in the news for both the dot-com crash and the current subprime disaster. Interesting, isn’t it?

Rothbard doubted (and I agree with him) that anyone can ever get an adequate theory that could fully explain these varying psychological details in the final playing out of the boom-bust process, but I’m sure he would have been happy to be proved wrong. If anyone thinks you can do a better empirical job than Rothbard, do it! Tell us how the current crisis will play itself out in all its empirical details (before the fact, please, not afterwards).

Good luck!

Anyway, these are the highlights of my talk – all of the details can be filled in by going to fairly obvious sources.

I hope this does make it clear why I consider the Austrian school, at least in its Misesian/Rothbardian form, to be far and away the most “empirical’ school of economics of all and am always bemused by attacks against Rothbard in particular for being too “a prioir.” Rothbard was absolutely obsessed with learning empirical details about not only economic history, but, in my personal experience, even physics.

However, he was indeed unwilling to pursue what my own mentor, Richard Feynman, dubbed “cargo-cult science”: i.e., social science that was filled with complicated math that looked like physics (to anyone ignorant of physics) but had nothing to do with the real world.

I’m afraid that too many people who want “empirical” social science really want “cargo-cult” science.

-------------------------------------------------------

Razib

---------------------------------------------

PhysicistDave

Razib-

I know you’re not a naïve positivist. I’m just trying to make the broad point that whether we are talking about Popper’s erroneous (and later retracted) criticisms of evolutionary theory or criticisms of Rothbard’s “a priorism,” you need to look beyond the soundbite version of methodology to see what is really going on.

Rothbard was actually a very empirical guy, though his research interests happened not to lie in the same direction as Becker's or the behavioral economists'.

Years ago, I actually played around (naturally enough, since I’m a physicist) with putting the stuff in my response to Jon and geek1 into mathematics. You don’t really gain anything, although if I wanted to get it published in mainstream econ journals, it would probably help to impress the editors and reviewers (as Feynman said, “cargo-cult science”). If you know what you are doing, you can make all sorts of cool oscillations occur, for example, but, if you know what you are doing, you also realize you are really putting that in by hand by how you structure the model equations.

Incidentally, the same thing happens even in physics, where some math is of course really necessary. I’ve seen physics papers published in prestigious journals where the math is simply wrong, but complicated enough that the reviewers and editors did not catch it. An old friend of mine, now a tenured professor at one of the country’s top universities, actually released a preprint of this sort years ago – utter nonsense, due to a trivial algebra error. Fortunately, I caught it for him and it was never published. If he had simply described what he was doing in words, instead of fooling around with unnecessary algebra, it would have been clear to him and anyone else that it was nonsense and that he should never have bothered to write the paper.

Based on experiences like this in physics, I have formulated Miller’s law: never use more mathematics in any paper or presentation than is absolutely necessary unless you wish to publicly make a fool of yourself.

Unfortunately, math doesn't save you. I once saw a "ground-breaking" paper (it would have been if it had been right!) published in the top US physics journal in which the authors "proved" their conclusion by dividing by a quantity that is provably equal to zero.

You’re supposed to learn that this is a no-no in first year algebra.

I patiently wrote up a letter to the authors and the editor explaining this.

The authors sent a verbally abusive response back to me; the editors were too dumb or too busy to remember that you cannot divide by zero.

The paper eventually sank into the oblivion it so richly deserved.

Believe it or not, this is actually the norm in academic physics. When I was a doctoral student, my thesis advisor told me not to bother to read the journals.

I do of course agree that in physics and engineering, it is nonetheless nice to try to cross-check yourself however you can, and I do strive for multiple different derivations to check an answer when I’m doing math. And, obviously, in much of physics and engineering, you simply have to use math: verbal arguments are sometimes simply not possible.

But I’d be hard-pressed to think of any place in econ where math helped, and I can think of a lot of places where it hurts. I’d be happy to see a counter-example, since I’m better at math than at words. (Of course, no one denies that you have to use statistics in econometrics, economic history, etc. I’m talking about “pure” theory.)

One of my blind spots in economics is rational-expectations theory: they use a lot of math, and it may actually be needed there. I don’t yet know enough about that area to judge.

