Jason Rosenhouse’s Whoppers: More Guidance on Reading Rosenhouse

1 Background

Jason Rosenhouse wrote a book of over 90,000 word for Cambridge University Press, which was released in May 2022 and titled The Failures of Mathematical Anti-Evolutionism. I wrote an 18,000-word review here on this blog that appeared in June 2022. It was simultaneously posted, in a serialized form, on Discovery Institute’s Evolution News & Views blog. I thought it was a pretty good review. A bit long, to be sure, but Rosenhouse got so many things wrong that it seemed worth spending the space to set things right.

I’ve reviewed my share of books over the years, in scholarly as well as more popular forums. I used to have an open door at First Things when Fr. Richard John Neuhaus ran it. I also had ready access to Books & Culture when John Wilson ran it. The very first review Wilson asked me to write, on Mark Steiner’s The Applicability of Mathematics as a Philosophical Problem, received the Evangelical Press Association first place award for 1999 in the category “Critical Reviews.” You can see several such reviews listed here. Go to Amazon, and you’ll find my review of Erik Larson’s The Myth of Artificial Intelligence, which to date remains the review voted most helpful to readers.

So when I saw that in early July of 2022 Rosenhouse had written a reply to my review, I decided to hold off reading or responding to it. I knew what to expect, and I knew if I read it, I would be spending more time responding to him. I was largely satisfied with what I had written. I had spent enough time reviewing the book. And I had other work that I needed to get back to. Moreover, Brian Miller responded to Rosenhouse’s reply at EvolutionNews.org, convincingly refuting it, at least from what I could tell.

2 No Concession Policy of Darwinists

But I wanted to see the reply for myself and finally had some time to do so. Rosenhouse’s reply weighs in at over 6,000 words, so at least the word count is coming down. With my response here, it comes down even further, though not by much. Although I’ve been doing intelligent design professionally now for three decades, I still experience surprise at how Darwinian critics of intelligent design remain obsessively committed to a no-concession policy. Intelligent design, it would seem, can never produce even one valid criticism of Darwinism. But surely, no theory is as good as they make it out to be. Surely, intelligent design proponents must have some valid points to make against natural selection. Alas, no.

So much in Rosenhouse’s book is careless, sloppy, giving no indication that he has carefully studied and adequately comprehended my work or that of my colleagues. No matter. Darwinism can do no wrong and intelligent design can do no right. Thus, in predictable Darwinian fashion, Rosenhouse turns the tables, insisting that my review is 100 percent in error: “It is a person [= yours truly] of rare talent who can write at such length without getting anything right.” Concede nothing is the Darwinist policy. I’ve seen it with Ken Miller. I’ve seen it with Eugenie Scott. I’ve seen it with many other Darwinists. And it’s on robust display in Rosenhouse’s book as well as in his reply to my review.

This attitude by Darwinists calls to mind for me the famous scene in the 1960s film The Guide for the Married Man, in which would-be philanderers are instructed to DENY, DENY, DENY when the least suspicion is raised that they are cheating on their spouse. Here’s the scene, with Joey Bishop standing in for the Darwinists:

3 Appealing to the Unwashed Middle

Before leaving academia for business, I used to lecture on intelligent design at colleges and universities, and often debate people on the Darwinian side. Michael Shermer and Michael Ruse were my most frequent debate partners. My philosophy at these debates was not to try to convince Darwinists that my views were correct. Nor was I particularly concerned about the intelligent design proponents—if they were proponents of ID, they had presumably put their necks on the chopping block and knew what was at stake, academically and culturally, in taking the side of ID. My challenge, rather, in these debates, was to win the unwashed middle—those who had not made up their minds—those who didn’t reside in the cloud cuckoo land of Darwinism. So this response is mainly directed at them.

Rosenhouse’s book is objectively bad. It purports to be a critique of mathematics as used by ID proponents and of my mathematical work in particular. Yet it betrays a lack of comprehension throughout. It makes a virtue of misrepresentation. Its aim is not to understand but to kill. In my review, I called Rosenhouse on his many failures in the book. It’s clear in his reply that he simply ignored the points I was able to score—points he made it easy for me to score because he did such a hack job. Read his book and read my review, and decide for yourself.

His reply, however, adds a new dimension to the debate. The reply, too, is objectively bad in the same sense as his book. But it adds a level of delusion that in reading it made my jaw drop. I’m not writing this for rhetorical effect. In the reply, he lets loose with two whoppers that make me question what planet he’s been living on. Indeed, I have to seriously wonder the degree to which Darwinists are in their right minds if they find in Rosenhouse a voice that speaks for them.

