Desperately Evading the Toughest Questions about Intelligent Design: A Review of Dembski's The Design Revolution

By Jeffrey Shallit

Posted April 11, 2004

William Dembski's latest book, The Design Revolution: Answering the Toughest Questions About Intelligent Design consists of 44 questions about intelligent design with short answers by Dembski -- each answer takes about 6 or 7 pages, on average. Critics of Dembski -- such as Mark Perakh -- who were looking forward to having their objections addressed will be disappointed. The Design Revolution is even more intellectually dishonest than I thought possible. The easy questions Dembski actually addresses are answered disingenuously; the really hard questions he avoids entirely. This book should have been titled Desperately Evading the Toughest Questions About Intelligent Design.

So many of Dembski's answers consist of evasion or dissembling that it would take a book as long as Dembski's own to catalogue them all. In this short essay I'll just list a few.

Consider chapter 5, "Religious motivation". Dembski says "let's be clear that design theorists oppose Darwinian theory on strictly scientific grounds". With nearly every prominent intelligent design supporter a conservative Christian, this isn't even remotely plausible. Consider three of the leaders of the ID movement:

• Phillip Johnson converted to fundamentalist Christianity after a mid-life crisis and made "Defeating Darwinism" (part of the title of one of his books) his new mission.
• Jonathan Wells candidly admitted that "Father's [Rev. Sun Myung Moon's] words, my studies, and my prayers convinced me that I should devote my life to destroying Darwinism..."
• Dembski himself, in a March 7, 2004 talk at the Fellowship Baptist Church in Waco, Texas, admitted his own motivation is religious. He said, "I think at a fundamental level, in terms of what drives me in this, is that I think God's glory is being robbed by these naturalistic approaches to biological evolution, creation, the origin of the world, the origin of biological complexity and diversity. When you are attributing the wonder of nature to these mindless material mechanisms, God's glory is getting robbed."

These do not sound like scientific grounds for opposing "Darwinian theory".

In a previous article in this blog, I showed how Dembski has a history of incorrect citation of quotations -- even when previously informed of his errors. Here's another example. On page 52, Dembski writes "Nor does philosopher Daniel Dennett help matters when, in Darwin's Dangerous Idea, he recommends quarantining religious parents who object to their children being taught evolutionary theory."

The only problem is that Dennett wrote no such thing. Dennett actually wrote, on page 519, "...those whose visions dictate that they cannot peacefully coexist with the rest of us we will have to quarantine as best we can, minimizing the pain and damage..." To find the people Dennett is actually referring to, one has to read the quotation in the context of the entire chapter. It is those (listed on pages 516-517) who insist on slavery, child abuse, discrimination, and fatwas on their opponents, all in the name of their religion. Dennett has informed Dembski that he did not say what Dembski attributes to him, and Dembski even sort-of retracted his bogus claim in 2000. Neverthless, it appears again in The Design Revolution with no indication of the controversy. This is the height of intellectual dishonesty.

On page 159, Dembski contrasts the information in one copy of a text with that in two copies. He writes "Nor for that matter do both copies together contain more information than any one copy individually. The mathematics justifying these claims is straightforward." However, using the definition of information most commonly used in computer science -- that is, Kolmogorov complexity -- Dembski's claim is incorrect. In fact, it can easily be proved (and I have assigned it in my undergraduate course on computation) that there exist infinitely many strings x for which there exists more Kolmogorov information in xx than x. Dembski was informed of this in December 2002, but he persists in making this bogus claim.

Any theory of information in which n copies of a text have no more information than a single copy is obviously wrong, because one can clearly store an arbitrary amount of extra information encoded in the number n which specifies the number of copies.

Here's another example of Dembski's disingenuous answers: On page 303, Dembski refers to some unnamed articles in scientific journals. He claims "They are all written by design theorists and are listed in the ISCID bibliography." Unfortunately for anyone who wants to check Dembski's claims, the URL provided is not available to just anyone -- only to paying members of Dembski's own society, the International Society for Complexity, Information, and Design. What has Dembski got to hide? Maybe those papers don't say anything about design at all, but how could we know?

Now let's turn to the questions that Dembski refused to address.

Let's start with the most obvious question: what, exactly, is design? You'd think this would be a basic question Dembski would have to address in a book about "intelligent design" -- and Del Ratzsch, an ID-friendly philosopher at Calvin College has taken the ID movement to task for this omission -- but he doesn't. Does design just mean pattern? Or is it necessarily teleological? Neither this book, nor previous books such as No Free Lunch, address this question.

Here are a few more questions Dembski doesn't address in The Design Revolution:

1. When estimating the probability of events, why do you use two different and incompatible methods, depending on whether the event was human-caused or not?
   As Wesley Elsberry and I have previously pointed out, when Dembski analyzes events that are clearly the result of human agency, such as the Caputo case, he estimates probability by assuming that events are independent and equally likely -- he ignores the known causal history of the events. However, when he analyzes other kinds of events, such as the generation of the protein URF13, his analysis is based on some hypothesized causal history. In one case in No Free Lunch, he even uses two different methods in analyzing the same example!
2. Why have you never acknowledged that a crucial calculation on page 249 of No Free Lunch is off by about 65 orders of magnitude, even though you were informed of this in 2002?
   Dembski has a curious history of ignoring corrections. In a previous article on this blog, I showed how he continued to use a quotation of Schopenhauer, even after he was informed of its very questionable provenance.

