in the actual range of environments they encounter.
One of the most perceptive summaries of Darwin’s argument was made by one of his fiercest critics. A man named Robert Mackenzie Beverley, writing in 1867, produced what he thought was a devastating demolition of the idea of natural selection. Absolute ignorance is the artificer, he pointed out, trying to take the place of absolute wisdom in creating the world. Or (and here Beverley’s fury drove him into capital letters), ‘IN ORDER TO MAKE A PERFECT AND BEAUTIFUL MACHINE, IT IS NOT REQUISITE TO KNOW HOW TO MAKE IT.’ To which Daniel Dennett, who is fond of this quotation, replies: yes, indeed! That is the essence of Darwin’s idea: that beautiful and intricate organisms can be made without anybody knowing how to make them. A century later, an economist named Leonard Reed in an essay called ‘I, Pencil’, made the point that this is also true of technology. It is indeed the case that in order to make a perfect and beautiful machine, it is not requisite to know how to make it. Among the myriad people who contribute to the manufacture of a simple pencil, from graphite miners and lumberjacks to assembly-line workers and managers, not to mention those who grow the coffee that each of these drinks, there is not one person who knows how to make a pencil from scratch. The knowledge is held in the cloud, between brains, rather than in any individual head. This is one of the reasons, I shall argue in a later chapter, that technology evolves too.
Charles Darwin’s dangerous idea was to take away the notion of intentional design from biology altogether and replace it with a mechanism that builds ‘organized complexity … out of primeval simplicity’ (in Richard Dawkins’s words). Structure and function emerge bit by incremental bit and without resort to a goal of any kind. It’s ‘a process that was as patient as it was mindless’ (Dennett). No creature ever set out mentally intending to see, yet the eye emerged as a means by which animals could see. There is indeed an adapted purposefulness in nature – it makes good sense to say that eyes have a function – but we simply lack the language to describe function that emerged from a backward-looking process, rather than a goal-directed, forward-looking, mind-first one. Eyes evolved, Darwin said, because in the past simple eyes that provided a bit of vision helped the survival and reproduction of their possessors, not because there was some intention on the part of somebody to achieve vision. All our functional phrases are top–down ones. The eye is ‘for seeing’, eyes are there ‘so that’ we can see, seeing is to eyes as typing is to keyboards. The language and its metaphors still imply skyhooks.
Darwin confessed that the evolution of the eye was indeed a hard problem. In 1860 he wrote to the American botanist Asa Gray: ‘The eye to this day gives me a cold shudder, but when I think of the fine known gradation my reason tells me I ought to conquer the odd shudder.’ In 1871 in his Descent of Man, he wrote: ‘To suppose that the eye with all its inimitable contrivances for adjusting the focus to different distances, for admitting different amounts of light, and for the correction of spherical and chromatic aberration, could have been formed by natural selection, seems, I freely confess, absurd in the highest degree.’
But he then went on to set out how he justified the absurdity. First, the same could have been said of Copernicus. Common sense said the world stood still while the sun turned round it. Then he laid out how an eye could have emerged from nothing, step by step. He invoked ‘numerous gradations’ from a simple and imperfect eye to a complex one, ‘each grade being useful to its possessor’. If such grades could be found among living animals, and they could, then there was no reason to reject natural selection, ‘though insuperable by our imagination’. He had said something similar twenty-seven years before in his first, unpublished essay on natural selection: that the eye ‘may possibly have been acquired by gradual selection of slight but in each case useful deviations’. To which his sceptical wife Emma had replied, in the margin: ‘A great assumption’.
Pax optica
This is exactly what happened, we now know. Each grade was indeed useful to its possessor, because each grade still exists and still is useful to its owner. Each type of eye is just a slight improvement on the one before. A light-sensitive patch on the skin enables a limpet to tell which way is up; a light-sensitive cup enables a species called a slit-shelled mollusc to tell which direction the light is coming from; a pinhole chamber of light-sensitive cells enables the nautilus to focus a simple image of the world in good light; a simple lensed eye enables a murex snail to form an image even in low light; and an adjustable lens with an iris to control the aperture enables an octopus to perceive the world in glorious detail (the invention of the lens is easily explained, because any transparent tissue in the eye would have acted as partial refractor). Thus even just within the molluscs, every stage of the eye still exists, useful to each owner. How easy then to imagine each stage having existed in the ancestors of the octopus.
Richard Dawkins compares the progression through these grades to climbing a mountain (Mount Improbable) and at no point encountering a slope too steep to surmount. Mountains must be climbed from the bottom up. He shows that there are numerous such mountains – different kinds of eyes in different kinds of animal, from the compound eyes of insects to the multiple and peculiar eyes of spiders – each with a distinct range of partially developed stages showing how one can go step by step. Computer models confirm that there is nothing to suggest any of the stages would confer a disadvantage.
Moreover, the digitisation of biology since the discovery of DNA provides direct and unambiguous evidence of gradual evolution by the progressive alteration of the sequence of letters in genes. We now know that the very same gene, called Pax6, triggers the development of both the compound eye of insects and the simple eye of human beings. The two kinds of eye were inherited from a common ancestor. A version of a Pax gene also directs the development of simple eyes in jellyfish. The ‘opsin’ protein molecules that react to light in the eye can be traced back to the common ancestor of all animals except sponges. Around 700 million years ago, the gene for opsin was duplicated twice to give the three kinds of light-sensitive molecules we possess today. Thus every stage in the evolution of eyes, from the development of light-sensitive molecules to the emergence of lenses and colour vision, can be read directly from the language of the genes. Never has a hard problem in science been so comprehensively and emphatically solved as Darwin’s eye dilemma. Shudder no more, Charles.
Astronomical improbability?
The evidence for gradual, undirected emergence of the opsin molecule by the stepwise alteration of the digital DNA language is strong. But there remains a mathematical objection. The opsin molecule is composed of hundreds of amino acids in a sequence specified by the appropriate gene. If one were to arrive at the appropriate sequence to give opsin its light-detecting properties by trial and error it would take either a very long time or a very large laboratory. Given that there are twenty types of amino acid, then a protein molecule with a hundred amino acids in its chain can exist in 10 to the power of 130 different sequences. That’s a number far greater than the number of atoms in the universe, and far greater than the number of nanoseconds since the Big Bang. So it’s just not possible for natural selection, however many organisms it has to play with for however long, to arrive at a design for an opsin molecule from scratch. And an opsin is just one of tens of thousands of proteins in the body.
Am I heading for a Lucretian swerve? Will I be forced to concede that the combinatorial vastness of the library of possible proteins makes it impossible for evolution to find ones that work? Far from it. We know that human innovation rarely designs things from scratch, but jumps from one technology to the ‘adjacent possible’ technology, recombining existing features. So it is taking small, incremental steps. And we know that the same is true of natural selection. So the mathematics is misleading. In a commonly used analogy, you are not assembling a Boeing 747 with a whirlwind in a scrapyard, you are adding one last rivet to an existing design. And here there has been a remarkable recent discovery that makes natural selection’s task much easier.
In a laboratory in Zürich a few years ago, Andreas Wagner asked his student João Rodriguez to
