Richard Dawkins was not included in Richard Dawkins' book: The Oxford Book of Modern Science Writing. The reason for the omission is obvious, so I rectify the "oversight" by including him in my list of good science writers.
I don't always agree with what Dawkins writes but there's no controversy about his ability to explain biology to the general public. He has a clear, crisp style that's easy to read and his arguments are well constructed. Part of his success is achieved by simplifying difficult concepts but this is also part of the problem since, in some cases, an over-simplification leads to misinterpretations.
Dawkins is also a master of metaphor but, sometimes the metaphors are misleading and can give an incorrect view of evolution (e.g. Climbing Mt. Improbable). I've chosen an excerpt from The Ancestor's Tale to illustrate Dawkins' skill at writing about science. This book is somewhat less polemical than his others, although it still has its fair share of strongly voiced personal opinions about evolution.
The passage below addresses "convergence," a favorite topic of theistic evolutionists such as Simon Conway Morris and Ken Miller. Dawkins has his own spin on the subject. He begins by addressing a question posed by Stuart Kauffman in 1985. Kauffman asked whether there are certain features of life that are easy to evolve. If so, we might expect these features to appear whenever life evolves. On the other hand ....
Those biologists who could be said to take their lead from the late Stephen Jay Gould regard all of evolution, including post-Cambrian evolution, as massively contingent—lucky, unlikely to be repeated in a Kauffman rerun. Calling it "rewinding the tape of evolution," Gould independently evolved Kauffman's thought experiment. The chance of anything remotely resembling humans on a second rerun is widely seen as vanishingly small, and Gould voiced it persuasively in Wonderful Life. It was this orthodoxy that led me to the cautious self-denying ordinance of my opening chapter; led me, indeed, to undertake my backwards pilgrimage, and now leads me to forsake my pilgrim companion at Canterbury and return alone. And yet ... I have long wondered whether the hectoring orthodoxy of contingency might have gone too far. My review of Gould's Full House (reprinted in The Devil's Chaplain) defended the unpopular notion of progress in evolution: not progress towards humanity—Darwin forfend!—but progress in directions that are at least predictable enough to justify the word. As I shall argue in a moment, the cumulative build-up of complex adaptations like eyes, strongly suggests a version of progress—especially when coupled in imagination with some of the wonderful products of convergent evolution.
Convergent evolution also inspired the Cambridge geologist Simon Conway Morris, whose provocative book Life's Solution: Inevitable Humans in a Lonely Universe presents exactly the opposite case to Gould's "contingency." Conway Morris means his subtitle in a sense which is not far from literal. He really thinks that a rerun of evolution would result in a second coming of man: or something extremely close to man. And, for such an unpopular thesis, he mounts a defiantly courageous case. The two witnesses he repeatedly calls are convergence and constraint.
Convergence we have met again and again in this book, including in this chapter. Similar problems call forth similar solutions, not just twice or three times but, in many cases, dozens of times. I thought I was pretty extreme in my enthusiasm for convergent evolution, but I have met my match in Conway Morris, who presents a stunning array of examples, many of which I had not met before. But whereas I usually explain convergence by invoking similar selection pressures, Conway Morris adds the testimony of his second witness, constraint. The materials of life, and the processes of embryonic development, allow only a limited range of solutions to a particular problem. Given any particular evolutionary starting situation, there is only a limited number of ways out of the box. So if two reruns of a Kauffman experiment encounter anything like similar selection pressures, developmental constraints will enhance the tendency to arrive at the same solution.
You can see how a skilled advocate could deploy these two witnesses in defence of the daring belief that a rerun of evolution would be positively likely to converge on a large-brained biped with two skilled hands, forward-pointing camera eyes and other human features. Unfortunately, it has only happened once on this planet, but I suppose there has to be a first time. I admit that I was impressed by Conway Morris's parallel case for the predictability of the evolution of insects.