Science and Theory

Science is all about theories, and science can't work without theories. Unfortunately, few non-scientists have any idea what a "theory" is.

While the popular notion equates "theory" and "guess", a scientific theory is a great deal more than a guess. It's a model that has been found to match reality. In particular, it's been found to be a better match than any other idea that's been offered. Theories are offered and rejected all the time, as they fail to meet the tests demanded by science.

There are two issues in creating good theory: (1) getting the modeling right and (2) using assumptions relevant to nature. Sometimes the former dooms theory, but in most instances it is the latter.

Not too long ago, Michael Behe and David Snoke published a paper that purported to show that complex systems would take an inconveniently time to evolve by the usual methods.

Unfortunately for Intelligent Design/Intelligent Origin Theory (ID/IOT), there are significant gaps in their model.

We believe there are some issues with the modeling in the paper, but investigating them is too complex for this essay. However, even a rough reading of the paper makes it clear that Behe and Snoke's work and the conclusions they draw are not relevant to nature.

So what are the problems?

Behe and Snoke are attempting to estimate how long and how large of a population it would take for a protein in the absence of selection to evolve a new binding site or other complex feature. They assume up-front that multiple amino acid substitutions would be required before the new feature can be preserved by natural selection. Behe and Snoke are modeling the formation of a completely new binding site in a duplicated protein. While this process is important in generating some kinds of new function, in the majority of duplicated proteins, existing binding sites are either modified to act on new substrates or new catalytic mechanisms. Duplication of enzymes and modification of their existing binding sites can produce quite complex pathways; for example the clotting cascade is due to duplication of proteolytic enzymes with a subsequent change in substrates.

(Parenthetical statements removed. Read the original.)

Basically, they are assuming that:

1. The protein system must have two specific functions
2. Its formation requires two specific mutations
3. The mutations must occur in two specific genes
4. The mutations must occur in specific locations
5. The mutations must occur in a specific population
6. The mutations are not subject to natural selection until both are in place
7. Only point mutations exist
8. The population of organisms is asexual

The first four items are part of the "one true sequence" model which ID/IOT assumes. In fact, when we look at real living things, we often find numerous different ways to carry out any given function. Given the variety of different ways different organisms will do the same thing, assumptions 1–4 are very hard to support.

I'll skip the discussion of real-world evolution in real-world living things and cut to the chase. Stipulating all of Behe and Snoke's assumptions, how improbable is the evolution of complex systems?

Despite using assumptions that render their model overly pessimistic, the population size and generation time that Behe and Snoke calculate is not prohibitive for the types of organisms (haploid, asexual) that it is most applicable to. The authors conclude that population sizes of 109 would require at least 108 generations to evolve a two-site MR feature (λ=2) under their model. And while this does seem prohibitive for large, multicellular eukaryotes, it's actually easily achievable for bacteria. A population size of 109 is what one would find in a very small culture growing in a lab; even small handfuls of dirt, or the average human gut, will contain populations in excess of this number. Bacteria reproduce quickly; under optimal conditions for E. coli, 108 generations will occur in less than 40,000 years, a geological blink of the eye. Given that there are about 5x1030 bacteria on Earth (Whitman et al. 1998), we should expect the evolution of novel MR features to be an extremely common event -- an average of many times per microsecond -- even if we accept Behe and Snoke's unrealistic assumptions. Since we can be confident that their numbers are a vast overestimate, Behe and Snoke have ironically demonstrated that the evolution of novel gene functions is not unlikely at all. And yet, it has been a long standing claim of the ID movement that the evolution of "novelty" simply cannot happen, period. Behe and Snoke have done us the favor of disproving this bogus notion once and for all.

(This isn't the first time a pro-creationism or pro-ID/IOT discussion has wound up supporting evolution. I've seen articles in Creation Research Quarterly that wind up being very nice discussions of known methods of natural selection in the wild.)