----------------------------------------------------

PhysicistDave

I largely agree with Bryan Caplan's comments on Rothbard's "America's Great Depression": to a large degree, they boil down to saying that Bryan (and I) would have written a somewhat different book. Specifically, I, like Bryan, would have given more attention to the effects of sticky wages combined with the contraction of the money supply – this is why, in my little essay above, I said that monetarism explains much of what happened after the Crash. As Bryan said, Rothbard did address this, but I would have given it more prominence than Rothbard did. Indeed, a lot of other issues were relevant too – the disastrous “beggar-my-neighbor” protectionist policies that wrecked the world trading system, cartelization of failing industries, etc. Rothbard did address some of these issues also.

I disagree with Bryan most strongly on the ability of entrepreneurs to predict the future actions of the monetary authority (the Fed) and the consequences of those actions. For one reason or another, anyone who reads the newspapers knows that they just don’t do this.

Of course, why they don’t is an interesting question. Perhaps it has something to do with the institutional structures – golden parachutes for CEOs, federal deposit insurance, etc.

While I’ve been in the role of defending Rothbard here – he was a very bright guy, and I liked him personally – I hope it is clear that I am not saying that he is beyond criticism. I could list a long string of topics on which I disagreed with him, ranging from Giffen goods to the philosophical status of natural law. But that only proves that either Rothbard or I was wrong sometimes, which is hardly news to anyone, given that he and I were both human.

 ------------------------------------------------

 PhysicistDave

tc wrote: I don't know much about Austrian theory, but are you saying that the Fed's actions caused the dotcom bubble? Having lived through it, I saw plenty of irrational herding behavior in myself and the people around me…

Perhaps I’d use the word “enabled” rather than “caused.”

One should never underestimate the power of human irrationality, herd mentality, etc. Social affairs are multi-causal. But if you already have, as the Chairman said, “irrational exuberance,” it does not help for the monetary authority to provide a congenial monetary environment that makes it easier for the capital markets to go nuts.

Your point is the one that I was making and that the Austrian economists have repeatedly made. Every boom is different in its details, psychological features, etc. – the Great Boom of the ‘20s was different from the “Nifty Fifty” of the ‘60s which was different from the dot com boom which differs from the current subprime crisis.

Human psychology and culture are so complex that I doubt that anyone will ever come up with a unifying theory that applies in great quantitative detail to each and every speculative boom.

But there does tend to be a commonality among the booms of an expansionist monetary policy on the part of the monetary authority (for the US, the Fed). While that is certainly not the sole cause of irrational speculation, it does fuel the fever and allows the booms to be sustained longer, and therefore ultimately causes more damage when the inevitable bust occurs.

Incidentally, as long as we have a fiat money system “managed” by a central bank that is ultimately responsible to a democratic political system, this unfortunate “accommodation” of irrational behavior is probably inevitable. Greenspan wrote about this years before he became the Chairman (and people who knew him personally said he continued to acknowledge it privately even while Chairman), but if he had acted according to his convictions, he would have lost his job.

Chairman Al really liked his job.

--------------------------------------------------------

chairmanK writes: Nobody would write a treatise on the sexual preferences of fruit flies without ample experimental data on how flies choose mates and copulate. Economics is just human behavior, and to correctly characterize human behavior, one must analyze real behavioral data collected on real human animals.

I take it you're not a scientist (e.g., a physicist like me)? If you were, you'd know that the naive positivism you espouse has not worked historically in the natural sciences, as Razib and I discussed earlier in this thread.

And, as also was discussed earlier, most of Friedman's theoretical work (e.g., the “quantity theory”) was also essentially a priori, like almost all economists (including Keynes, Ricardo, Smith, etc.). I know of no significant results in the history of economics that were arrived at through the sort of positivistic approach you suggest.

Of course, maybe I missed something -- I'd be happy to hear your examples if you have any.

Every now and them some ignoramus makes the suggestion that mathematicians too should stop their silly a priori theorizing and simply use the experimental method like all sensible people do. This sort of person could not understand a need to prove Fermat's Last Theorem given the huge number of examples that had been shown empirically to satisfy the theorem.

I think it is fair to say that no one who understands mathematics makes such suggestions.

