Bill Dembski, one of the leading proponents of modern intelligent design “theory,” is an interesting animal. But before turning to him, I should explain why I am telling readers of a freethought magazine about the latest attempt by religiously inspired intellectuals to advance their pseudo-scientific agenda. Surely few readers of Free InquIry would give much credence to whatever Dembski or his colleagues at the Discovery Institute have to say. And therein lies the problem. Too often I have observed that skeptics and freethinkers assume certain positions (such as discounting anti-evolution writings) without knowing the intricacies of the arguments involved. Some of my evolutionary biologist colleagues have made the same mistake, setting out to debate a creationist without any preparation because “surely he is a mistaken fool and it will take just a few minutes to show the audience the error of his ways.”
Actually, it takes more than a few minutes. Indeed, anti-science writers are getting more sophisticated in their arguments, which requires correspondingly increased sophistication on the part of skeptics and freethinkers to understand why those arguments are fallacious. After all, the very difference between our position and the other side is that we strive to consider matters rationally and to do the hard work of refuting rampant nonsense, while most of what they do is to keep producing that nonsense at a fast enough rate to keep their followers from thinking at all.
That said, let us briefly examine one of the latest from Dembski’s magic hat of tricks: his idea that the so-called no free lunch theorem (NFL) is a death blow to the theory of evolution by natural selection. Dembski is apparently so convinced of this that he has written a whole book on the topic, a creationist bestseller hailed as the intelligent design movement’s equivalent of Darwin’s Origin of Species (notice the implicit recognition of the intellectual stature of the latter).
The NFL is a legitimate, if esoteric, bit of novel mathematical theory elaborated by David Wolpert and William Macready. It says that the average performance of any algorithm over the class of all optimization problems is no better than blind search. Let me unpack the last sentence so that we can see what it means and why Dembski thinks—erroneously—that it refutes Darwinism.
Optimization problems are mathematical problems (with plenty of practical applications) where one is trying to maximize (or minimize) a certain quan-tity given certain constraints on the factors affecting the situation of interest. One famous example is the so-called traveling salesman problem, in which one has to determine the shortest route connecting a series of cities. For a long time, the goal of researchers in optimization theory was to find an algorithm that would do better than blind search regardless of the particular problem at hand. Wolpert and Macready showed that this, alas, is not possible.
Why should evolutionary biologists care, one might ask? Dembski thinks they should because evolution by natural selection can be thought of as an “algorithm” to optimize a particular quantity, namely the organism’s fitness. In Dembski’s reading of the NFL, Wolpert and Macready have shown that natural selection cannot possibly produce adaptation. If this is correct, then the Holy Grail of creationism—a mathematical demonstration of the impossibility of Darwin’s theory—has been found!
If you smell a trick somewhere, you are completely right, but it is important to understand exactly where the reasoning goes bad. First, notice that the NFL concerns the average performance of algorithms over all possible optimization problems. It turns out that most of these problems are characterized by chaotic mathematics (in the technical term of chaos theory), which is the main reason behind the conclusion of the NFL: when the mathematics is mostly chaotic (i.e., the behavior of the system is unpredictable from one moment to the next) then a search algorithm obviously will not be able to do better than a random stab. The NFL does not say that a specific algorithm is not better than chance when applied to a specific problem, as Dembski erroneously implies. This mistake on his part is enough to refute the contention that the NFL demonstrates the impossibility of natural selection. But wait, there’s more!
It turns out that natural selection is not even an optimizing process at all, which means that evolution does not fall under the category of problems covered by the NFL to begin with. Natural selection does not increase the fitness of organisms over time: if this were so, then currently living animals would be much more “fit” than, say, those that lived hundreds of millions of years ago. But there is no meaningful sense in which mammals are more fit than dinosaurs; each species is suffi-ciently (not optimally) well adapted to its own environment, a rather modest outcome of evolution. Furthermore, it turns out that the very process of evolution changes the environment itself, which in turn changes the nature of selection in a continuous feedback loop that happens to violate one of the conditions of the NFL, an additional reason why the latter does nothing to refute evolutionary theory.
In the end, one has to conclude that Bill Dembski either: (a) does not understand the no free lunch theorem; (b) does not understand the theory of evolution; or, (c) is in bad faith when he claims that the former refutes the latter. Of course, a combination of the above is also possible, and I will leave the reader to sort that out.
Massimo Pigliucci is associate professor of ecology and evolutionary biology at the University of Tennes see and author of Denying Evolution: Creationism, Scientism and the Nature of Science.