We have arrived back at the conclusion of the previous chapter. An elephant is a huge digression within a computer program written in DNA language. An ostrich is another kind of digression, an oak tree is another. And, of course, a human is another. We are all TRIP robots, all von Neumann machines. But how did the whole process start? To answer that, we have to go back a very long time, more than 3,000 million years, probably as long as 4,000 million years. In those days the world was very different. There was no life, no biology, only physics and chemistry, and the details of the Earth's chemistry were very different. Most, though not all, of the informed speculation begins in what has been called the primeval soup, a weak broth of simple organic chemicals in the sea. Nobody knows how it happened {282} but, somehow, without violating the laws of physics and chemistry, a molecule arose that just happened to have the property of self-copying — a replicator.
This may seem like a big stroke of luck. I want to say a few things about this luck’. First, it had to happen only once. In this respect, it is rather like the luck involved in colonizing an island. Most islands around the world, even quite remote ones like Ascension Island, have animals. Some of these, for example birds and bats, got there in a way that we can easily understand, without postulating a great deal of luck. But other animals, like lizards, can't fly. We scratch our heads and wonder how they got there. It may seem unsatisfactory to postulate a freak of luck, like a lizard happening to be clinging to a mangrove on the mainland which breaks off and drifts across the sea. Freakish or not, this kind of luck does happen — there are lizards on oceanic islands. We usually don't know the details, because it is not a thing that happens often enough for us to have any likelihood of seeing it. The point is that it had to happen only once. And the same goes for the origin of life on a planet.
What is more, as far as we know, it may have happened on only one planet out of a billion billion planets in the universe. Of course many people think that it actually happened on lots and lots of planets, but we only have evidence that it happened on one planet, after a lapse of half a billion to a billion years. So the sort of lucky event we are looking at could be so wildly improbable that the chances of its happening, somewhere in the universe, could be as low as one in a billion billion billion in any one year. If it did happen on only one planet, anywhere in the universe, that planet has to be our planet — because here we are talking about it.
My guess is that life probably isn't all that rare and the origin of life probably wasn't all that improbable. But there are arguments to the contrary. One interesting example is the ‘Where are they?’ argument. Imagine a South Pacific race whose island is so remote that in all the oral history of the tribe no canoe has ever found another piece of inhabited land. The tribal elders speculate as to the likelihood that there is life outside the island. The ‘we are alone’ faction has a powerful argument in the fact that the island has never been visited. Even if {283} the tribe s travelling is limited to canoe-range, shouldn't there be other tribes who have progressed to more advanced boats? Why have they never come?
In the case of inhabited islands on Earth, they have all been visited by now, and today there must be few people so remote that they have not seen an aeroplane. But our island planet in the universe has never, as far as we know from properly authenticated accounts, been visited. More significantly, for the last few decades we have been equipped to detect radio communications from far away. There are about a million stars within the radius that radio waves could reach in a thousand years. A thousand years is a short time by the standards of stars and geology. If technological civilizations are common, some of them will have been pumping out radio waves for thousands of years longer than we have. Shouldn't we have heard some whisper of their existence by now? This is not an argument against life of any kind existing elsewhere in the universe. But it is an argument against intelligent, technically sophisticated life being spaced densely enough to be within easy radio range of other islands of life. If life when it starts has anything other than a low probability of giving rise to intelligent life, we might take this as evidence that life itself is rare. An alternative conclusion to this chain of reasoning is the bleak proposal that intelligent life may arise quite frequently, but typically only a short time elapses between the invention of radio and technological self-destruction.
Life may be common in the universe, but we are also at liberty to speculate that it is exceedingly rare. It therefore follows that the kind of event we are seeking, when we speculate about the origin of life, could be a very very improbable event: not the kind of event that we can expect to duplicate in the laboratory and not the kind of event that a chemist will deem ‘plausible’. This is an interesting paradox, spelled out in full in a chapter of The Blind Watchmaker called ‘Origins and Miracles’. We could be actively seeking a theory with the specific property that, when we find it, we shall judge it highly implausible! Looking at the matter in one way, we might even be positively worried if a chemist manages to support a theory of the origin of life which, using ordinary standards of probability, we judge to be plausible. {284} On the other hand life seems to have arisen during die first half billion of the Earths 4.5 billion years; we've been here for eight parts in nine of the Earths age and my intuition is still that the arising of life on a planet is not all that unexpected an event.
An origin of life, anywhere, consists of the chance arising of a self-replicating entity. Nowadays, the replicator that matters on Earth is the DNA molecule, but the original replicator probably was not DNA. We don't know what it was. Unlike DNA, the original replicating molecules cannot have relied upon complicated machinery to duplicate them. Although, in some sense, they must have been equivalent to ‘Duplicate me’ instructions, the language’ in which the instructions were written was not a highly formalized language. such that only a complicated machine could obey them. The original replicator cannot have needed elaborate decoding, as DNA instructions and computer viruses do today. Self-duplication was an inherent property of the entity's structure just as, say, hardness is an inherent property of a diamond, something that does not have to be ‘decoded’ and ‘obeyed’. We can be sure that the original replicators, unlike their later successors the DNA molecules, did not have complicated decoding and instruction-obeying machinery, because complicated machinery is the kind of thing that arises in the world only after many generations of evolution. And evolution does not get started until there are replicators. In the teeth of the so-called ‘Catch-22 of the origin of life’ (see below), the original self-duplicating entities must have been simple enough to arise by the spontaneous accidents of chemistry.
