20 March 2016

Some Proper Philosophy of Science

Noah Smith and Jason Smith have been writing a series of blog posts focusing on three main ideas: Empiricism, paradigm shifts, and 'negative empiricism' (which is, as far as I'm concerned, equivalent to Postpositivism, which would basically be synonymous with 'whatever Karl Popper said,' but whatever).

Regarding empiricism, I'm not sure why this is getting any attention from people discussing philosophy of science in the 21st century. Empiricism is most closely associated with David Hume -- i.e. an enlightenment philosopher -- and, while it serves as the backdrop for newer philosophy of science, is neither new nor worthy of attention when considering a new philosophy of economics. It seems that people have confused empiricism with caring about empirical evidence when in reality empiricism is an epistemology that elevates experience over a priori deductions (which are characteristic of rationalism). For a more in-depth look into this, see "Rationalism vs. Empiricism" in the Stanford Encyclopedia of Philosophy. Needless to say empiricism alone does not qualify as a philosophy of science.

Jason somewhat repairs this by arguing for "negative empiricism" which he defines as "theory rejection." Here my only issue is that Jason has almost literally argued for Postpositivism, which is exactly that: hypothesis rejection. Basically think Popper; what really matters for a theory is falsifiability. Interestingly, Popper essentially solves the empiricist (Humean) problem of induction by arguing that empirical evidence in favor of a hypothesis is irrelevant and falsification is all that matters. In this sense, Jason is basically a Postpositivist by arguing for "theory rejection" based on empirical evidence. Noah Smith also weighs in, suggesting that economics was in "an age of great theoryderp" before the IT revolution which (according to him, not me) allowed for theories to be invalidated. To his credit, Noah acknowledges that "you can be a pure theorist and still subscribe to empiricism - you just don't believe in theories until they've been successfully tested against data," which is, in my opinion, the way most of economics currently is and will remain forever.

Both Smith's talk about paradigm shifts (further solidifying my suspicion that neither of them are empiricists and that they did not just come forward in time from 1748), which are attributed to Thomas Kuhn (another 20th century philosopher of science). In Kuhn's view, science operates in a single paradigm until the usefulness of said paradigm is exhausted at which point there will be a revolution. Both of the Smith's seem to be arguing for such a regime change, whether it is Jason's Information Transfer Framework or Noah suggesting that economics is experiencing a wave of "new empiricism" (OK, every time I read the word empiricism in this topic I shudder, please stop calling it that). Either way, each of them sees economics as being on the verge of or currently undergoing a revolution which begs the question of whether or not the current paradigm is really exhausted. I'll leave this to others to determine, but I personally don't see the field as being in a dead-end right now.

So, really, this post all comes down to two requests: please stop calling Postpositivism Empiricism and please stop trying to get economists to accept an 18th century epistemology instead of actually arguing for a new philosophy of economics (or, at least, coming clean and saying you are a Postpositivist).


  1. Jason claims that there's a difference between his view and that of Popper though: he says a strict Popperian would view Newtonian Mechanics as having been falsified, whereas in his view it's still useful at certain scales, and thus doesn't deserve to be called "falsified."

    If Jason is accurate about Popper's views vs those he's laid out, I think he has a point there: I view science in general as producing a series of improving approximations to reality, each approximation (in the case of physics say) best suited to a different scale or region of validity. This is how engineers view the world. The idea of tossing out an approximation because in a region it overlaps with another approximation, its error is larger seems silly (unless the approximation with the smaller error in that overlap region has characteristics to recommend it over both regions).

    You might be interested in what CalTech physicist Sean M. Carroll had to say about the concept of falsifiability. I was initially surprised by his take. I'd like him to elaborate on that a bit. I think the key might be the phrase "directly observable." For example, Carroll has argued for the many worlds interpretation of quantum mechanics because (he says) it's the simplest: it requires no wave function collapse mechanism. Wave functions don't collapse in that view (called Everettian), instead the universe splits. So he's using Occam's razor in a surprising way there: arguing that we should just accept what the wave equation tells us and not add complication to it... even though the consequence is there may be "many worlds." However, some of the competing interpretations (most of which add extra dynamics to the wave equation to explain a collapse associated with measurements) are, in principle, testable so it's conceivable we could test their validity some day. He argues that Everett should be the null hypothesis because it doesn't add anything. This is FAR from my area of expertise, so I'm just summarizing what I've heard him say on various occasions.

    But you may view all that as far from the concerns of social scientists. I'm not sure.

    Sean describes himself as a "philosophy friendly" physicist (and he sometimes answers emails).

    1. I thought this link was broken, but Sean does elaborate a bit there. Also I notice at the bottom of that original link I gave you (to preposterous universe), there was somebody there named "Danny Hillis" arguing that cause and effect should be retired (I suppose) (for some reason it takes Edge.org forever for my browser to load, so I can't tell you what Hillis says yet... maybe in an hour).

    2. Re: many worlds: you might ask why we don't test for those if that's what the Everettian view predicts... well, it's because according to that view those different worlds can't interact from the time the split occurs going forward (from what I understand).

    3. I would say that Jason is somewhere between being an actual postpositivist and a pragmatist; he won't say that Newtonian Mechanics is falsified because it is still useful [read pragmatic], but, in the grand scheme of things, he's pretty much a postpositivist.