Brother, Can you Paradigm?

I just read an interview with my friend Tim O'Reilly in which he approvingly cited Thomas Kuhn's "The Structure of Scientific Revolutions". There are some books so bad, but so plausible and influential, that periodically trashing them in public is almost an obligation. The really classic stinkeroos of this kind, like Karl Marx's Das Kapital, exert a weird kind of seduction on otherwise intelligent people long after their factual basis has been completely exploded.

Yes, Kuhn's magnum opus is one of these. When I was a bright and naive young sprat, full of zeal to correct my fuddy-duddy elders, I loved Kuhn's book. Then I reread it, and did some thinking and fact-checking, and discovered that it is both (a) deeply wrong, and (b) dishonest.

First, deeply wrong. Kuhn's basic model that paradigm changes are generational — you have to wait for the old dinosaurs to die — is dramatically falsified by the history of early 20th-century physics. Despite well-publicized exceptions like Einstein's refusal to accept "spooky action at a distance", the record shows us that a generation of physicists handled not one but two major paradigm shifts in their lifetimes — relativity and quantum mechanics — quite smoothly indeed.

Later in the 20th century, the paradigm shift produced by the discovery of DNA and the neo-Darwinian synthesis of evolutionary theory didn't require the old guard to die off before it was accepted, either. More recently, the discovery of things like reverse transcriptase and "jumping genes", which broke two of the central dogmas of genetics, were absorbed with barely a ripple.

I found many other examples once I started looking. It turns out that the kind of story Kuhn wants to tell is quite rare in the hard sciences. There are a few examples of paradigm shifts that fit his model — my personal favorite is Wegener and the continental-drift hypothesis — but they are the exception rather than the rule. Most theoretical upheavals, even most very radical ones, happen rather smoothly.

The soft sciences are a somewhat different story, and the reasons for this are revealing. Look at the post-Freudian upheaval in psychology or the clashes between social contructivism and the evolutionary-psych crowd and you will see something much more like a Kuhnian shock going on (I suspect we've got another one coming in linguistics when Noam Chomsky kicks off). But these fields are vulnerable largely to the extent that they are not science — that is, when the dominant model is poorly confirmed or untestable, and holds largely for reasons of politics and/or the dominance of a single charismatic personality.

One of the most pointed criticisms of Kuhn is that his book is a sort of soft-science imperialism, an attempt to project onto the hard sciences the kind of incoherence, confusion, and political ax-grinding we see in (say) sociology or "political science". In doing so, it does real science a profound disservice.

The dishonesty in the book is that Kuhn evades the question of whether paradigm shifts are an emic or etic phenomenon. In fact, he does this so neatly that it's possible to read the whole thing and not notice that the largest central question about the nature of paradigm shifts is being dodged. Do they change the world or just our description of it? Kuhn hints at a radical sort of subjectivism without ever acknowledging what that would actually mean.

Kuhn got me interested in the cultural history of science when I read this book around 1971. But the more I studied it, the more I became convinced that Kuhn's thesis is simple, appealing, and wrong. Among many other flaws, he erects a binary distinction between "normal" science and paradigm-shattering earthquakes that is not really sustainable except through a kind of selective hindsight. It plays to our human tendency to want to make heroic narratives out of history, but it misrepresents science as it is actually practised and perceived by the people who do it.

(For a demolition of Kuhn that focuses less on the factual holes in his thesis and more on the historical and logical flaws, see this New Criterion article.)
posted by Eric at 3:32 PM