I pointed out to Dembski that evolutionary biologists expect to find complexity in living things, so when he applies his mathematical litmus test and declares that living things have complexity, he's not exactly sending evolutionary biologists into a tailspin. Yes, he said, but his "explanatory filter" not only distinguishes between complexity and randomness; it distinguishes between "specified complexity" (the kind that he says would result from intelligent design) and "cumulative complexity" (the kind that might result from the mere accretion of adaptations via natural selection). And he has discovered that animals have the former!

Fine, I said, but what he calls "specified complexity"—and asserts is not producible via natural selection—is exactly the kind of complexity that evolutionary biologists tell us they do expect natural selection to produce. But they're wrong, he said: They can't successfully explain how natural selection could produce certain complex adaptations.

Of course, I said, it's true that they could conceivably be wrong about this. But the only way to find out is to argue it out on a case-by-case basis—argue about eyeballs, blood clotting, and other complex adaptations. Dembski can't prove they're wrong just by asserting that they're wrong. Yet his "explanatory filter" argument, so far as I can tell, is just a mathematical assertion that they're wrong.

The more I argued that his logic was circular, the more the two of us seemed to just be going around in circles. So we moved on to another dimension of his position. He concedes that natural selection could be the "conduit" for the creation of specified complexity. But it could only serve this purpose if it was "front-loaded" with certain kinds of "informational input." Indeed, he said, it is conceivable that once you had one-celled life, all the requisite information had already been fed in (though he doubts it would all reside in the cell's genome).

What is the nature of this "informational input"? Here things got pretty sketchy (though, clearly, this is where the hand of God would enter the picture). But let me now throw out one broad sense in which I think Dembski is right—a sense so broad that, I think, any open-minded evolutionary biologist should agree that he's right.

A few self-replicating one-celled organisms, by themselves, cannot get natural selection started. They have to have something else—an environment. And the environment has to have certain features, such as available energy and various physical laws. What's more, for evolution to have produced the particular forms of life we see on this planet, all kinds of other, more specific things have to have been true of the environment (solar energy, water, etc.).

Now, you can call all these features of the environment "information" if you want. In fact, the eminent geneticist Theodosius Dobzhansky once said that natural selection is a process that transmits information from environment to genome. (That is, the genes natural selection preserves, by being adaptively compatible with the environment, amount, in a sense, to an implicit description of it.)

Let me put the point more abstractly. Natural selection is an algorithm. (Even arch-anti-creationist Daniel Dennett calls it that.) And no algorithm can do anything more than take one kind of information and turn it into another kind. In that sense, the information that was eventually transmuted into today's biosphere had to exist in some form prior to today—and some of the information, clearly, was "outside" of natural selection itself in the sense of being in the environment.

So have Dembski and I found common ground? Um, no. He says that the kind of front-loaded information I'm talking about is not the kind he's talking about.

But I don't see why—for theological purposes—it should really matter. If you believe (as I do) that natural selection was likely to produce intelligent beings, and you believe that natural selection is basically an algorithm, or conduit, that has transformed pre-existing information into these intelligent beings (in the concrete sense described above), then isn't the hand of some kind of God an equally plausible conjecture regardless of what the exact nature of the front-loaded information is? Anyway, that's my view.

By the way, Dembski will present his argument in a book called No Free Lunch, to be published this summer.