First Impressions of Dembski's "Intelligent Design"

by Wesley R. Elsberry

I picked up my copy of William A. Dembski's "Intelligent Design: The Bridge Between Science and Theology" at the "Evolution and Providence" conference. Having read Dembski's "The Design Inference" and a number of other essays of Dembski's, my expectations were somewhat different from the actuality.

One might expect that "ID" would present an integrated and high-level approach to what "intelligent design" might mean and signify for both science and religion. While it does broach and discuss many issues, where it seems to fail expectation is in the "integration" part.

Dembski has chosen to collect edited versions of a number of separate essays together rather than produce the contents of this volume from scratch. Only the first chapter, on signs, and the appendix, on criticisms of ID, appear to be new for this book. While this obviously allowed Dembski to produce the volume quickly, it detracts from the reading experience. Various concepts from chapters 5 and 6, for example, are explicated redundantly. Other concepts are referred to with differences in terminology that reflect the evolution of Dembski's ideas, but which when encountered in the confines of one volume simply appear confusing or inconsistent.

The one chapter that I have read in detail so far is chapter 6, which derives from Dembski's presentation at the 1997 "Naturalism, Theism, and the Scientific Enterprise" conference. I was especially on the lookout for changes in Dembski's critique of natural selection. And change there was, but not quite in the way I anticipated. Back in 1997, Dembski attempted to place an upper bound on the rate at which natural selection might incorporate information into a population. Dembski's approach then was to note the difference between fecundity and survival and use the ratio of those numbers to ascribe a maximum number of bits per generation that could be attributed to the action of natural selection. Bill Jefferys, a UT Austin astronomy professor, took Dembski to task over this interpretation at the conference. The amount of information, Jefferys pointed out, was not proscribed by how many offspring typically died in each generation. The information content of an evolutionary novelty was essentially independent of the ratio that Dembski used.

Dembski's reaction to Jefferys' criticism, to his credit, has been to remove that particular problematic approach. However, Dembski has not replaced it with anything, nor revised his discussions and conclusions concerning natural selection. Dembski treated his analysis of information accumulation due to natural selection as if it were an entirely separable and disposable part of his original argument. Since Dembski's argument seeks to eliminate other possible routes by which biological organisms may accumulate information in order to conclude abiotic infusion of information, this leaves rather a large hole in his argument. The discussion of the process of information accumulation via natural selection is not separable when one is applying argument by elimination.

I had hoped that Dembski might expand his analysis of natural selection in this volume, but so far that appears not to be the case. Back in 1997, Dembski promised that we would see his full-blown technical discussion of natural selection in section 6.3 of "The Design Inference". Section 6.3 of TDI includes no such thing. Nor does any other part of TDI. In September 1999, Dembski promised that his in-prinicple refutation of natural selection would appear in a book he was working on called "Redesigning Science". This continues a habit of Dembski's of deferring critical discussion of particular topics by reference to some work to appear in the future. As Dembski notes for coin flips and die rolls, single events do not give us a basis for distinguishing between chance and design. But as Dembski continues to put off critics with references to works that are not at present available, this starts to look less like an occasional happenstance and more like a deliberate strategy.

Dembski's Appendix presents a series of briefly stated criticisms of "intelligent design" and Dembski's rebuttals of each.

The first item is the criticism that "intelligent design" is simply another case of a "god of the gaps" fallacy. Dembski's response is that in certain cases the ability of science proceeding via methodological naturalism to discover explanations not only is limited, but that we can recognize that we have reached that limit. Dembski mentions alchemy and the existence of mythical creatures as instances of cases where naturalistic science not only has reached a limit but can recognize it.

There are at least two problems with Dembski's rebuttal. The first is that Dembski does nothing to demonstrate that "intelligent design" is a particular instance of the general case that he argues from. Even if we accept Dembski's formulation, his argument does not establish that all events without current scientific explanation should be rejected as having possible explanation in that mode. Each instance requires justification, and this is something that Dembski does not do. The second, and I think more serious, problem is that Dembski's characterization of "proscriptive generalizations" appears misleading. For example, Dembski invokes the existence of unicorns as something for which scientists are justified in a "proscriptive generalization" that none such exist based upon the level of effort that has gone into finding and classifying organisms. By Dembski's argument, because scientists have put in enough effort to uncover mammals of the size of a unicorn if any such existed, we can issue the "proscriptive generalization" that no unicorns exist. This seriously misrepresents the basis for saying that the unicorn as conceived in modern art does not exist somewhere waiting for discovery. It is not primarily a matter of effort that has gone into describing mammalian species, but rather the incompatibility of unicorn features with equine characters. (It should be noted that certain breeders have been displaying one-horned goats as "unicorns", and decry the horse-based imagery as a corruption of medieval knowledge.) Similarly, alchemy fails to get funding not because the efforts of medieval alchemists produced nothing, but rather because the basic principles of alchemy make counter-factual predictions and have been falsified. The basis for the "proscriptive generalization" that alchemy does not work is not just a matter of level of effort, as Dembski implies.

The second objection that Dembski addresses is that "design" can explain everything, and thus explains nothing. Dembski's response is that critics using this objection confuse and conflate "intentionality" with "design". This rebuttal boils down to a rather cheap literalism concerning the construction of the objection. Because "design" does not apply to "everything", though "intentionality" might, Dembski says that the critics have hit the wrong target. The problem with such a rebuttal is that it leaves the suitably restated underlying criticism completely untouched. "Design" can be used as an explanation for events where current knowledge fails, without particular independent evidence that "design" actually applies, and thus tells us no more than that current knowledge fails to provide a compelling explanation.

Dembski's discussion of "intentionality" simply seeks to deploy "intentionality" as a ready scapegoat for "design"'s failings. But something that Dembski mentions along the way deserves attention. "Design" implies "intentionality", though the obverse is not true. This is simply left lying. Yet quite a lot of argument goes on in philosophical circles concerning how one can objectively infer "intentionality" in another agent. This discussion can be exceedingly contentious in the field of animal cognition. Dembski has, apparently incidentally, produced a logical framework that claims to be able to produce just such a result, but appears not to wish to promote its use in the community that could best put it to use. We often know that an agent has acted. If we then use Dembski's Explanatory Filter to yield "design" for some event, then we also obtain an argument for a particular intention (the one underlying the specification used) on the part of the agent. This represents a legitimate possible use of Dembski's framework within the scientific community, but we do not see Dembski or other ID proponents attempting to push this technology where it might do some good.