New Edition of Dembski's Design Inference

Last year, I published a book called The Failures of Mathematical Anti-Evolutionism with Cambridge University Press. As the title suggests, I analyze all of the major mathematical arguments put forth by creationists and intelligent design (ID) proponents and find them all to be very poor. Inevitably, I devote a fair amount of space to William Dembski’s arguments.

So I naturally raised an eyebrow upon hearing that Dembski has now released a revised and expanded edition of his 1998 book The Design Inference, now coauthored with Winston Ewert. I wondered if he would have anything worthwhile to say in response to my book, and it was gratifying to see that my name shows up numerous times in the index. I will look forward to seeing what he has to say, once I have acquired a copy of the book. However, Dembski has posted an excerpt from the book’s Introduction, and I was delighted to see that, perhaps inadvertently, he has simply conceded two of the main points I raised against his ideas.

Recall that Dembski’s catchphrase is “complex, specified information.” Roughly, this means that if something has a low probability of arising by chance and instantiates an independently describable pattern, then intelligent design is implicated at some point in its causal history. In my book, I argued that, to the extent that he intends to apply this to biological structures, he has no hope of establishing either complexity or specificity. We have nothing like the information we need to carry out a meaningful probability calculation, and we have no background information relevant to distinguishing design-suggesting patterns from those we simply impose on nature (like seeing a dragon in a fluffy, cumulus cloud).

More specifically, I noted that when Dembski calculates the probability of something like a flagellum, he explicitly assumes that an “irreducibly complex” (IC) system cannot evolve by gradual means. This means his argument is completely parasitic on Michael Behe’s claims about IC systems: If those claims are correct, then they are all by themselves a strong argument against evolution. A probability calculation adds nothing. But since those claims are completely wrong, no calculation based on them will have any meaning. For specification, I noted that Dembski presents an elaborate mathematical apparatus, complete with notation, jargon, and Greek letters, which is meant to allow us to distinguish real patterns from artificial ones. But then he never uses that machinery for anything, but instead just says that biological structures are obviously specified by their function.

With that set-up, here is Dembski from his Introduction:

Tacitly in the first edition of The Design Inference and explicitly in its sequel, No Free Lunch, I argued that natural selection and random variation could not create the sort of complexity we see in living things. My approach in applying the design inference to biology was to piggyback on the work of design biologists Douglas Axe and Michael Behe. They had identified certain subcellular systems . . . that proved highly resistant to Darwinian explanations.

Precisely my point! “Piggyback” is a more polite term than “parasitic,” but my term is more accurate. It is the claims made by Behe and Axe that are doing all the work in the argument that evolution as currently understood is seriously deficient. Dembski’s calculations are irrelevant, in addition to being incorrect.

Continuing from the last quote, Dembski writes:

Our joint task was to put plausible numbers to these systems so that even factoring in Darwinian natural selection, the probability of those systems arising was exceedingly small. Note that the specification of these systems, as in their exhibiting the right sort of a pattern for a design inference, was never in question. The issue was always whether the probabilities were small enough.

Dembski’s calculations do not, in even the slightest way, “factor in” Darwinian natural selection. Instead, they use certain claims made by Behe and Axe to justify ignoring natural selection altogether. Once again, it is those claims, and not Dembski’s pointless calculations, that are doing all the work.

Moreover, note how dismissive he is of establishing “specificity” in his idiosyncratic sense. This, again, was precisely my point. Dembski and his supporters tell the world he has a mathematically rigorous way of distinguishing real patterns from artificial ones. But that mathematical machinery plays no role at all in his actual argument that the flagellum, say, is specified. Apparently, to specify a flagellum it is sufficient to say, “Kinda looks like an outboard motor.” Nothing rigorous about it. And since biologists argue that natural selection crafts functional systems as a matter of course, I would say the question of specificity is very much in question.

If this excerpt is representative of the rest of the book, then I do not think his critics have anything to worry about.

Let me finish with an extended analogy. Suppose someone tells you they drew five cards from a shuffled deck and that all were the same suit. He asks you to work out the probability of that happening. No problem, that is a basic calculation. But then he gives you a plot twist. He tells you that his deck has millions of cards and dozens of suits, and that he will not be any more specific than that. At this point you laugh and say you need more information to find the probability, thinking that you had just stated the obvious.

But now an ID proponent charges into the room. He says, “Of course you can work out the probability. Just assume there are exactly one million cards and forty suits, with the same number of cards in each suit.” He then scribbles down elaborate calculations involving factorials and binomial coefficients.

Astonished that anyone would seriously put forth such a thing, you reply that his calculation is silly since it is based on entirely arbitrary assumptions. But he just ignores you and says, “And not only is this improbable, but it’s also specified! I mean, seriously, five cards of the same suit. What an obvious pattern!”

At this point you are face-palming and rolling your eyes. You point out that since we know nothing about the distribution of suits in this deck, we have no idea if it is remarkable to get five cards of the same suit. You say, “For all we know one suit is so overrepresented in the deck that getting five of the same suit is close to inevitable.” Again, you are surprised you need to point this out.

But the ID proponent is undaunted. He shouts, “You’re just a dogmatic, anti-religious tool of the scientific mafia.” Then he runs out of the room and writes books telling the world about how he has revolutionized mathematics, biology, and philosophy. His sycophants start calling him the Isaac Newton of information theory. Meanwhile, you are left wondering what just happened.

End scene. I think that captures the frustration scientists feel when addressing ID proponents. Except that there is one point where my analogy does not go far enough: In the analogy, the assumptions underlying the calculation were merely unjustified. In Dembski’s writing, the assumptions are plainly false.

Dembski packs an astonishing amount of error into such a short excerpt, and there is much more to comment on than what I have presented here. But we will save that for another day.