Sheer vs. Real Possibilities: A Response to Allen Orr

By William A. Dembski


Allen Orr reviewed my book No Free Lunch in the Summer 2002 issue of the Boston Review. Orr's review is available at The response below is at the request of the Boston Review and will be appearing in a subsequent issue.



Though Allen Orr raises many specific objections to No Free Lunch, they ultimately hinge on a question he addressed to me directly in his review: "What's gained by replacing a mysterious material order with an equally mysterious designer?... It would have been interesting to hear Dembski's [response]." Underlying Orr's question is David Hume's concern that design offers no advantage explaining the organization of a material system when that system can equally well be conceived as organizing itself.


In No Free Lunch I argue that material systems are not capable of organizing themselves into complex specified structures apart from intelligence. In particular, I argue that since biological systems exhibit specified complexity, intelligence is involved in their production (just when and how is a matter for further study). Hume's concern is therefore met with Aristotle's distinction between nature and design. Nature produces things by self-organization and generation; design (the Greek techne, usually translated "art") produces things by impressing form from without. Thus, to use one of Aristotle's examples, acorns have it within themselves to produce oak trees, but raw wood does not have it within itself to produce ships. As Aristotle put it, "The art of ship-building is not in the wood." Indeed, it requires design.


But how can we tell when a material system is capable of organizing itself and when it requires the addition of design? In biology, how can we tell whether the Darwinian selection mechanism is capable of reorganizing existing biological structures into vastly more complex ones? In No Free Lunch I lay out an information-theoretic apparatus for answering such questions. Orr, to be sure, thinks that this apparatus cannot bear the weight I put on it, especially in biology. It is instructive, however, to see why Orr thinks that. Ultimately, we part company over what it means to attribute a capacity for self-organization to material processes (and specifically to the Darwinian mechanism).


To attribute to something an ability or capacity (or for that matter an inability or incapacity) is not nearly as straightforward as Orr seems to think. Indeed, this accounts for Orr's light dismissal of Michael Behe's work on irreducible complexity. Behe has argued that irreducibly complex biochemical machines, by being composed of numerous parts each of which is necessary for the system's function, could not be produced by gradual Darwinian means. To say that such systems "could not be produced" is to attribute an inability or incapacity to the Darwinian mechanism. Yet Orr is able to imagine scenarios in which parts gradually get folded into such systems -- parts that, though not originally necessary, become necessary over time. Thus, according to Orr, the intelligent design community must admit that "Behe's chief claim was wrong. Irreducible complexity is accessible to Darwinism."


To say that irreducible complexity is accessible to Darwinism means, obviously, that Darwinism is able to access systems displaying that property. But what does it mean to attribute that ability or capacity to Darwinism? Does it simply mean that Darwinists are able to imagine how a Darwinian process might lead to such systems? Or does it mean that Darwinists have realistically assessed the Darwinian selection mechanism's ability, as actually operating in nature, to produce irreducible complexity. Ability wears many faces, characterized by different forms of possibility (derived from the Latin posse, meaning to be able). Orr, along with much of the Darwinian community, is satisfied with a very undemanding form of possibility, namely, conceivability. So long as he can conceive a Darwinian pathway to irreducible complexity, nature trumps design.


Behe, by contrast, requires a much more demanding form of possibility in assessing the ability of the Darwinian mechanism to produce irreducible complexity. For Behe, it's a probabilistic form in which highly improbable, functionally specified structures cannot happen by chance. This weds Behe's work on irreducible complexity to mine on specified complexity. Both Behe and I understand chance here very broadly, and thus include the Darwinian mechanism of natural selection and random variation. The logical force of our argument purports to be the same as "You can't walk into a Las Vegas casino and get a hundred double zeros in a row playing roulette." There's a sheer possibility that this could happen by chance, but not a real possibility.


Now chance in a casino is much more tractable than chance in nature. Hence I need to develop an information-theoretic apparatus for handling chance quite generally. I also need to develop bridge principles for mapping this apparatus onto biology. I do both in No Free Lunch. Orr is dissatisfied with my treatment. But the reason for that dissatisfaction has nothing to do with the coherence and validity of my apparatus or its applicability to biology. The reason, rather, is that for Orr, Darwinism has the alchemical property of transforming sheer possibilities into real possibilities. One can see this at several places in his review.


For instance, he remarks that Behe's "chief claim was that Darwinism just couldn't get here from there [i.e., achieve irreducible complexity]." According to Orr, that claim is "dead wrong." Why? Not because Orr provided detailed, testable, causally specific instances of the Darwinian mechanism producing irreducible complexity, but because he can imagine a Darwinian scenario that gives rise to irreducible complexity. Orr's criterion for possibility is conceivability. This enables him to sidestep the real limitations confronting the Darwinian mechanism and focus instead on the sheer possibilities that an unbridled imagination can create for the Darwinian mechanism. This is also why he doesn't like my demand for "causal specificity," which requires Darwinists to demonstrate that their selection mechanism actually has the causal power to produce irreducible complexity. Instead, Orr substitutes a much weaker demand for "historical narrative," which in the case of Darwinism degenerates into fictive reconstructions with little, if any, hold on reality.


The subtext of Orr's review, though unintended, is that Darwinism is not a solution to the problem of biological complexity but an exercise in delusion by which evolutionary biologists convince themselves that they've solved the problem when in fact we haven't.