Contra Gelernter on Darwin

David Gelernter recently wrote an essay on Giving Up Darwin that is not obviously stupid. Dr. Gelernter, in many ways an astute thinker, does not commit obvious stupidities – but I have had to call him out before for allowing himself to be blinded by a hunger for epistemic gaps that fit the shape of religion. Apparently it is, alas, time to do that again.

The central argument of Gelernter’s essay is that random chance is not good enough, even at geologic timescales, to produce the ratchet of escalating complexity we see when we look at living organisms and the fossil record. Most mutations are deleterious and degrade the functioning of the organism; few are useful enough to build on. There hasn’t been enough time for the results we see.

Before getting to that one I want to deal with a subsidiary argument in the essay, that Darwinism is somehow falsified because we don’t observe the the slow and uniform evolution that Darwin posited. But we have actually observed evolution (all the way up to speciation) in bacteria and other organisms with rapid lifespans, and we know the answer to this one.

The rate of evolutionary change varies; it increases when environmental changes increase selective pressures on a species and decreases when their environment is stable. You can watch this happen in a Petri dish, even trigger episodes of rapid evolution in bacteria by introducing novel environmental stressors.

Rate of evolution can also increase when a species enters a new, unexploited environment and promptly radiates into subspecies all expressing slightly different modes of exploitation. Darwin himself spotted this happening among Galapagos finches. An excellent recent book, The 10,000 Year Explosion, observes the same acceleration in humans since the invention of agriculture.

Thus, when we observe punctuated equilibrium (long stretches of stable morphology in species punctuated by rapid changes that are hard to spot in the fossil record) we shouldn’t see this as the kind of ineffable mystery that Gelernter and other opponents of Darwinism want to make of it. Rather, it is a signal about the shape of variability in the adaptive environment – also punctuated.

Even huge punctuation marks like the Cambrian explosion, which Gelernter spends a lot of rhetorical energy trying to make into an insuperable puzzle, fall to this analysis. The fossil record is telling us that something happened at the dawn of the Cambrian that let loose a huge fan of possibilities; adaptive radiation, a period of rapid evolution, promptly followed just as it did for the Galapagos finches.

We don’t know what happened, exactly. It could have been something as simple as the oxygen level in seawater going up. Or maybe there was some key biological invention – better structural material for forming hard body parts with would be one obvious one. Both these things, or several other things, might have happened near enough together in time that the effects can’t be disentangled in the fossil record.

The real point here is that there is nothing special about the Cambrian explosion that demands mechanisms we haven’t observed (not just theorized about, but observed) on much faster timescales. It takes an ignotum per æque ignotum kind of mistake to erect a mystery here, and it’s difficult to imagine a thinker as bright as Dr. Gelernter falling into such a trap…unless he wants to.

But Dr. Gelernter makes an even more basic error when he says “The engine that powers Neo-Darwinian evolution is pure chance and lots of time.” That is wrong, or at any rate leaves out an important co-factor and leads to badly wrong intuitions about the scope of the problem and the timescale required to get the results we see. Down that road one ends up doing silly thought experiments like “How often would a hurricane assemble a 747 from a pile of parts?”

To get a better handle on the problem, it helps to ask the kind of question D’Arcy Thompson did in his monumental 1917 book “On Growth and Form”: why is a hen’s egg round?

The shape of an egg can be neatly described by a parametric equation in three variables, but neither that formula nor those parameters are encoded in the chicken genome. The chicken genome describes a relative simple production rule about the timed release of various egg-component chemicals; that rule doesn’t know anything about the spatial organization of the result.

What happens instead is a dance between the construction steps and the diffusion physics of the chemicals. The egg shape is supplied by the principle of least action. The chicken genome’s recipe captures – incorporates – this physics without actually coding it.

Thus, if you derange the egg-formation recipe with point mutations, the outcomes are limited by the physics. You may abort egg formation entirely, or you may get ellipsoids with differing sizes or shapes. What you won’t get is cubes or Klein bottles. Random variation in the egg-production genome doesn’t produce random variation in the shapes of eggs – it produces sharply constrained variation. The design space that mutations of the recipe are exploring is many orders of magnitude smaller and more continuous than you’d expect from a “pure chance” account.

Gelernter makes a similar mistake when he asks “Starting with 150 links of gibberish, what are the chances that we can mutate our way to a useful new shape of protein?” But this is never a question evolution has to answer. The nearest correct question would be “Starting from 150 links of a protein we know is already selected for usefulness because it’s already expressed in an organism, what are the chances we can mutate to something else useful?”

Again…the physics of van der Waals forces mean that a small change in coding for a protein is likely to produce a small change in its folding. As with eggs, point mutations are highly unlikely to jump a large distance in expressed phenotypic design. And – this is the point – they are thus unlikely to jump far away from a design that is productive for something.

The question Gelernter actually asked is a silly straw man that depends for its apparent force on the reader having no intuitions about the effects of a history of successful adaptation – or of the constraining role of extragenetic natural laws – at all.

Gelernter himself is definitely not stupid or ignorant enough to fall into this kind of error when he’s thinking clearly. From which we can only conclude that, on this subject, he refuses to think clearly.