And I also think it is fair to say that no one with a decent understanding of economics would make the suggestion you have made.

But I am open-minded on issues of methodology. Prove me wrong.

I’ve listed earlier on this thread various well-established, interesting results in economics arrived at through the traditional non-positivist methods of economics.

Now, it’s your turn. Tell us of all the wonderful, solid, interesting results arrived at through the positivist approach. you advocate.

Go for it.

Give us your best shot.

I’m actually hoping you come up with something – I like learning new things.

-----------------------------------------------------

Mencius

razib, if there is one book that will make you change your mind: it is Rothbard's Man, Economy, and State. The only way to judge AE is to actually work through it. It is not hard. It is actually much simpler than the mathematical models that have entranced you. And it makes better predictions. (Checked the news today?) Sometimes simple is good.
Let me try to explain the relationship between psychology and praxeology. Razib, chairmanK, both you guys program, n'est ce pas? You are familiar with the word "orthogonal"? In that case, the answer is simple: praxeology is orthogonal to psychology. They do not conflict any more than TCP conflicts with IP.

When Misesians think praxeologically, they abstract over psychology. They try to derive principles which apply to systems of subjectively motivated agents whatever their desires may be. Praxeological conclusions, at least if they are right, are independent of psychology.

Do you think that no such conclusion can be drawn? Here is a simple example. Try and refute it.

A basic principle of praxeology is that an option cannot have negative subjective value. That is, a subjectively motivated agent, whatever its motivation, whether it is a human or a little green alien, cannot rationally prefer a position in which it has more options to a position in which it has less.

If you can go to the mall or play chess, your life is better than if you can only play chess. "But wait," you say. "What about the psychological stress of making a decision? Isn't that a cost? Doesn't it factor into my utility curve?"

(There is no such thing as a utility or indifference curve. The entire trope illustrates the abominable 20C habit of talking about undefinable, immeasurable and unquantifiable concepts as though they were definable, measurable and quantifiable. Hello, cargo cult science. Or more precisely, cargo cult empiricism.)

But actually, Austrians have not missed the psychological burden of an option at all. They merely insist that you include it in the problem statement. If you subjectively prefer to be relieved of the decision of whether to go to the mall or play chess, that relief is a good, and when you set up your problem without it and then later tried to shoehorn it in, you erred.

The fascinating thing about Austrian economics, then, is that it is absolutely independent of human psychology. The Austrian theory of the business cycle may be correct or incorrect - you can check it for yourself, just as you can check any proof. It is not a hypothesis, it is a deductive construct. (I am not endorsing all the conclusions of the Mises-Rothbard school - in fact, I think some of them could use a little updating.)

But if the derivation of the ABCT is correct, it works just as well for the nine-armed octopi in the Large Magellanic Cloud as well as for humans here on earth. If AE in general is correct, monetary policy is the same thing for the octopi. Term transformation will not work any better for them than it works for us. Und so weiter.

From a broader standpoint, I think Popperian science occupies too large a place in your epistemology. "Science" has come to mean two distinct things: any procedure for arriving at highly reliable knowledge, and one specific procedure for arriving at highly reliable knowledge. The latter is constantly threatening to crowd out the former, generating the well-known lamppost effect.

Perhaps if we call the former "reason" and the latter "science," we could say that (Popperian) science is a valid epistemology because reason can deduce - aprioristically - that it should be so. The scientific method itself, for example, would be valid even if no science had ever been done. It is not at all the product of empirical analysis.

In other words, it is not that reason works because reason is scientific. It is that science works because science is reasonable. But this - as Dave points out - doesn't prevent other things from being reasonable. Socrates is not a cat, or even a Kat.

In case it isn't obvious, I should also mention why Austrians prefer to apply praxeology than the scientific method to economics.

The reason is that it is impractical to apply the scientific method to economics, because people are not fruit flies and do not fit in a Petri dish. "Empirical" is a telling word. What it means is "we don't have the budget to perform an actual controlled experiment."

If you could perform significant controlled experiments on human societies, this would be another path to the conclusions that praxeology produces. You cannot. But you can certainly pretend to. You can certainly treat uncontrolled correlations as evidence that some real mechanism exists which matches your model. (Can you say "Phillips curve," boys and girls?) If this isn't what Feynman meant by cargo cult science, well, um, it should have been.