But before getting to the two whoppers, buried in his reply are two substantive points worth addressing. They came up in my review, received comment in the reply, and deserve some additional comment here. They concern (1) the connection between irreducible and specified complexity and (2) the role of the environment in supplying information to the Darwinian process.

4 Irreducible versus Specified Complexity

For Rosenhouse, irreducible and specified complexity are both bogus notions in that they purport to place some limitation on Darwinian natural selection when, of course, no such limitation can ever apply because of natural selection’s presumed wonder-working ability to build novel adaptations and transform one species into another. Yet, for Rosenhouse, specified complexity is the more bogus notion of the two.

Michael Behe, in introducing the concept of Irreducible complexity, used it to argue that irreducibly complex systems could not evolve gradually. But once it’s known, or assumed, that these systems could not evolve gradually, what’s the point in doing a probability calculation, as is inherent in specified complexity, to show that these systems are unlikely to evolve? If something can’t happen or is assumed to be unable to happen, then it is improbable. What further need for a calculation? As Rosenhouse puts it in his reply, “It is really Michael Behe’s claims about irreducible complexity that are doing all the work. The probability calculations do nothing to strengthen the argument.”

I addressed this point in my review, but let’s have another go at it. Consider Sisyphus. As long as you can remember, he’s been rolling a rock up a hill, only to have it roll back down before it gets to the very top, which, let’s assume, is a stable equilibrium, so if he gets it to the very top, it will stay there (though he never does). What is the probability that Sisyphus will get the rock up to the very top? As a historical or inductive probability, it is quite low. All your life, you have been seeing him try to get the rock up there and somehow it never quite gets there.

That historical probability for Sisyphus is the same type of probability as inherent in Mike Behe’s assessment of Darwinian processes being unable to build irreducibly complex molecular machines. All the attempts by biologists to trace a detailed Darwinian pathway of how an irreducibly complex system might emerge from an evolutionary precursor performing a different function have failed.

Richard Lenski, for instance, has run tens of thousands of generations of E. coli, and produced no novel irreducibly complex system. The record of failure of evolutionary biologists in their inability to provide detailed Darwinian pathways for irreducibly complex systems is as complete as Sisyphus’s efforts to get the rock to the top of the hill. If you disagree, please provide an irreducibly complex system, its precursor system performing a different primary function, and then the step-by-step path of how to get from one to the other. Silence? Crickets?

By contrast, specified complexity gets at the nuts and bolts of the probabilistic hurdles that render an evolutionary transition intractable. To continue with the Sisyphus analogy, specified complexity would look not at Sisyphus’s record of failure so much as the types of obstacles he faces in getting to the top and how those might render getting to the top improbable.

For instance, perhaps in rolling the rock up the hill, most of the path is clear and unproblematic, but at one point there’s a bump so that given his strength he just can’t get over the bump. Or perhaps, there are multiple bumps, where he’s got a positive probability of getting over each bump, but when all these probabilities get combined, he’s bound not to get over all the bumps. Or perhaps he gets tired, running out of steam, as he moves up the hill, so that bumps lower on the hill would be no problem, but by the time he gets up the hill, they do become a problem, and his probability of getting over all of them approaches zero.

The point to appreciate is that such a probability analysis of Sisyphus adds to our understanding of his failure. His record of failure is enough to justify assigning a low historical probability to his being able to roll the rock to the very top of the hill. But an empirically based probability of his failure needs to look at the particularities of the probabilistic hurdles that he’s facing. The same holds for irreducible complexity. There’s a long record of failure by biologists to explain how these systems might evolve. Specified complexity attempts to understand the probabilistic particulars that could explain the record of failure.

But specified complexity is not merely a supplement to irreducible complexity. Not all biological systems are irreducibly complex. In consequence, specified complexity can assess the evolvability of biological systems that are not irreducibly complex. For instance, the beta-lactamase enzymatic system that Doug Axe examined (described at greater length in my review) is not in any clear sense irreducibly complex, but it is analyzable probabilistically and exhibits specified complexity.

One more analogy to try to nail all this down. Again, I write for the unwashed middle and have no expectation of assuaging Rosenhouse. Consider a bridge. It’s stood for 100 years, faced all kinds of weather and hardship, and has remained imperturbable. And yet one day it suddenly collapses. Before its collapse, we might think that its probability of continuing to stand was quite high, and so the probability of collapse was quite low. Given its collapse, is it therefore safe to say that a highly improbable event happened?

