Thinking over the problem from a variety of perspectives, I've arrived at the following point of view:
1) Intelligent design is necessarily committed to the existence of at least one supernatural being.
2) Intelligent design is not necessarily committed the view that this being must be God.
3) Intelligent design does not succeed as an argument for the existence of God.
4) Intelligent design fails to pass muster as a scientific theory.
5) Intelligent design should be mentioned in science classrooms in high schools and colleges.
My argument for (1) is that intelligent design theorists themselves go so far as to define "intelligence" as "that which is irreducible to the combination of chance and necessity." In this approach, basically reading Plato against Monod, they define intelligence as non-natural. Although biological intelligent design (BID) leaves open the purely formal possibility of design by aliens, this only raises the further question as to where those aliens came from. Either they did evolve, in which one can well ask why that hypothesis is not available for terran life as well, or else they were also intelligently designed. By contrast, cosmological intelligent design (CID) explicitly requires that the fine-tuning argument licenses the inference that there must be at least one being which cannot be described by the laws of the physics of this universe, and that sounds close enough to "supernatural" for it to count. (But see my note (1) below.)
My views for (2) and (3) are basically indebted to the insights of Hume and Kant, though I think that Kant sees the problem in a more general and deeper way that Hume does. (Briefly: Hume only provides a criticism of the argument from design when cast in the form of an argument of analogy; Kant provides a criticism of the argument from design in any form.)
My argument for (4) hinges on the importance of testability. Consider these two hypotheses (thanks to John Pierot of Thoughts in a Haystack for the discussion, based on work I've read by Eliot Sober):
H1: There exists a supernatural being which is solely responsible for the natural order and which wants (or would have wanted) everything to be purple.
H2: There exists a supernatural being which is solely responsible for the natural order and which wants (or would have wanted) there to be exactly as many different, and different types, of purple things as there are.
Firstly, notice that both hypotheses are claims about supernatural beings, and both make claims about the relation between supernatural beings and observable phenomena. But there's a crucial difference. If H1 is true, then there is a set of observable phenomena -- "everything being purple" -- which can be shown to be false. So H1 is testable -- that is, it can be tested, and it can be shown to be false according to the test.
But compare that with H2. The observables entailed by H2 are indistinguishable from observations that can be made independent of H2. There are exactly as many different purple things, and as many different types of purple things, as there are -- regardless of the truth of H2. Since the truth or falsity of H2 makes no difference in what can be expected, it is not testable.
My contention, then, is that intelligent design is like H2 -- it claims that there exists some intelligent being which wanted to produce either the universe (CID) or life (BID) exactly as it is. This yields zero explanatory insight, or what is the same thing, it is not testable.
Having said all that, one might reasonably expect me to argue that intelligent design has no place in the science education of our public schools, colleges, and universities. Here I depart from what is basically an entrenched consensus, on the following grounds.
Firstly, "teaching the controversy" -- that is, bringing the full weight of evidence and reasoning to bear -- is an excellent opportunity to teach students how to think scientifically and not merely master a body of scientific knowledge. Unfortunately, science educators are not themselves taught in how to teach epistemology. The easy solution is to ignore the problem. The hard solution, and the better one, is to regard the "teach the controversy" movement as an excellent opportunity for the NCSE (National Center for Science Education) to create a program on how to teach science educators how to teach epistemology and philosophy of science for high school and college students. It's a public secret that we are not teaching students how to think scientifically, we're not teaching students how to appreciate and enjoy what science can and cannot do, and the "teach the controversy" strategy is an opportunity to change that.
Secondly, consider the argument raised by Crispin Sartwell in favor of teaching intelligent design in schools (here or here). In brief, he argues that while intelligent design (or, for that matter, creationism -- though he doesn't mention creationism by name) is anti-naturalistic, and for that matter, anti-scientific, that doesn't mean that it should be excluded from science education. And I agree.
Consider, Sartwell asks us, someone who is deeply religious, and who believes, as part of her faith, that neo-Darwinian explanations are incompatible with that faith. That person is going to be very upset if her child is subjected to nothing but the neo-Darwinian point of view in a school that is supported by her school taxes. But more interestingly, Sartwell points out that it's completely reasonable for her to be upset. It would be puzzling if she weren't upset, given what she believes! Now, does her voice deserve to be heard in the classroom?
I think that the answer is "yes" -- "yes," even though intelligent design is not a scientific theory. That's because science education has be about more than instilling in young minds all the latest scientific theories. It has to be about training people about how to be scientifically informed citizens in a pluralistic and (supposedly) democratic society. And for that reason, anti-naturalistic -- even anti-scientific -- voices should not be excluded. To exclude those voices from the classroom -- or, to put it more pointedly, to silence those voices within the classroom -- is tantamount to isolating science from its social and cultural and historical and political context. And doing that is a failure of science education.
Having said that, I would take issue with Sartwell in one serious respect. On his view, the naturalistic world-view, which is (for him and for me) the scientific world-view (but see note (2) below), is the stance of "reason" -- in contrast with that of "faith." For one thing, I'm not happy with any simplistic, Enlightenment-era contrast of "reason" and "faith." For another, I'm not happy with the identification of science and reason.
If we take "reason" here in a broadly Sellarsian sense -- in the famous phrase, taking part in "the game of giving and asking for reasons" -- then it seems clear to me that anyone who deliberately excludes him or herself from that game has positioned him or herself outside the range of views that are available for dialogue in the public sphere. So if one takes up an anti-scientific position, and in doing so puts oneself outside of the space of reasons altogether, then no, I don't think that position is entitled to a public voice. (I say that even though the division between public and private is itself highly contested.) And I take it as fairly obvious that someone such as P.Z. Myers (Pharyngula) holds precisely that view: that scientific methods and rational inquiries are simply co-extensive. (But see note (3) below)
Therefore, in order to maintain that anti-naturalistic/anti-scientific voices deserve to be heard, I conclude that one can position oneself outside of science without thereby positioning oneself outside of "reason" altogether, and that means that the identity between "science" and "reason" should be rejected.
(1) However, one might consider the multiverse hypothesis. Suppose there are infinitely many possible universes. Each universe is defined by a set of values of different physical constants. Possibly, a fully developed physics could explain the nature of universes as such. Call this a general physics and the physical laws for each universe are defined by a particular physics. Then a supernatural agent, in keeping with the tradition of theological speculation, would be an agent that cannot be defined in terms of a general physics.
(2) Of course, the sharp contrast should be noticed between my view and that of the advocates of intelligent design. On their view, the identification between the domain of scientifically explicable phenomena and the domain of natural phenomena is precisely what ought to be rejected!
(3) I'm not entirely sure if Myers can afford to be as committed to materialism as a metaphysical position as he advertises himself as being, for the simple reason that it has proven to be exceedingly difficult, if not impossible, to adequately account for mathematics and for ethics in materialist, or physicalist, terms. Myers is best understood, I contend, as an Enlightenment rationalist whose basic commitments are to science, to liberalism, and to secularism. Whether this is ultimately a compelling view, or whether the work of Nietzsche, Adorno and Horkheimer, Foucault, Wittgenstein, and Rorty shows how deeply problematic this view is, is a matter I shall address in a subsequent posting.