   Nearly two years ago, I told Dembski that his calculation on page 297 of No Free Lunch (estimating the centered formula about a third of the way down the page) was wrong. The very same calculation is also used on page 299 to estimate the "perturbation probability for the [sic] bacterial flagellum" -- which is the centerpiece of the book. Now, this calculation isn't rocket science; the formula just involves multiplying and dividing some binomial coefficients and powers. Even an undergraduate should be able to perform it correctly. But Dembski didn't. He said this formula evaluates to approximately 10^-288, whereas it's actually about 10^-223.

   Now it's true that nearly every biologist who has read Dembski's computation of the probability of "the [sic] bacterial flagellum" has rejected it as specious nonsense, because the formula is based on a random assembly scenario that is wildly unrealistic. But in addition to that, Dembski's calculation error makes a flagellum 10⁶⁵ times less likely than the result of his own his bogus scenario.

   Even stranger for someone who has Ph. D. in mathematics, Dembski claims his calculation is based on "Stirling's formula". Now Stirling's formula is a classical approximation for the factorial function, and factorials are one way to evaluate binomial coefficients. Indeed, Stirling's formula would be good for a back-of-the-envelope calculation. But there's no need to rely on approximations for doing Dembski's calculation. Any decent computer algebra system, such as Maple or Mathematica, can compute the exact value of the formula on page 297 -- or for that matter, the exact value of the quotient of sums that the formula is based on -- in a matter of microseconds.

   The point isn't the sloppy calculation. The point is that Dembski was informed of this error nearly two years ago, but he has never acknowledged it or corrected it in any way. Why not? You'd think that an error of 65 orders of magnitude in the centerpiece calculation of the book would merit a correction somewhere. For that matter, why is there no errata page for No Free Lunch?

   Maybe Dembski should take his own advice, given on page 51 of The Design Revolution. There he wrote: "How can a scientist keep from descending into dogmatism? The only way I know is to look oneself squarely in the mirror and continually affirm, I am a fallible human being. I may be wrong. I may be massively wrong. I may be hopelessly and irretrievably wrong--and mean it!" [emphasis in original]

3. Why have you not acknowledged that your mathematical "proof" on pages 152-154 of No Free Lunch that "natural causes cannot generate CSI" is flawed, since (among other errors) it claims it applies to all functions f, but actually it assumes that the function f is known to the intelligent agent in question?
   This "proof" is crucial to Dembski's grandiose "Law of Conservation of Information". But it's not a proof at all. As Elsberry and I showed in our paper, the "proof" contains several errors. Another error is the conflation of "information" with Dembski's CSI and the blithe assertion that the pair (i,f) contains at least as much "information" as j, if f(i) = j. Dembski has known of these criticisms for nearly two years, but he has never addressed them.
4. Why have you never seriously addressed the work of artificial life researchers, who routinely find in their simulations the kinds of novelties you claim are impossible?
   Artificial life researchers such as Karl Sims and Thomas Ray carry out simulations of evolution on computers, using tools such as mutation, natural selection, and recombination. They routinely find the kinds of innovations that Dembski claims are impossible. Sims, for example, in his paper "Evolving 3D Morphology and Behavior by Competition", published in Artificial Life IV, showed how simple 3-D robots could evolve very interesting and novel strategies for locomotion and fighting, even though those strategies were not programmed in by the programmers.

   But Dembski never seriously addresses these findings. In The Design Revolution, the index contains an entry for Thomas Ray, but but when one looks at the indicated page (page 67), there is a discussion of John Ray, but nothing about Thomas Ray. Karl Sims is nowhere to be discussed.

5. Why do you continue to conflate your term "specified complexity" with Davies' use of the term, when Davies is clearly referring to events with high Kolmogorov complexity, whereas you are referring to events with low Kolmogorov complexity?
   This is one of Dembski's rhetorical tricks. He sometimes calls his main concept "specified complexity" and sometimes calls it "complex specified information". He then exploits his own confusing terminology by conflating his own terms of art with well-known uses of the terms "complexity" and "information" when it is useful.

   In particular, Dembski has located other occurrences of the term "specified complexity" in the literature, and then implies that his concept is the same as these other uses. But it is not. In Davies The Fifth Miracle, for example, he makes it clear on page 116 that by "complexity" he means high Kolmogorov complexity (although he attributes it to Chaitin). (What precisely Davies means by "specified" is less clear.) But for Dembski, specified complexity corresponds to low Kolmogorov complexity (see, for example, No Free Lunch, p. 144). Pretending these are synonyms when in fact they are antonyms is deceptive.

The bottom line is that The Design Revolution has little that is new. Much of the content has been lifted, in some cases word-for-word, from Dembski's previous work. His answers to questions are a masterpiece of evasion and misdirection. And he refuses to answer the really tough questions about intelligent design. The folks who endorsed this book -- from Rick Santorum to Robert George to David Berlinski -- should be ashamed of themselves.

Originally posted at The Panda's Thumb.

* * *