Those versed in the use specified complexity as a tool for disentangling the probabilities underlying various systems would say that such historical probabilities are of little interest now that the bridge has collapsed. Rather, we need engineers to examine the wreckage to see if there were any tell-tale signs of weaknesses in the bridge that would increase its probability of collapse. The probabilities in this case would be empirical and structural rather than historical. Specified complexity substitutes actionable empirically and structurally based probabilities for historical probabilities.

5 The Environment as a Source of Information

In Rosenhouse’s book, he claims that “natural selection serves as a conduit for transmitting environmental information into the genomes of organisms.” (p. 215) I addressed this claim briefly in my review, indicating that conservation of information shows it to be incomplete and inadequate, but essentially I referred him to technical work by me and colleagues on the topic. In his reply, he remains, as always, unpersuaded. So let me here give another go at explaining the role of the environment as a source of information for Darwinian evolution. As throughout this response, I’m addressing the unwashed middle.

Darwinian evolution depends on selection, variation, and replication working within an environment. How selection, variation, and replication play out, however, depends on the particulars of the environment. Take a simple example, one that Rosenhouse finds deeply convincing and emblematic for biological evolution, namely, Richard Dawkins’ famous METHINKS IT IS LIKE A WEASEL simulation (pp. 192–194 of Rosenhouse’s book). Dawkins imagines an environment consisting of sequences of 28 letters and spaces, random variations of those letters, and a fitness function that rewards sequences to the degree that they are close to (i.e., share letters with ) the target sequence METHINKS IT IS LIKE A WEASEL.

So what’s the problem? The problem is not with the letter sequences, their randomization, or even the activity of a fitness function in guiding such an evolutionary process, but the very choice of fitness function. Why did the environment happen to fixate on METHINKS IT IS LIKE A WEASEL and make evolution drive toward that sequence? Why not a totally random sequence? The whole point of this example is to suggest that evolution can produce something design-like (a meaningful phrase, in this case, from Shakespeare’s Hamlet) without the need for actual design. But most fitness functions would evolve toward random sequences of letters and spaces. So what’s the difference maker in the choice of fitness? If you will, what selects the fitness function that then selects for fitness in the evolutionary process? Well, leaving aside some sort of interventional design (and not all design needs to be interventional), it’s got to be the environment.

But that’s the problem. What renders one environment an interesting source of evolutionary change given selection, variation, and replication but others uninteresting? Most environments, in fact, don’t lead to any interesting form of evolution. Consider Sol Spiegelman’s work on the evolution of polynucleotides in a replicase environment. One thing that makes real world biological evolution interesting, assuming it actually happens, is that it increases information in the items that are undergoing evolution. Yet Spiegelman demonstrated that even with selection, variation, and replication in play, information steadily decreased over the course of his experiment. Brian Goodwin, in his summary of Spiegelman’s work, highlights this point (How the Leopard Changed Its Spots, pp. 35–36):

In a classic experiment, Spiegelman in 1967 showed what happens to a molecular replicating system in a test tube, without any cellular organization around it. The replicating molecules (the nucleic acid templates) require an energy source, building blocks (i.e., nucleotide bases), and an enzyme to help the polymerization process that is involved in self-copying of the templates. Then away it goes, making more copies of the specific nucleotide sequences that define the initial templates. But the interesting result was that these initial templates did not stay the same; they were not accurately copied. They got shorter and shorter until they reached the minimal size compatible with the sequence retaining self-copying properties. And as they got shorter, the copying process went faster. So what happened with natural selection in a test tube: the shorter templates that copied themselves faster became more numerous, while the larger ones were gradually eliminated. This looks like Darwinian evolution in a test tube. But the interesting result was that this evolution went one way: toward greater simplicity.

At issue here is a simple and yet profound point of logic that continually seems to elude Darwinists as they are urged to come to terms with how it can be that the environment is able to bring about the information that leads to any interesting form of evolution. And just to be clear, what makes evolution interesting it that it purports to build all the nifty biological systems that we see around us. But most forms of evolution, whether in a biology lab or on a computer mainframe, build nothing interesting.