------------------------------------

ChairmanK writes: I study neuroscience. I am accustomed to working with mountains of data and little theory. I believe that I can gain astonishing insights into how animals process information, simply by looking at patterns in the experimental data. I use mathematics to do this, but the mathematics alone do not give me any useful results.

I agree with you that a priori reasoning is an essential part of science. I appreciate your example of the Darwinian theory of natural selection, which I believe must be trivially true, even though the direct experimental evidence is not overwhelming. Likewise, I believe that economists have deep reasons for believing the non-empirical explanatory theories that they invoke. Simple, elegant theories help us to understand real data.

As I have tried to explain above, the correct term for Mises’ approach (and really all economic theorists' historically) would be “deduction from well-established, broadly robust empirical facts.” Unfortunately, that does not have the zing of “a priorism,” but it does capture more accurately what Mises (and all economic theorists) actually did. Rothbard did make this point explicitly to try to correct misconceptions due to Mises’ use of the term “a priorism,” but, as this lengthy thread so well demonstrates, Rothbard did not do an adequate job of getting the word out.

Your own field of neuroscience is a very young science, and therefore it is not surprising that you are dealing with mounds of data.

But the goal of science is indeed “deduction from well-established, broadly robust empirical facts.”

More mature sciences, such as economics and some parts of physics, have actually achieved this.

As I mentioned above, statistical mechanics may come closest to economics in this respect. Given some extremely broad principles – basically conservation of energy and a weak form of time reversal invariance – you can deduce statistical mechanics quite rigorously. (If any mathematician is lurking, yes, I know you also need some form of ergodicidty, but I also know it is hard to avoid ergodicity in any sort of realistic system. I’m using “rigorous” as natural scientists use the term, not mathematicians.)

To the best of my knowledge, no competent physicist has ever though it necessary to empirically check statistical mechanics: we use it all the time, but if some experimentalist claims to have experimentally disproven the Maxwell-Boltzmann distribution, we would just chuckle, pat him on the head, and recommend psychiatric treatment.

We cannot yet do this in all fields of physics, but the goal of physics is to be like economics.

It is indeed true that if you can find a community of humans all of whom would much rather pay higher prices than lower prices for everything that they buy, then that will wreak havoc with the law of supply and demand. But, as long as most people, other things being equal, usually prefer a low price over a high price, the law of supply and demand will stand. And, if an experimental economist claimed to ‘disprove” the law of supply and demand, economists would quite appropriately think he was more than a little incompetent.

To take another example from physics, are you old enough to remember the “cold fusion” brouhaha? When it came out, my cousin, who is a banker, asked me if it was legit. I told him that almost certainly the claimed experimental results were either fraud, or, more likely, incompetence.

Almost all physicists said the same thing before any experimental checking had been done. We were of course correct.

Yes, economics, like any empirical study is not and cannot be strictly “a priori.” But, for practical purposes, like many areas of physics, it is. Again, a more accurate description would be “deduction from well-established, broadly robust empirical facts” rather than “a priorism,” but in practical terms, most people would call my behavior as a physicist in the cold fusion case or in statistical mechanics “a priori,” even though strictly speaking it is “deduction from well-established, broadly robust empirical facts.”

Someday, we can all hope, neuroscience will reach the same level of maturity already attained by economics and some areas of physics.
-------------------------------------------------

PhysicistDave

I think what Feynmann meant by "cargo-cult science" was specifically loading up on mathematical mumbo-jumbo without any actual empirical basis for the math but just to impress people with how “scientific” you were. I know he was especially disdainful of Samuelson’s “Foundations of Economic Analysis,” which he reportedly read cover to cover (unlike some people here who are criticizing Mises without having read “Human Action”!).

I never talked with Feynmann about this in detail, although I did know him well. (However, I actually did hear his original speech about this in person.)

Incidentally, Feynmann had the same attitude towards physics – he thought that one clear concept was worth a hundred pages of math. He had a true gift for seeing through the math to the underlying physics (hence, “Feynmann diagrams”). I aspire to the same, but then I’m no Feynmann.

