Book Reviews
Origins & Design 17:2

Alchemy, NK Boolean Style

At Home in the Universe: The Search for the Laws of Self Organization and Complexity
Stuart Kauffman
New York: Oxford University Press, 1995, 321 pp.

William A. Dembski

There's an old joke about the philosopher Rudolf Carnap and his method of doing philosophy. According to the joke, Carnap's method was to begin any philosophical investigation with the statement "Consider a formal language L." As the good logical positivist he was, Carnap desired the precision inherent in formal languages. Unfortunately, precision has its price. Formal languages are not natural languages and the problems expressible in formal languages need not connect to actual problems in the real world. With formal languages the question ever remains whether they adequately capture the subject under investigation.

Appropriately modified, the joke about Rudolf Carnap can be retold about Stuart Kauffman and the scientific method he employs in At Home in the Universe. According to the modified joke, Kauffman's method is to begin any scientific investigation with the statement "Consider an NK Boolean network." Indeed, throughout At Home in the Universe just about every real-world problem gets translated into a toy-world problem involving NK Boolean networks. As with Carnap's formal languages, NK Boolean networks have the advantage of complete logical precision. But they also suffer the disadvantage of losing touch with reality. And it is this disadvantage which ultimately proves the undoing of Kauffman's project.

In broad terms, Kauffman's project is quite simple. Kauffman wants to account for the origin and development of life. Now there is a dominant scientific theory that purports to do just this, namely, the neo-Darwinian synthesis. Nonetheless, Kauffman is dissatisfied with it. According to Kauffman the twin pillars of the dominant theory--mutation and natural selection--are inadequate. Kauffman wants therefore to supplement these twin pillars with yet a third pillar, what he calls "laws of self-organization."

Unfortunately, just what these laws of biological self-organization are is at this point far from evident. The subtitle of At Home in the Universe quite properly reads The Search for the Laws of Self-Organization and Complexity. To date this search has been unsuccessful. Indeed, Kauffman admits that he is searching for laws he has yet to find. An obvious question therefore arises: Whence Kauffman's confidence that such laws even exist?

Kauffman's confidence rests in an analogy. In the non-linear dynamics of physics and the simulations of computer science, Kauffman finds self-organizational scenarios that are suggestive of what might have happened in biology. Kauffman's project therefore is to use non-linear dynamics and computer simulations to massage our intuitions, make the search for laws of self-organization seem plausible, and ultimately facilitate the discovery of such laws. And of course, the main analytic tool for carrying out this project is his NK Boolean networks.

Before considering the details and merits of Kauffman's project, it will help to understand the motivation behind it. Throughout his book Kauffman makes very clear that he is after a non-mysterious account of the origin and development of life. For Kauffman such a non-mysterious account is one that appeals only to natural laws and is devoid of any reference to a "master choreographer" (p.208). Appeals to intelligent design are therefore ruled out from the start.

All the same, Kauffman is not wholly without a sense of mystery and the sacred. At the end of his book Kauffman encourages us to "reinvent" the sacred. Indeed, a religious impulse underlies Kauffman's rejection of strict Darwinism, with its exclusive dependence on mutation and selection. As Kauffman sees it, strict Darwinism makes the universe a giant test tube within a stochastic chemistry lab. To reinvent the sacred, Kauffman needs the universe to be more than a test tube--it needs to be our home. And for the universe to be our home, our place in the universe must be assured. Laws of self-organization hold such a promise. Thus Kauffman will write, "I would rather life be expected in this unfolding since the Big Bang than that life be incredibly improbable in the timespan available" (p. 304).

Let us now turn to the details and merits of Kauffman's project. To strict Darwinists like Richard Dawkins and Daniel Dennett, Kauffman's project will seem thoroughly misguided. Indeed, what is the point of positing unknown laws of self-organization to explain biological phenomena when these same phenomena have already been adequately explained in terms of mutation and selection? Kauffman's project casts doubt where there ought to be certainty. Kauffman's project problematizes what ought not to be a problem. For the strict Darwinist, laws of self-organization are leeches whose only effect is to sap a perfectly good theory (i.e., the neo-Darwinian synthesis) of its vitality.