The logical point at issue here is one the philosopher John Stuart Mill described back in the nineteenth century. He called it the “method of difference” and laid it out in his System of Logic. According to this method, to discover which of a set of circumstances is responsible for an
observed difference in outcomes requires identifying a circumstance that is present when the outcome occurs and absent when it doesn’t occur. An immediate corollary of this method is that common circumstances cannot explain a difference in outcomes

So if selection, variation, and replication operating within an environment can produce wildly different types of evolution (information increasing, information decreasing, interesting, uninteresting, engineering like, organismic like, etc.), then something else besides these factors needs to be in play. Conservation of information says that the difference maker is information built into the environment.

In any case, the method of difference shows that such information cannot be reducible to Darwinian processes, which is to say, to selection, variation, and replication (because these are common to all forms of Darwinian evolution). Darwinists, needless to say, don’t like that conclusion. But they are nonetheless stuck with it. The logic is airtight and it means that their theory is fundamentally incomplete. For more on this, see my article with Bob Marks titled “Life’s Conservation Law” (especially section 8).

6 Whopper Number One: Seeing Patterns in Biology Is Like Seeing Dragons in the Clouds

With these points of clarification aside, let’s now turn to Rosenhouse’s whoppers. Whopper number one is this: According to Rosenhouse, to see the bacterial flagellum as a bidirectional motor-driven propeller is to see a pattern that in no essential way differs from seeing clouds shaped like a dragon. I’m not making this up. Thus in his reply he writes: “How can we be confident that in using function as a specification we are not doing the equivalent of looking at a fluffy, cumulus cloud and seeing a dragon?” As if to leave no doubt, he repeats the point: “Saying of a flagellum that it resembles an outboard motor is comparable to saying of a cloud that it resembles a dragon.” Seriously?!

Since the flagellum gets so overused in the debate between ID and Darwinism, let’s change the system. Consider the leaf hopper, Issus coleoptratus, which uses toothed mechanical gears to achieve its amazing jumping ability. Here’s a brief video about its gear-based jumping system. Note that these really are mechanical gears. They are not merely like gears. They are not a metaphor for gears. They don’t meekly aspire to be gears. They’re gears.

Yet, according to Rosenhouse, all such systems, despite exhibiting clear engineering patterns, require no fundamentally different type of explanation from clouds that resemble dragons. Natural selection produces the one, natural weather the other. Here’s one example of a dragon in the clouds (a fire-breathing dragon, no less). The internet is filled with such examples.

If you can’t appreciate the difference between these two types of patterns, and if you take seriously that they don’t require completely different orders of explanation, then the real world may not be your home. My notion of specification grapples with the difference. You may not like my formulation of it. Fine. But if you can’t form and take seriously some such distinction, then you inhabit an alternate reality. And even though Darwinian group think, powerful as it is, can keep you imagining that you are in your right mind, your really need to do a sanity check.

7 Digression: The Shermer Delusion

How can some Darwinists, like Rosenhouse, dismiss engineering-like patterns in biological systems as having no more force than seeing human and animal images in clouds? It seems that what’s behind this is the addling effect of evolutionary theorizing on human psychology. As indicated above, I’ve debated professional skeptic Michael Shermer before various college audiences. Shermer dismisses the use of patterns to eliminate chance by appealing to evolution, claiming that evolution has hardwired us to see patterns that convince us of design even when design is in fact absent. In other words, evolution biases us to err on the side of false positives, seeing design in patterns where design is not actually present. Here is how he makes such an argument (in his 2006 book Why Darwin Matters, pp. 38–39):

Perceiving the world as well designed and thus the product of a designer … may be the product of a brain adapted to finding patterns in nature. We are pattern-seeking as well as pattern-finding animals… Finding patterns in nature may have an evolutionary explanation: There is a survival payoff for finding order instead of chaos in the world, and being able to separate threats (to fight or flee) from comforts (to embrace or eat, among other things), which enabled our ancestors to survive and reproduce. We are the descendants of the most successful pattern-seeking members of our species. In other words, we were designed by evolution to perceive design.

Shermer is here suggesting that the patterns we find in nature (as opposed to artefacts created by humans or aliens) are simply imposed by us on nature on account of our evolutionary conditioning, and thus signify nothing about any real underlying design of the world. But that can’t be right. Clearly, nature could present us with patterns that reliably point to the activity of an intelligent agent outside of nature. Take, for instance, a pulsating star that acts as an oracle, communicating messages, in English, about matters of great consequence to the inhabitants of the earth. Assume the star is millions of light years away, and yet is communicating in real time, responsive to questions as they are posed by humans on earth.

