For someone who doesn’t care much for experimental studies and data analysis, I often come up with experimental tasks or statistical surveys for other people to do. Here is one: Take a bunch of graduate students of nuclear physics and ask them about the role of falsifiability in science. Then take a bunch of graduate students of macro-economics and ask them the same question. My guess is that the physicists will not know what you are talking about, and at any rate wouldn’t have come accross this issue during their studies, whereas the economists will know all about Popper, logical positivism, demarcation and other latin words I never heard of.
I thought about this comparison when I watched this diavblog. As usual with bloggingheads I lost interest quickly, but it’s worth watching for a while, if only to keep track of the fluctuations in the frequency of David Levine’s blinks while the other head is talking. Best Levine’s quote from the part I watched — `You will have to fill me on which these three axioms are’, responding to the assertion that `nothing is more true than that the three axioms of rational choice have all been falsified’. I actually replayed this part to retrieve the quote, and I still managed to forget the three axioms.
Anyway, back to my (testable !) prediction. Suppose it turns out to be correct. Then, since I am the one that made the prediction, Popper tells us that the fact that it was verified renders my explanation for it more credible. And the explanation is that physicists are robins and economists are penguins.
Let me elaborate. Richard Feyman famously said that philosophy of science is as useful to scientists as ornithology is to birds. It’s an awesome observation, and I would like to apply it specifically to the part of philosophy of science that deals with the demarcation problem, or the distinction between science and non-science. And I’ll start with the physicists, because our whole idea of science is essentially `the stuff that physicists do’. Popper was trying to pin down what are the special characteristics of the physics enterprise. Thus, the fact that physics is science is almost a tautology. Every child knows it as every child knows that robin is a bird. That’s why Popper’s (or anybody’s) answer the demarcation problem is useless to physicists.
Things are different with penguins. A layman might not recognize them as birds. In order to prove that they qualify, penguins have to appeal to some standards that describe what are the special properties of birds that distinguish them from humanities mammals. That’s why penguins care about ornithology. And that’s why economists care about falsifiability.
Rosenberg is right it’s a bit ironic that economists adopt falsifiability as a litmus test for science, since philosophers of science have generally rejected it. But the irony I think is on the philosophers, not the economists. How would the ornitologists feel if the birds that actually read their papers would only be interested in what was written eighty years ago ?
5 comments
October 23, 2009 at 1:20 am
Theoretical Economics as Math « Algorithmic Game Theory
[…] Economics as Math Eran Shmaya, in a beautiful recent blog post, adds some sociological angles to the economists’ never-ending introspection of whether their […]
October 23, 2009 at 8:41 am
michael webster
Feynman was doing philosophy of science, just badly.
I looked at the divablog – Rosenberg is just wrong. Long before Tversky/Kahneman people were giving good counter examples to the axioms of cardinal utility. Allais and Ellsberg to name two. Allais was there from the beginning. A lot of work was done to relax the axioms to account for this behavior and yet still have something identifiable as an expected utility model.
Recent behavioral economics is less rigorous as the underlying axiom/axioms are not specified. Some of the work is interesting, some of it is crap.
But in preference theory, there was always a strong interest in behavior.
October 26, 2009 at 2:02 pm
Anonymous
>> `the stuff that physicists do’
Except for the debate about string theory and its uselessness due its inability to be pinned down to any specific prediction in an empirical sense (or its inability, with one single exception, to make empirical predictions that are novel)… so, is it science (with some empirical content) or just abstract mathematics?
November 14, 2009 at 4:59 pm
noam neer
string theory may not have predictions, but it still can be a useful part of science (as I see science) if it gives consistent unification of all other theories. of course, even simpler physical theories are yet to become mathematically consistent.
another option is that string theory won’t unite quantum theory with general relativity, but it will be able to explain the dozens (hundreds?) of constants in the standard model of elementary particles.
December 27, 2010 at 5:32 pm
Robert Johnson
I’m an engineer, so I can’t help but approach the philosophy of science from the narrow vantage point of usefulness. Most knowledge is useful, in a very practical sense. Knowledge about physics is useful because it is used to produce the transformative technology that has created the modern world. Knowledge about the human body and about pharmaceuticals is useful because it can be used to produce effective surgery and medicine.
The usefulness comes from the ability to make accurate predictions. Technology is simply using accurate predictions to produce desired effects. E.g. if I can predict how a lever will behave, I can make one to help me do work that I want to do.
So what can be said about ‘scientific’ propositions that don’t produce predictions? Either they don’t represent any new knowledge (highly likely), or they represent new knowledge that we don’t know how to make use of (less likely, but still sometimes the case).
From that line of reasoning I’m inclined to go on to guess that string theory and big chunks of macro econ probably don’t represent much real knowledge.