Hello, Stuart Kauffman | Evolution News

In an earlier post, I noted that theoretical biologist Stuart Kauffman’s definition of life implicitly includes irreducible complexity. 

Kauffman is one of the more interesting participants in the debates about the nature of life, since he takes seriously the perplexing theoretical details needed to explain self-replicating, purpose-driven systems. He is also one of the more sophisticated critics of the “modern synthesis” of neo-Darwinism, and he has expressed the sentiment that it is deeply important for those who are convinced of evolution to understand why some people reject it in favor of intelligent design arguments. An Evolution News contributor has even credited Kauffman’s writings with helping to free him from the prison of a materialist worldview. 

But all that does not mean that Kauffman is sympathetic to intelligent design. On the contrary, he is quite opposed to it. Kauffman has explicitly addressed the irreducible complexity argument for design (albeit in a rather cursory and dismissive way), and since his work has been cited here pretty frequently recently, I think it would be worthwhile to answer him on that point.  

Kauffman’s Second Argument

Kauffman begins his attempt at a refutation by invoking the old cooption or “exaptation” explanation for irreducible complexity. But I don’t want to beat that dead horse¹ here. Instead, I would like to respond to Kauffman’s second argument, which is more unusual.² He says: 

Furthermore, as I’ve described before, we can have no probability measure for the evolution of the biosphere into its Adjacent Possible, since we do not know all the possibilities, hence we do not know the sample space of the process, so cannot construct a probability measure. Therefore, the calculations of improbability that the ID proponents make are vacuous. 

If I understand Kauffman correctly,³ he is arguing that while any given functional molecular system may be individually unlikely to evolve, we don’t know how many kinds of (perhaps equally unlikely) functional molecular systems are theoretically possible. If the number of possible functional systems is high enough, then the odds of getting one of all the many possibilities might not be excessively low — even though each system might, individually, have very low odds of appearing.

For example: Suppose the odds of getting the mutations needed for a bacterial flagellum at some time in the course of life’s history are one in a gazillion. That’s very improbable. However, what if there are half a gazillion similar molecular machines that could have appeared but didn’t — each one just as unlikely as the flagellum, but also just as extraordinary? In that case, the odds of getting one of those functional machines would be just one in two, a 50 percent chance! Then, although each extraordinary machine would indeed be individually unlikely, it would not be unlikely that one of them would appear. We could say, “Well, it was gonna be something, and it just happened to be the flagellum.” 

A Possible Protest

Of course, we could protest that there might not be half a gazillion possible flagellum-like machines. But that’s Kauffman’s point: we can’t know. Since the number of theoretically possible functional molecular machines is undefinable, there is no way to calculate how many possibilities are out there, floating around in Platonic space. So there is no use in trying to calculate the odds of one of them appearing by chance. 

This is a very clever argument. In fact, it is a bit too clever. It explains too much. After all, the same argument can be used to explain away anything that appears to be designed.

To see what I mean, let’s take a different scenario,⁴ and see how Kauffman’s argument applies in it. 

Imagine that two astronauts have just landed on a distant planet. We’ll (completely arbitrarily) call them “Mike” and “Stuey.” They step out of their spacecraft, and stand there in awe, surveying the desolate alien landscape. After a minute, Mike happens to look down at the ground at his feet, and what should he see, but (wonder of wonders!) a group of identical stones arranged neatly in the shape of a message. It reads:

“Hello, Mike and Stuey. Welcome to my planet.”

Mike: Well, I’ll be jiggered. It looks like there’s intelligent life here, and it knew we’d be coming!  

Stuey: [glances down at the message] I doubt it. We have no reason to believe this planet is habitable, and it certainly appears devoid of life. 

Mike: Uh… okay, sure. But then, how did this message get there? 

Stuey: Those lines of stones? I’ll admit that they’re extraordinary. But stones can be arranged in a lot of ways. There’s no reason they couldn’t have emerged by luck. 

Mike: Ha! I bet if we do the math, we’ll find that the odds of the stones being arranged in the shape of that exact message is practically infinitesimal. Let’s see here… [Mike takes out a pocket calculator]

Stuey: Don’t bother doing the math. You’re forgetting that there might be any number of ways to arrange the stones that you would find equally extraordinary. Indeed, perhaps there is an infinite number of strange and unlikely arrangements! So while the odds of getting this particular extraordinary arrangement of stones may indeed be vanishingly small, how can you be sure that the odds of getting any such extraordinary arrangement is small? 

