Evolution Icon Evolution
Intelligent Design Icon Intelligent Design

Does Tree-Like Data Refute Intelligent Design?

Image: Darwin’s Tree (1837), a sketch from his First Notebook on Transmutation of Species, via Wikimedia Commons.

As we’ve seen in two recent posts (here and here), in 2009 FORA.tv asked Richard Dawkins to give the single fact that he believed would refute “creationism.” I responded to Dawkins’s claim that the genetic data forms a “perfect hierarchy — a perfect family tree.” But even if genes don’t fit a “perfect tree,” they can often be organized into a pattern that is “treelike” to some limited extent. This is in part because the computer programs that evolutionary biologists use to do phylogenetic analysis presuppose common descent and thus necessarily generate treelike patterns irrespective of the genes sequences that are fed into them.

For the Sake of Argument

But let’s say, for the sake of argument, that the data generally fit a tree, even if imperfectly. Does this mean, as Dawkins goes on to suggest, that the designer has “deceive[d] us in the most underhanded and devious manner” in order to make it look like different species are genetically related when they really aren’t? Not at all — because of the principle of common design, where intelligent agents re-use parts and components that perform common functions in different designs. 

Everyday examples of this include wheels being used on both cars and airplanes, or touchscreen keyboards used on both phones and tablets. Because certain components often “go together” in engineering — e.g., wheels always come with axles, or keyboards always come with the same set of keys — you would expect that even a set of designed structures could be organized into a somewhat “treelike” pattern. I once took shirts in my closet — all intelligently designed — and scored their characteristics (e.g., buttons vs. no buttons; short sleeve vs. long sleeve; collar vs. no collar) and I found that this intelligently designed dataset of my shirts could be fit into a tree quite well. As I wrote in Debating Darwin’s Doubt:

Critics would never say we can only refute common ancestry if the data is so bad that it’s “random.” Non-common-ancestry-based datasets could frequently have higher-than-random Cis [CI = Consistency Index, a measure of the degree to which a dataset is “treelike”].

For example, I blindly grabbed 12 shirts from my closet and scored them for 16 traits (e.g., buttons, zipper, solid color, etc.). While scoring the data, I immediately observed non-random correlations that would give my dataset a decent CI. For example, shirts with buttons are often associated with a solid color, a pocket, and a collar. My tree had a CI of 0.76 — much higher than ~0.25, what G. J. Klassen et al. (1991) say we should expect from a random dataset of that size. Yet common ancestry did not generate the shirts in my closet! What does this show? A dataset with a higher-than-random CI doesn’t necessarily imply common ancestry. Why? Common design predicts re-usage of parts in a non-random manner that fulfills design constraints required by the system. 

Debating Darwin’s Doubt, p. 124

The ability to fit a set of data into a “treelike” pattern is not necessarily incompatible with intelligent design. Nonetheless, it’s clear that a lot of data does not fit a treelike pattern. Computer scientist Winston Ewert applied the concept of “common design” to propose a “dependency graph” model of organismal relationships. This was based upon the principle that software designers frequently re-use the same coding modules in different programs. In a paper in BIO-Complexity, Ewert tested his model by comparing the distribution of gene families in nine diverse organisms to a treelike pattern predicted by neo-Darwinism versus a dependency graph distribution used by computer programmers. The results are only preliminary, but they suggest that a common design-based “dependency graph” model fit the genetic data from these species some 103000 times better than a Darwinian evolutionary tree. Common design is a potentially superior explanation compared to universal common ancestry for the distribution of much genomic data.