The “Drunkard’s Walk” Theory of Complexity

In his latest book, Full House, (1996), Stephen Jay Gould posits what he characterizes as a “drunkard’s walk” model to account for the evolution of complexity. This is a rather surprising argument, coming from such a sophisticated and articulate student of evolution.

Gould finds “almost chilling” E.O. Wilson’s testimonial about the functional improvements that have been associated with the evolution of complexity in nature. “We grasp at the straw of progress (a dessicated ideological twig) because we are still not ready for the Darwinian revolution” (p.29). As in some of his other recent writings, Gould wishes to deny the relevance of any notion of general “improvement” as a significant aspect of the evolutionary process. He seeks to undermine the traditional paleontological conceit that there has been “progress” (culminating of course in humankind), or that the so-called “trends” in evolution imply that something has gotten (normatively) better.

Gould has a point, of course, but objections to the notion of any “driven” trend toward betterment or progress (orthogenesis) go back to Darwin himself. Perhaps Gould is speaking to the “structuralists” — the proponents of evolution as a self-organized, self-propelled process. This is a minority view that currently appeals mainly to certain biophysicists, for obvious reasons. What is new — Gould’s conceit if you will — is the assertion that the evolutionary process is essentially random in its overall course and that the simplest forms of life — bacteria — represent the modal trend.

For instance, Gould attacks Cope’s Rule (after the 19th century paleontologist Edward Drinker Cope), which holds that most lineages tend to increase in size over time. This is an artefact, he says, of the circumstance that life began at the extreme “left wall” (or tail) in the distribution of biological size and complexity. Therefore, any subsequent directional change could only be toward the “right wall” (greater complexity). Furthermore, complexity in living systems is not a product of natural selection; organisms merely “wander” into complexity (like a “drunkard’s walk”).

There is a kind of “just-so” quality to this notion (to borrow a metaphor from Gould). It implies that systematic size/complexity increases in nature could occur without being “tested” and winnowed by natural selection. On the contrary, any such changes always entail bioeconomic “costs” (energy, for instance) that have to be offset by at least equivalent “benefits.” (There are no free lunches in nature.)

So intent is Gould on making his anti-progressive case that he even allows himself to fall into a logical trap — a rare event for such a stellar thinker. In light of some recent, suggestive work by various colleagues, Gould claims that “a small overall tendency toward decreasing complexity may characterize the history of most lineages” (p.200). This is absurd on the face of it. If life began in extreme simplicity, it had to get more complex before it could become less complex. So, if there is any residue of complexity left in nature, the overall trend at the margin had to be in that direction (unless there is some statistical sleight of hand going on).

Indeed, in the process of unpacking his most important (and eccentric) example — the disappearance of .400 hitters in baseball — Gould inadvertently undermines his own argument. Gould’s key point is that, despite appearances, baseball hitters have not gotten worse over the years. This trend is a result (in his words) of an “improvement in general play” — i.e., pitching and fielding. What’s that? Did he say “improvement” — a “progressive” advance in functional performance rather than a “drunkard’s walk”? So it seems there is a role for functionally-based changes after all, at least in the evolution of baseball.

Category: Publications