Mike: [has already wandered off to look for the alien] 

The analogy of finding a message on the ground is popular as a reductio ad absurdum argument against claims of randomness in nature because it’s very easy to see in this example that the design inference is reasonable. Typically, someone who wants to refute the argument will contend that biological structures are different from messages on the ground in significant ways, or that the mechanism of random variation and natural selection makes them more probable. I won’t get into those arguments here, however, because that is not the tack Kauffman is taking this time. He’s not denying the improbability, complexity, or even the specificity of biological structures. Nor is his argument an explanation for how unguided evolutionary processes can construct such systems. 

With Stuey and Mike

Rather, Kauffman is calling into question the improbability of improbable things in general. His argument does not require Darwinism, and in fact would work just as well if he were claiming that complex life had simply appeared in a puff of smoke, by a chance arrangement of atoms. It’s a very clever argument — which would render him very non-functional if he actually followed it in his everyday life. If Kauffman were with Stuey and Mike, I have no doubt that he would take Mike’s side, and conclude that some inexplicable entity was trying to get their attention. 

I think that anyone who honestly considers the tale of Stuey the astronaut should be able to see that there is something seriously wrong with Stuey’s line of reasoning. But it may be less obvious exactly what is wrong with it. 

After all, Stuey was quite correct in saying that it is not possible to calculate the number of possible “extraordinary” stone-arrangements. Likewise, Kauffman is (no doubt) correct that there is an unlimited number of potential highly specified functional molecular structures.

The trouble is, having infinite chances does not necessarily make something infinitely likely — or even somewhat likely. We know intuitively that the ways to arrange stones into a message, infinite though they be, are dwarfed by the ways of not arranging the stones into a message.  Think of it like an infinite line running through an infinite plane. If God threw a dart randomly at the plane, he would have an infinitely low chance of hitting the line — even though the line is infinite! An infinite number can still be an infinitesimal fraction of the total probability space. 

Therefore, it is not enough to posit that the possible number of complex molecular machines is undefined or potentially limitless. What matters is what relative proportion of the total possibilities is comprised of functional machines. Kauffman is implicitly arguing that this proportion could be unexpectedly high. 

A Hypothesis About Probability Space

The first problem with that explanation is that it contradicts all observations about the universe outside of biology. And since Kauffman’s hypothesis is a hypothesis about probability space, it is not unreasonable to look for evidence of it in the universe in general, not merely in biology. That is to say, if the probability space were as favorable to the production of complex machines as Kauffman thinks — so probable that you could sometimes get them by chance in biology — then you would expect them to appear by chance in the non-biological world as well. Yet, as far as we have seen, they don’t. Once we exclude (1) living organisms and (2) intelligently designed artifacts… complex machines simply don’t exist in the observed universe. 

Therefore, to maintain the plausibility of his hypothesis Kauffman would need to additionally argue that there are certain features of organic life, which set it apart from everything else, that make complex machines likely to be seen in life while remaining unlikely in everything else.

I would be very interested to know what Kauffman thinks those differences are. If I were tasked with solving the problem, I would suggest that perhaps complex machines do appear by chance in non-living systems, but the fact that natural selection multiplies them when they appear in self-replicating systems is enough to make them likely to be observed by us when they appear in those systems, while remaining unlikely to ever be observed when they appear in non-living / non-self-replicating systems. The problem with all such explanations is that no such thing has been theoretical demonstrated, much less empirically observed. That means Kauffman’s argument is at best an argument from ignorance — there could be the right number of theoretical machines in the probability space to make them likely to be seen in living systems yet unlikely everywhere else. 

Explanations of this kind are a good way to avoid undesirable conclusions, but they are not often used in practical situations. That is because they violate the principle of parsimony, or Occam’s razor, the rule-of-thumb that says that you should defer to the simplest explanation and not invoke any unnecessary entities. Occam’s razor is not a fool-proof rule — sometimes a phenomenon really does have a strange and convoluted explanation — but it is impossible to function in life as a rational being without generally adhering to this principle. 

