Dembski's mathematical achievements

Two previous entries on this blog by John Lynch have discussed the scientific output (or lack thereof) of two intelligent design superstars, Jonathan Wells and Michael Behe. Despite claims that both of these ID supporters are actively engaged in research, Lynch documents that they have published little or no scientific research in the last six years. Now let's look at the record of another one of ID's superstars, William Dembski.

The principal review journal in mathematics is Mathematical Reviews and its online version, called MathSciNet. Both are projects of the American Mathematical Society. The description of MathSciNet states that it is "a comprehensive database covering the world's mathematical literature since 1940." And it really is comprehensive: about 70,000 new reviews are added each year.

I searched MathSciNet for Dembski's publications. Exactly four are listed: a paper called "Uniform probability" that was published in the Journal of Theoretical Probability in 1990; a survey article called "Randomness by design" that appeared in the philosophical journal Noûs in 1991; his 1998 Cambridge University Press book The Design Inference, and his 2002 book No Free Lunch. That's it.

Dembski's CV lists one other scientific publication (a 1990 article in the Journal of Statistical Computation and Simulation that was not reviewed by Mathematical Reviews). It also lists a preprint entitled "Random Predicate Logic I" that Dembski posted in 2002 (although it was apparently written in 1990).

To understand how sparse this output is, you need to know that the average research mathematician publishes something like 1-2 research papers each year. Mathematicians at small colleges typically publish less because they have more teaching duties, while those with postdoctoral positions or research positions typically publish more. Dembski received his Ph. D. in mathematics in 1988. By this time, a typical university mathematician would have published something like 15-30 papers in the peer-reviewed mathematical literature; Dembski has published two. (I do not count the paper in Noûs since that journal is a philosophy journal and the paper has no original mathematical research in it.)

Of course, the number of published papers is not the only measure of scientific output. A good researcher could publish a small number of papers with large impact. So it is worthwhile to see how often Dembski's papers have been cited in the scientific literature. I used the ISI Web of Science (previously called Science Citation Index) to see how often Dembski's work was cited. His 1991 Noûs article has been cited five times (once by Beckwith in the Harvard Journal of Law and Public Policy and four other citations, including one in Paleobiology, but none in mathematics journals); his 1990 Journal of Theoretical Probability article has been cited twice (once again by Beckwith and once by L. Olsen in the Mathematical Proceedings of the Cambridge Philosophical Society); his 1990 article in Journal of Statistical Computation and Simulation has been cited three times (once again by Beckwith, once by Sober, and once by Barbour -- none in mathematics journals). Since important papers often receive dozens or even hundreds of citations, this suggests that Dembski's work has not been very influential among practicing scientists or mathematicians.

Of course, Dembski has been very busy with other projects. His CV lists many articles and reviews in religious periodicals such as First Things and Princeton Theological Review. I do not criticize his activities in other areas; after all, he is a man of many interests. It is the inflation of his mathematical and scientific credentials that I find inappropriate.

For example, Dembski is frequently touted as an expert on information theory; his colleague Rob Koons has called him "the Isaac Newton of information theory". But how many research papers has Dembski published on information theory? According to MathSciNet, none. (By contrast, Aaron D. Wyner, an expert in information theory who died in 1997, has 64 entries in MathSciNet stretching over 40 years, for an average of 1.6 entries per year.)

Dembski himself states in this interview in Christianity Today that he "became something of an expert in the study of randomness". But how many original research papers has Dembski published on randomness? According to MathSciNet, none (or one, if you count the survey in the philosophy journal Noûs). By contrast, Avi Wigderson, a colleague of mine who really is an expert in randomness, has 103 entries in MathSciNet (of course, not all of those are specifically about randomness).

Dembski's supporters will no doubt argue that his books represent original research. But what do mathematicians have to say about his work? David Wolpert, one of the inventors of the "No Free Lunch" theorems that inspired the title of Dembski's 2002 book, wrote a rather uncomplimentary review for Mathematical Reviews, saying that his work "is written in jello". I have criticized Dembski's mathematics here and here and here, but Dembski has never found the time to reply.

Dembski may be a fine theologian; I don't have the expertise to judge. But it is clear that hyped accolades such as "the Isaac Newton of information theory" have yet to be earned. In terms of mathematical output, Dembski is far below the median. ID advocacy appears to be an excellent way to derail a promising scientific or mathematical career.