Glenn Branch's picture

Falsifiability Dembski-Style, Part 1

Physical model of a bacterial flagellum

“Is intelligent design falsifiable? Is Darwinism falsifiable? Yes to the first question, no to the second question.” You might think that it’s Michael Behe again; after all, he’s on the record as having asserted, “ID is quite susceptible to falsification” while “Darwinism seems quite impervious to falsification.” (He said so in his “The Modern Intelligent Design Hypothesis” [2001]; I argued that he was mistaken in “Falsifia-behe-lity,” and noted in “Falsifia-behe-lity on the Edge” that he seems to have quietly abandoned the claim by the time of his second book, The Edge of Evolution [2007].) But no. Instead, it’s William A. Dembski, the chief theoretical architect of the “intelligent design” movement, in his No Free Lunch (2002). You might think that, since Dembski, unlike Behe, is trained in philosophy, that his argument will be clearer, crisper, and perhaps even not obviously misguided. But, again, no.

After offering what seems like a straightforward position, as quoted above, Dembski immediately backpedals, writing, “I am being a bit fast and loose in my use of falsifiability here,” and adding, “I am using the term ‘falsifiable’ not just in the strict sense where claims get eliminated because they are demonstrated to be false but also in the loose sense where claims get eliminated because they lack warrant or are superfluous.” The reference to “warrant” here is puzzling: warrant is a notion from epistemology that applies to beliefs, while falsifiability is supposed to be a semantic notion applying to propositions (including theories): including warrant in the definition of falsifiability will require revising it to apply to beliefs (or otherwise complicating the discussion immensely). In a very similar passage on falsifiability in Dembski’s The Design Revolution (2004), there is no mention of warrant, so I’m inclined to ignore it.

Beyond discarding the mention of warrant, The Design Revolution also improves on No Free Lunch in introducing the term “refutability” for “falsifiability...in the loose sense”: “Theories are refuted when, for whatever reason, they get beaten down and rejected—not because they are demonstrably false.” That’s vaguely expressed, and there’s no formal definition given, but the idea is apparently that a theory is refuted if and only if it’s either falsified or shown to be superfluous; it’s refutable if and only if it’s either falsifiable or showable to be superfluous. “Intelligent design” in particular is refutable not because it is falsifiable but because it is showable to be superfluous. Given a naturalistic explanation of the bacterial flagellum (seen above), intelligent design would be dispensed with “on the general grounds that one does not invoke intelligent causes when undirected natural causes will do. In that case Occam’s razor would finish off intelligent design quite nicely.”

Grant, for the sake of argument, that “intelligent design” is refutable, as Dembski claims. So what? Dembski suggests that refutability is consonant with Popper’s ambitions for falsifiability: “The main point of Popper’s criterion of falsifiability is not so much that scientific claims must have the possibility of being demonstrated false as that they must have the possibility of being eliminated as the result of new evidence...that is the point of refutability.” Now, Popper was interested in falsifiability because he thought that it provided a criterion for demarcating science from non-science. According to Popper, as the Stanford Encyclopedia of Philosophy summarizes,

if a theory is incompatible with possible empirical observations it is scientific; conversely, a theory which is compatible with all such observations, either because, as in the case of Marxism, it has been modified solely to accommodate such observations, or because, as in the case of psychoanalytic theories, it is consistent with all possible observations, is unscientific.

Consequently, in revising falsifiability to refutability, the effect of Dembski’s proposal is to suggest that “intelligent design” is scientific. (In The Design Revolution, he discusses falsifiability and refutability under the general head of testability, commenting, unexceptionably, “For a theory to be immune to evidence from nature is a sure sign that we are not dealing with a scientific theory.”) Since “intelligent design” is so often accused of not constituting a scientific theory, often on the grounds that it is unfalsifiable, it’s unsurprising that Dembski wants to provide a framework according to which it is a scientific theory. But refutability isn’t going to do the trick.

Why? Well, consider the charge that refutability misclassifies the examples that Popper cited as unfalsifiable and thus non-scientific theories: astrology, the Marxist theory of history, Freud’s and Adler’s psychoanalytic theories, and the like. For example, assume for the sake of argument that Popper accurately described Freud’s and Adler’s psychoanalytic theories as not determinately offering any predictions at all. In that case, Occam’s razor could be invoked, without any ado, to dispense with the entities—id, ego, superego, inferiority complex, and so on—distinctively posited by those theories, which would themselves be dismissed as superfluous. So are those theories refutable and thus scientific after all? That would seem to show that refutability misclassifies Popper’s examples, and can’t be taken as a good criterion for scientific status; if so, then the fact that “intelligent design” is refutable isn’t going to be of any interest.

Dembski might reply that he specified that refutability involves “the possibility of being eliminated as the result of new evidence”; that there’s no new evidence that underlies the judgment of superfluity for Popper’s examples; and that those theories are therefore not refutable. But then the challenge is to demonstrate that “intelligent design” isn’t in the same boat. Dembski asserts that “intelligent design” would be rendered superfluous by the provision of a naturalistic explanation of the bacterial flagellum and is therefore refutable, but he never argues that it isn’t of itself superfluous considered on its own—in the same way that Popper’s examples are—without reference to any rival explanation. Showing that “intelligent design” is falsifiable would suffice, but he already abandoned that project. It’s not obvious that there is any non-question-begging way for him to save the position otherwise. As for “Darwinism,” it will have to wait for part 2.