Hello, Mom

For example, if you call your parents’ landline and your mom’s voice answers on the other end, it is theoretically possible that you are speaking to a terminator robot from the future that is able to perfectly mimic your mother’s voice. In fact, it might be very difficult to rule it out as a possibility. However, it would be unnecessary as an explanation, because you already have a perfectly good explanation for the phenomenon of your mom’s voice on the phone — namely, your mom.  

In the example of the astronauts, the problem is a bit harder. While you have (probably) observed your mom at some point, the astronauts in the story had not observed any being on the new planet that could have written the message. However, they did know, at least, that intelligent beings are capable of writing such messages. It didn’t matter that the explorers had no way of guessing who or what the message-writer might be. Since an already-observed phenomenon — a message-writer — was available as an explanation, there was no reason to invoke the unattested explanation of there being, perhaps, “a relatively high proportion of meaningful stone arrangements in the probability space.” 

What Needs Explaining

In the problem at hand, we have found something that needs explaining: adaptive functions that are only produced by a mind-bogglingly precise and complex orchestration of different parts, and therefore would not be likely to form incrementally by natural selection and random variation. Like the astronauts, we have not actually observed any being in the universe that could have orchestrated those specific systems. But we do know, with the most reliable and intimate sort of knowledge possible, that conscious agents are capable of coordinating different parts with foresight to produce a desired function. It doesn’t matter that (based on the evidence of biology alone) we have no way of guessing who or what the designer might be. Since an already-observed phenomenon — planning by a conscious agent — is available as an explanation, there is no reason to invoke the as-of-yet-unobserved explanation of there being “a relatively high proportion of functional molecular system arrangements in the probability space.” 

This is a very in-the-weeds way of saying something very intuitive. Actually, it’s the whole reason Darwin’s explanation was necessary in the first place. Before Darwin, natural philosophers who wanted an alternative to theism had to invoke mere chance to explain life,⁵ and this was not widely persuasive — the inherent improbability of complex, functional systems was a grave challenge to a belief in an undesigned universe. Darwin knew that, and he took the challenge seriously. If Stuart Kauffman is really serious about understanding the resistance to Darwinism, he should take it seriously, too.

Notes

1. In 1996, the evolutionary biologist H. Allen Orr wrote a highly negative review of Michael Behe’s Darwin’s Black Box, which was mostly devoted to trying to refute Behe’s argument that irreducible complexity is evidence of intelligent design. This is what Orr had to say about exaptation: “[W]e might think that some of the parts of an irreducibly complex system evolved step by step for some other purpose and were then recruited wholesale to a new function. But this is also unlikely. You may as well hope that half your car’s transmission will suddenly help out in the airbag department. Such things might happen very, very rarely, but they surely do not offer a general solution to irreducible complexity.” It doesn’t take an ID advocate to see that the exaptation is pretty outlandish as an explanation for the formation of irreducibly complex systems, and I suspect it only became popular once it became apparent that there was no better explanation available.  
2. Though the philosopher David Berlinski (now a Discovery Institute Senior Fellow) made a similar argument in a 2002 Commentary essay.
3. Kauffman is presumably not saying that there is no way to calculate the number of possible DNA-base pair arrangements for a given length of genetic code, since that is easy to calculate. He could merely be saying that an evolving organism could be constrained so that it can only mutate in a relatively small number of ways, including towards the (in-and-of-itself improbable) base pair arrangement that would produce the irreducibly complex system in question. However, that would merely push back the question — the constraints on the evolutionary process, rather than the final product, would then be the improbable thing. It would appear that the evolutionary process itself was designed, like a marble sent rolling along a fixed track. But we haven’t found any such constraints yet; so that’s a bridge to cross we when come to it. 
4. The illustration of finding an unlikely artifact on an apparently uninhabited planet is not original to me, but, regrettably, I can’t remember who first used it. Coincidentally, David Coppedge imagined a similar scenario in an article here yesterday.
5. See, for example, Lucretius, On the Nature of Things 5.416-431 and 5.821-854, on the formation of the first complex organisms by the random movement of atoms; also Cicero, On Divination 2.21, on the possibility of improbable events occurring.