I do think it is fair to expand the phrase “cargo-cult science” to describe all sorts of silly “physics envy" in the social sciences, most especially when that physics envy exhibits a woeful lack of understanding of the actual procedures of physics itself!

I hope it is clear from all this discussion that I do not deny the possibility that either advanced math or detailed empirical studies may someday produce some important results in economics. I just do not know of any historically. Of course, as I have emphasized, Mises’ foremost student, Rothbard, acknowledged that broad well-established empirical facts were indeed at the foundation of economics, and Rothbard, who after all, unlike most of his critics and most “mathematical” economists, actually had a degree in mathematics, also did use some math in his presentation (e.g., in the classic Austrian textbook, “Man, Economy, and State”).

But Rothbard, perhaps because he actually did know mathematics, also did try to obey what we might call “Feynmann’s principle”: use no more math than necessary.

-----------------------------------------------------------

Mencius

The goal of AE is to explain the recurrent patterns of behavior that we see over and over again in networks of independently motivated, intelligent individuals. The product of this exercise is verbal by definition. If you try to achieve it with the mathematical techniques of Marshall and his heirs, what you get is a string of formulas, plus a press release explaining it in English.

The former is rigorously derived, but its assumptions are extremely hard to corral, collate and check. The latter is (typically) incredibly sloppy and ill-reasoned, and derives its authority from the former in classic papal style.

The Austrians, like the economists of all centuries but the last, prefer to take both in one step. Thus their work is far more open to criticism and debate - the adversarial process that actually makes the "scientific method" work.

Did your training in mathematical economics convince you that maturity transformation is a normal, healthy part of a stable financial system? Try my Austrian-lite explanation of the problem, which I think any intelligent person can read. Though it helps if they understand division.

Probably the most important product of the Austrian edifice is not the ABCT, but simply the conclusion that any quantity of money is adequate - there is no need for the money supply to expand "to meet the needs of trade."

This can be explained simply without any math:

Demand for currency is indirect - it is motivated not by the monetary commodity itself (gold, engraved green paper, Yap stones, etc), but by the goods the commodity can be exchanged for. [This is not completely true under a gold standard, because gold has direct uses, but even under a gold standard most demand for gold is indirect.]

Thus, as Hume pointed out, currency can be redenominated neutrally - adding or subtracting zeroes, or multiplying by any other value - as long as the change is uniform, it affects all those who hold or owe the currency equally. For example, contracts must be redenominated as well.

Thus, we can define all balances of a currency as fractions of the total quantity extant. If that quantity is fixed, the fractional and traditional notations are identical. If that quantity changes uniformly, the traditional representation changes and the fractional representation does not.

In the fractional representation, however, nonuniform creation of new currency is equivalent to a transfer of money from those who did not get the new currency, to those who did. Ie, illegal counterfeiting is theft. Legal monetary creation is taxation.

The punch line is that we have no reason at all to believe that healthy trade and industry depend on involuntary transfers. So we have no reason to believe, with Friedman, that the money supply must expand to match "economic growth."

Of course this observation, which as you can see has nothing to do with psychology, was known not only to Hume, but to all 19C economists. (I believe the first at least in modern times to observe it was Cantillon.) It is also basically obvious. But if today's neoclassical economists understand it, they keep that very, very quiet.

There is a reason for this, which is that the relationship between 20C economics and the 20C state is unusually close. The former sees its primary role as suggesting policy choices to the latter. Thus some friendliness is to be expected.

The monetary models that the likes of Keynes and Fisher produced, which of course were vastly less sophisticated than what PhDs today turn out, were primarily designed to obscure this rather obvious fact. Ie, they allowed state and quasistate institutions to engage in behavior which in the past had been conceived as prima facie evidence of official malfeasance (debasing and coin clipping), and present it as the latest, greatest science, an update of the old "orthodox" fuddy-duddy conventional wisdom.

To hear the products of this tradition describe the gold standard as "crank" economics is sort of like being in the Louvre when a crazy homeless person pulls down his pants, craps in his hand, and rubs it all over the Mona Lisa, while museum security and a roomful of tourists applaud. I don't believe that any further empirical evidence is required to distrust mathematical economics. Your mileage, of course, may vary.

---------------------------------------------------------------