## The Growth of Complexity

Dembski's specified complexity is an important piece of a broader creationist argument: that the complexity of biological organisms is so large that they could not have arisen through natural processes; hence, intelligent intervention is needed to explain them. We have now seen where this argument breaks down. Specified complexity is not a well-defined quantity, and there is no law of conservation of information that ensures that complexity cannot grow over time.

Where, then, does the complexity of living things come from? We cannot answer this question until we have a satisfactory definition of the word complexity. If complexity means Kolmogorov complexity, then there is nothing to explain since random processes can generate as much complexity as we like. If complexity means specified anti-information as discussed previously, then as much complexity as we like can be generated by simple computational processes.

How do simple computational processes arise in nature? This is an interesting question whose answer we are just beginning to understand. It turns out there are many naturally occurring tools available to build simple computational processes. To mention just four, consider the recent work on quantum computation (Hirvensalo 2001), DNA computation (Kari 1997), chemical computing (Kuhnert et al. 1989, Steinbock et al. 1995, Rambidi and Yakovenchuk

2001), and molecular self-assembly (Rothemund and Winfree 2000). Furthermore, it is now known that even very simple computational models, such as Conway's game of Life (Berlekamp et al. 2003), Langton's ant (Gajardo et al.

2002), and sand piles (Goles and Margenstern 1996) are universal and hence compute anything that is computable. Finally, in the cellular automaton model, relatively simple replicators are possible (Byl 1989).

Dembski's claim that CSI is a trademark of intelligent agents is therefore suspect. CSI provides no way to separate the actions of intelligent agents from the results of simple, naturally occurring computational processes. Indeed, intelligent agents themselves could well be naturally occurring computational processes.

## Post a comment