Why I'm not a Behe fan: conclusion and a challenge

About 2 months ago, I finished a series on Michael Behe's latest book, The Edge of Evolution. I concluded that it was a terrible book, displaying significant errors of both fact and judgment. The book's main argument is a population genetics argument, and Behe seems to have little knowledge or understanding of that difficult subject. The book is a joke, and I believe it will someday be seen as one of the more disastrous mistakes made by the ID movement. But I think it's important to distinguish between Behe's errors (which reflect on his scientific credibility and on his decision-making habits) and his thesis. His book is full of mistakes, but that doesn't mean that his proposal is known to be false. So I'd like to make it clear what my verdict on his book actually is, then present an outline of one way to actually test Behe's hypothesis.

1. In The Edge of Evolution, Behe correctly identified a biological process – the generation of genetic variants that lead to evolutionary change – as a likely focus of deliberate design. Having concluded that common descent is true, he reasoned that the trajectory of change through the tree of life might be expected to show evidence of non-random direction. Design, as he and others in the ID movement conceive it, might be manifested in the pattern by which the tree of life came to be. (Some might go as far as to say that it must be manifested in such a way, but I don't think Behe suggests this.) My point is that there is nothing stupid, irrational, or unscientific about Behe's reasoning. So, Behe conceived a hypothesis, which I will restate as follows:

• Based on the consideration of life's complexity, specifically on the consideration of the integrated complexity that characterizes the molecular machinery of the cell, it is proposed that random mutation and subsequent selection cannot fully account for the evolutionary development of biological systems.
• Consequently, it is proposed that the process of mutation is non-random.

Again, I find nothing outrageous or stupid about the hypothesis, or even its rationale. Molecular machines are astoundingly complex and integrated, and I do think it's reasonable to wonder how such things can come about without the aid of a superintelligence. In other words, Behe's proposal is not inherently incoherent or otherwise easily dismissed. Might the machinery of life have emerged through non-random processes? Sure. EoE is a joke, but not because the proposal is a joke.

EoE is a joke because Behe seems not to have even attempted to establish the strength of the hypothesis. Very little of the book is devoted to this central concern, and those sections that take up the task are so laughably wrong that they have led me to question Behe's scientific integrity. (Sorry, no apologies: the errors are too basic, and the proposal too world-altering, to give someone who is vying for scientific immortality a pass on standards of scientific conduct.)

But this is important: Behe's failure to even attempt an honest defense of his proposal does not imply that the proposal has been falsified. It hasn't. It remains possible that the development of biological machines – especially in the early days of the tree of life – was characterized by a non-random, directed trajectory. (I happen to doubt this, but that's not relevant here.) Behe's book is a failure, but his hypothesis stands.

So here we are: an interesting and potentially revolutionary hypothesis has been advanced. It has a certain explanatory appeal, and it has unquestioned relevance for believers of many kinds. It is empirical and rational. And, I maintain, it is testable, at least in principle. And so I'm offering to collaborate on a real effort to test it.

2. Behe's proposal leads to certain types of testable predictions. He claims that the genetic changes that underlie certain levels of evolutionary change occurred non-randomly. In other words, he claims that there is a dramatic mismatch between rates of genetic mutation and rates of evolutionary change. His efforts in EoE were ridiculously inadequate. Here is an outline of an approach that could succeed.

• One major mistake that Behe made was to devote most of his attention to a "case study" in which significant genetic change did not occur. His case study was poorly suited to his purpose, but even if it had been better conceived it would be worthless. We can't learn about how evolution works by analyzing examples in which it didn't occur. (Well, of course it did occur in Behe's case study, but the changes that he claims are non-random are different by his own definition.)
• So, any approach to the detection of non-random influences on evolutionary change needs to focus on case studies that actually involve the relevant level of evolutionary change. Examples should be easy to find, by considering the tree of life and the branching levels at which one would hypothesize non-random change.
• The evolutionary lineage(s) selected for analysis should be fairly well-documented, so that the nature of the relevant common ancestors can be reasonably inferred. This probably means that much deeper lineages (such as eukaryotes or even multicellular eukaryotes) would not make good subjects of analysis. Since Behe is pretty sure that design characterizes differences at the level of class (and deeper), this concern is not a barrier to addressing his hypothesis, at least at those levels of divergence. The tetrapod lineage could serve well, but there are any number of evolutionary trajectories that could be considered.
• Within the selected lineage(s), one or more evolutionary changes would be selected for genetic analysis. Changes could be simple (such as the molecular evolution of a particular protein of interest) or more complex (such as the development of a particular attribute like teeth or feathers or lungs), and could even include the sum total of the genetic changes in a lineage, but must be amenable to genetic description. Most importantly, the evolutionary changes that are analyzed must be associated with the specific design postulate. The goal is to examine the genetic changes underlying an evolutionary transition that Behe would identify as designed.
• Once the genetic changes of interest have been identified, analysis can proceed the way Behe pretended to proceed in EoE: inferred mutational trajectories can be considered in the light of estimated mutation rates and estimated generation numbers. If non-random mutation is clearly necessary for the evolutionary changes in question, it should be apparent that even the simplest mutational paths leading to change are well beyond the explanation of random mutation.

My description makes the undertaking sound straightforward, and in principle it is, but of course such examination of even a relatively simple evolutionary change is a significant and demanding project. Inferring the genetic makeup of the common ancestor is a project all by itself, and constructing postulated mutational pathways is the kind of work that occupies many professional biologists full-time. (Consider the work of Joe Thornton and his group, considered among the best analyses of this kind.) Estimates of generation number will span huge ranges even after the most careful consideration of the variables.

But this is the work that any real scientist and scholar would know has to be done. Behe's hypothesis is completely untested, and only the kind of study that I have outlined can change that. I invite any scholar with interest in undertaking this project to contact me. I would be interested in joining a collaborative effort to test the non-random mutation hypothesis, and I have some significant resources that could be brought to bear on the problem. This is a serious offer, and I would encourage readers to forward it to anyone who might be interested in discussing the details.