Clearly, signals like this coming from outer space would exhibit patterns that leave no doubt about their intelligent origin. But the intelligence in these signals would also be beyond this world, given that physics limits communication by the speed of light, precluding communication taking place, as here, instantaneously. Such signals would provide decisive confirmation of SETI, the Search for Extraterrestrial Intelligence. But it would not be an intelligence localized in space and time that’s sending signals with the help of advanced technology. In this example, the intelligence doing the communicating makes itself evident in nature (via a design inference) but operates beyond the constraints of nature.  

In any case, to refuse to attribute patterns to design on the grounds that our brains are hardwired by evolution to overinterpret design merely begs the question. That’s because some patterns are indeed rightly interpreted as signaling design. We all recognize a valid distinction between patterns that convincingly demonstrate design and patterns that result from unreliable psychological factors, such as an overactive imagination: seeing human faces in the clouds, burnt toast, or soap films, examples of the sort that Shermer—and Rosenhouse—use to discredit intelligent design. Shermer’s claim that we are hardwired by evolution to be pattern-seeking, pattern-finding animals does nothing to draw this distinction.

8 Whopper Number Two: Probability Theory is Irrelevant

The other mega whopper in Rosenhouse’s reply is the claim that probabilities can never be used to assess (and thus potentially to question) how and whether Darwinian evolution can bring about novel biological forms or adaptations. He writes: “We have no probability estimates for the evolution of these systems [such as the eye or flagellum]. That is because probability theory is fundamentally the wrong tool for this particular job. That’s the whole point! Not only can you not rigorously calculate the probability of evolving a particular complex system, you cannot even estimate it in any reasonable way.”

And how does Rosenhouse justify such a massive claim? By citing a single sentence from the Harvard biologist Michael Nowak: “You cannot calculate the probability that an eye came about. We don’t have the information to make this calculation.” Now it’s certainly true that if we were at an early point in the history of life, when no eyes existed, we would be in no position to say how likely eyes would be to evolve by natural selection or, for that matter, how likely it would be for a designer capable of producing eyes in fact to produce them.

But it’s quite another thing to look at existing eyes and ask how probable it would be for them to evolve from various precursors. Such questions may be difficult, especially given the vast complexity of the eye, which is why design theorists like Doug Axe focus on simpler, more tractable systems. But to suggest that probability is inapplicable to such questions of biological complexity is absurd.

But the problem is worse for Rosenhouse, much worse. All physical processes can be modeled by deterministic or stochastic processes. In fact, stochastic processes subsume the deterministic case by simply collapsing all probabilities to zero and one. The problem, then, is that if you deny that probabilities apply to a physical process (and that includes biological processes such as Darwinian evolution), you’ve abjured science—you no longer have a scientific theory. You are then simply spinning fairytales, imagining that what happens is what you want to have happen, but with no tether to reality.

I made this point at the end of a 2014 talk on conservation of information that I gave at the University of Chicago. My former advisor Leo Kadanoff convened that talk, and at the end he backed me up by saying that the ball was in the Darwinists’ court if they wanted to contend that Darwinian processes can avoid the constraints of conservation of information and probability. The talk was recorded and you can watch it here:

9 An Appeal to Sanity

I’ll close with a story. My wife used to set up psychiatric units across the US. She would travel to various states, stay there a few months, set up a unit, hire people, get it self-sustaining, and move on to the next place (she worked for a company based out of Dallas). While overseeing a unit in Arkansas, it was her responsibility also to get people admitted to the unit. The highlight of her time there was an older woman who would sit on the porch of her house and fire her shotgun at local police. It seems she wasn’t aiming to hurt them, but eventually it got to be a bit much.

So my wife was tasked with bringing her before a judge to determine whether she was a danger to herself and others, with the intention then of having her committed to my wife’s psych unit. For some time during the proceeding, this woman seemed lucid, to the point that the judge was looking at my wife as if to say “Why did you bring her here in the first place?” She was articulate. She described her house, her living arrangements, and the relationship between her and her dog “Baby.” Much as with Darwinism, it all seemed so reasonable, so plausible, so sane.

And then the woman let loose with a whopper. She described how Baby took such good care of her, how Baby would go to the fridge, how Baby would prepare a sandwich for her, how Baby would even cut off the crust from the bread just the way she liked it. At that point my wife, with relief, saw the judge look up, nod, and sign the commitment papers.

As you read Rosenhouse’s reply, your reaction ought to be that of the judge. If it isn’t, you should consider (but of course won’t) whether Darwinism has addled your brain.