Yet for a growing number of dissenters, who view mutation and selection as inadequate to account for the full diversity of biological phenomena, Kauffman's project is full of promise. Our choices are, after all, limited. The order we observe in living systems is either exogenous or endogenous--either it is imposed from without by an intelligent cause or it arises spontaneously from the intrinsic properties of matter. For Kauffman exogenous order is mysterious and to be rejected out of hand. Hence the order must be endogenous. And since mutation and selection are inadequate to account for the order, still some third factor must be invoked to yield an adequate explanation of life. According to Kauffman this third factor is the laws of self-organization.

At this level of generality Kauffman's project remains full of promise. The laws of self-organization that Kauffman envisages derive their inspiration from the non-linear dynamics of physics and the simulations of computer science. Here we have extremely active areas of research that regularly exhibit emergent properties of complex systems. Moreover, if Kauffman is correct in assuming that the order we observe in living systems is endogenous, then it makes perfect sense to treat life as an emergent property of a complex system.

If we delve a little deeper, however, Kauffman's project evaporates into a promissory note with little or no substance backing it up. Kauffman is notoriously slippery about defining terms. Order and complexity, the very things his hard-sought laws of self-organization are supposed to produce, are themselves never defined. Consider Paley's watch and a rock on a beach. Both can be viewed as highly complex. Indeed, the instructions to construct a watch and the instructions to construct an exact replica of an irregularly hewn stone will both be vast. Moreover, both have an order. How does the complexity and order of the one differ from the other? Kauffman never clarifies what he means by complexity and order, or how these concepts apply in one case but not another.

Or consider Kauffman's ubiquitous NK Boolean networks. An NK Boolean network is a set of N nodes, each of which is assigned a value of 0 or 1, and which in being reassigned values depends functionally only on the values already assigned to K fixed nodes (K < N). No doubt, Kauffman's NK Boolean networks are capable of exhibiting many interesting behaviors. But to call the nodes "genes," as he is wont to do, and then take what he interprets as self-organizational behavior by the Boolean network as evidence for self-organization in the formation and development of life is utterly gratuitous.

Nowhere does Kauffman even attempt to establish a correspondence between the mathematical models he runs on his computer and the actual processes matter must undergo to form a biological system. I find this omission unconscionable, for it represents a descent into mysticism worse than any Kauffman claims to avoid. Kauffman will write, "it is not implausible that life emerged as a phase transition to collective autocatalysis once a chemical minestrone, held in a localized region able to sustain adequately high concentrations, became thick enough with molecular diversity" (p. 274).

This is not science, but alchemy (cf. p. 277 where Kauffman actually uses the word "alchemy" to describe what he is doing). Indeed, once Kauffman leaves the pristine world of mathematical modeling and computer simulations, and turns to the messy world of matter in motion, he can do no better than alchemy. Kauffman's laws of self-organization must do their self-organizing all by themselves. A supracritical mixture of diverse molecules (Kauffman's "chemical minestrone") operating according to laws of self-organization must--if he is right--be able to work the magic of life. Get the proper mixture and life will emerge.

Nor can Kauffman's approach ever get beyond alchemy. A very damning admission occurs when Kauffman considers a rather large NK Boolean network in which N equals 100,000. Previously Kauffman has contended that life constitutes an attractor for an autocatalytic set of chemicals. Such an auto-catalytic set will be exceedingly more complicated than the NK Boolean network he is now considering. And yet Kauffman will admit that finding an attractor even for this Boolean network will be all but impossible: "I cannot show you an attractor in such an unfathomable state space" (p. 100). If Kauffman's stylized mathematical models are unfathomable, how much more nature herself?

Apparently oblivious to how it undercuts his program, Kauffman repeats this admission later on, and even more forcefully: "It is one thing to talk about supracritical reaction systems blasting off into the outer space of chemical creativity, but all confined to a computer model. It is quite another thing to fathom what might go on in a real chemical system" (pp. 118-9). And fathom it he never does. Kauffman has not one thing to say about real chemical systems except the unsubstantiated assertion that the right mixture of chemicals governed by laws of self-organization will yield life. But the laws of self-organization he cites are unknown. And even if they were known, there is no reason to think that we could ever apply them (after all, we know the laws that govern toy examples like Boolean networks, and can't even apply them there). Alchemy plus inscrutable laws of self-organization will ever remain alchemy.

All the problems inherent in the origin and development of life are still there after one finishes the book.

Copyright © 1996 William A. Dembski. All rights reserved. International copyright secured.
File Date: 11.14.96