My claim about specified complexity needs to be fleshed out. Completeness is a very general result; it tells us nothing about how chance and necessity combine for the sort of creativity leading to organisms or artifacts, only that such creativity is possible. It tells us Dembski must be wrong, but it does not say where his mistake lies.
How can we get creativity out of the raw novelty that randomness provides? How can we incrementally build up complex machines? One way to do so has been known for some time: Darwin's mechanism. Biologists know, in considerable detail, how this mechanism works. Although formal demonstrations of Darwinian mechanisms gradually increasing information have so far been confined to simple simulations (Schneider 2000, 2002), there is no reason that such results cannot be generalized to more-complex and -realistic biological scenarios.
In fact, a striking development in recent decades has been the way in which Darwinian thinking has taken root outside biology. In particular, in AI and cognitive science, Darwinian approaches to the mind have become prominent. We have Gerald Edelman's (1992) neural Darwinism; the Darwin machines and variation and selection in the brain proposed by William Calvin (1996); memes, a nongenetic form of replicating information operating in the realm of culture (Blackmore 1999, Aunger 2002); Daniel Dennett's (1995) multiple levels of Darwinian mechanisms depending on processes competing to assemble our stream of consciousness; and more. Researchers in machine intelligence are increasingly relying on Darwinian mechanisms to introduce creativity into machines—even beginning to explore art and original engineering designs (Fogel 2000, Bentley 1999). AI is no longer an enterprise devoted to canned, preprogrammed strategies; it includes open-ended, evolutionary behavior.
This is not to say that human creativity is even close to being fully explained. Darwinian variation and selection are almost certainly vital parts of the picture; but other mechanisms, such as conceptual blending (Fauconnier and Turner 2002), are bound to be crucial for the rich creativity we find in ourselves. We are just beginning to find out. Even now, however, we can say with some confidence that human intelligence is not something separate from a world of chance and necessity.
ID proponents like to portray not just evolutionary biology but also AI and cognitive science as stagnant fields, unable to overcome deep, persistent problems. They argue that conventional research cannot overcome those problems because of these fields' commitment to inadequate theories such as Darwinian evolution. Admitting intelligent design as a separate principle, they say, will clear the way, leading to the required breakthroughs. The actual developments in these fields, however, are very different. We keep making progress in understanding not just biology but also human intelligence itself in terms of chance and necessity.
What, then, of no free lunch, of METHINKS IT IS LIKE A WEASEL, of all the ways in which Dembski argues that chance and necessity can do no more than shuffle existing information? Dembski's mistake is subtle but straightforward: he conceives of evolution as a way to search for a solution to a predetermined problem. It is nothing of the sort. Darwinian evolution is creative precisely because nothing is predetermined and everything may be randomly modified.
If evolution truly was a search for a high point on a fixed fitness landscape, in the manner of a genetic algorithm, Dembski's argument might be plausible. In that case, allowing a machine to make random decisions would not change what it is capable of solving and what it is not. Then the kinds of search procedures that would work well or not would depend on the fitness landscape, so we might be tempted to think a Darwinian mechanism introduces no genuine novelty. The problem is set; therefore, finding the solution becomes a matter of letting the information inherent in the problem bubble up to the surface.
As biologists point out (Orr 2002), evolution is not at all like a search on a fixed fitness landscape. Living populations are not searching for a solution to a preset problem. Their fitness landscape is continually changing, and this change is largely due to the other organisms that make up an important part of an organism's environment. Even an organism's own reproductive strategy alters the fitness landscape. All that is important is being able to reproduce, and what works best at any one moment is not likely to remain so forever, since competitors are themselves always changing.
The no-free-lunch theorem that Dembski relies on does not apply when the fitness landscape changes in a way that depends on the population (Wolpert and Macready 1997). Indeed, being able to randomly alter strategy is important since competitors may adapt to any set strategy and exploit it. And a prime way to generate increasing complexity in biology is to have evolutionary arms races (Dawkins 1986, 178). In an arms race, competing populations can climb smooth hills of fitness constructed by the competition itself, using variation and selection.
Physicists working to explain complexity also recognize the importance of this point. Old-fashioned creationists often challenge evolutionists to explain how, in a world tending to disorder because of the second law of thermodynamics (see chapter 7 in this book), biological order is supposed to increase without intelligent intervention. Cast in physical terms, Dembski's arguments about chance and necessity being able only to preserve or degrade specified complexity are a close cousin of the creationists' second-law argument.
Spontaneous ordering of the sort we see in evolution can take place in systems driven away from thermodynamic equilibrium. One example is the universe after the big bang. The expansion of space means that the maximum possible entropy of the universe increases faster than the actual entropy. This gap creates opportunities for order to form. Evolution, in fact, works for just this reason (Edis 1998b). As species diversify, the diversity actually realized increases more slowly than does the number of all possibilities (Brooks and Wiley 1988).
In other words, assembling complexity through chance and necessity depends on an expanding set of possibilities. It requires a changing world, one in which, by accident, history can take a genuinely new path to the exclusion of others. In contrast, if we have a set destination, history is merely about success or failure. If all we had was a search for the best spot on a fixed fitness landscape, evolution could not even take hold, let alone be genuinely creative.
Was this article helpful?