Once the first spontaneous replicators existed, evolution could proceed apace. It is in the nature of a replicator that it generates a population of copies of itself, and that means a population of entities that also undergo duplication. Hence the population will tend to grow exponentially until checked by competition for resources or raw materials. I'll develop the idea of exponential growth in a moment. Briefly, the population doubles at regular intervals, rather than just adding a constant number at regular intervals. This means that there will soon be a very large population of replicators and hence competition between them. It is in the nature of any copying process that it is never quite perfect: there are random errors in duplication. {285} Therefore there would arise varieties of the replicator in the population. Some of these variants would have lost the property of self-duplication and their particular form would not have been retained in the population. Other variants happened to have some property that caused them to be duplicated more rapidly or more efficiently. They became consequently more numerous in the population. Since they would have been competing for the same raw materials as rival replicators, as time went by the average, typical replicator type in the population would continually have been supplanted by a new and a better average type. Better at what? Better at replicating, of course. Later, this improvement would take the form of influencing other chemical react
ions so as to facilitate self-replication. Eventually, the influence would have become sufficiently complicated that an observer, had there been one (there wasn't, of course, for it takes billions of years to evolve anything that you could call an observer), might have described the process as the decoding and obeying of instructions. And if that same observer were asked what the instructions mean, he would have to reply that they mean ‘Duplicate me’.
There are undoubted difficulties in this story. Among them I have already alluded to the so-called Catch-22 of the origin of life. The larger the number of components in a replicator, the more likely it is that one of them will be miscopied, leading to complete malfunctioning of the ensemble. This suggests that the first, primordial replicators must have had very few components. But molecules with fewer than a certain minimum number of components are likely to be too simple to be capable of engineering their own duplication. Ingenuity has been expended on reconciling these two apparently incompatible requirements — with some success, but the argument becomes more mathematical than is suitable for this book.
The original replication machines — the first robot repeaters — must have been a lot simpler than bacteria, but bacteria are the simplest examples of TRIP robots that we know today (Figure 9.3a). Bacteria make their livings in a great variety of ways, from a chemical point of view a far wider range of ways than the rest of the living kingdoms put together. There are bacteria that are more closely related to us than they are to other, strange kinds of bacteria. There are {286} bacteria that obtain their sustenance from sulphur in hot springs, for whom oxygen is a deadly poison, bacteria that ferment sugar to alcohol in the absence of oxygen, bacteria that live on carbon dioxide and hydrogen, giving out methane, bacteria that photosynthesize (use sunlight to synthesize food) like plants, bacteria that photosynthesize in ways that are very different from plants. Different groups of bacteria encompass a range of radically different biochemistries compared with which all the rest of us — animals, plants, fungi and some bacteria — are monotonously uniform.
Bacteria of several different kinds got together, more than a thousand million years ago, to form the ‘eucaryotic cell’ (Figure 9.3b). This is our kind of cell, with a nucleus and other complicated internal parts, many of them put together from intricately folded internal membranes, like the mitochondria which I briefly pointed to in Figure 5.2. The eucaryotic cell is now seen as derived from a colony of bacteria. Eucaryotic cells themselves later got together into colonies. Volvox are modern creatures (Figure 9.3c). But it is possible that they represent the kind of thing that went on more than a thousand million years ago, when our kind of cells first started to band together into colonies. This ganging up of eucaryotic cells was comparable to the earlier ganging up of bacteria into eucaryotic cells and the even earlier ganging up of genes into bacteria. Larger and more densely packed gangs of eucaryotic cells are called metazoan bodies. Figure 9.3d shows a comparatively small one, a tardigrade. Metazoan bodies themselves sometimes gang up into colonies that themselves behave somewhat like individuals (Figure 9.3e).
I said that an elephant was a huge digression on a Duplicate Me program, but I could have said mouse instead of elephant and huge would still have been the right word. A volvox has a few hundred cells. A mouse is a large edifice of perhaps a billion cells. An elephant is a colony of about 1,000 trillion (1015) cells, and each one of those cells is itself a colony of bacteria. If an elephant is a robot carrying its blueprint about, it is an almost unthinkably large robot. It is a colony of cells but, since those cells carry copies of the same DNA instructions, they all cooperate, working together towards the same end of duplicating their separately identical DNA data. {287}
Figure 9.3 Increasing levels of organization among life forms: (a) individual bateria; (b) advanced — eucaryotic — cell with nucleus, originally evolved from a colony of bacteria; (c) volvox, a colony of differentiated eucaryotic cells; (d) a more densly packed and populous colony of differentiated eucaryotic cells, a tardigrade. A human body is another such colony — a colony of colonies, since each of our cells is a colony of bacteria; (e) a colony of individual organisms: a swarm of honey bees — a colony of colonies of colonies. {289}
Of course an elephant isn't a particularly large thing, on any absolute scale. Compared with a star it is small. I meant large by comparison with the DNA molecules that the elephant is designed to preserve and propagate. It is large compared with the replicating elephant-makers that ride around inside it.
To get an idea of scale, imagine that human engineers built a giant mechanical robot in which they could ride, like the Greeks in their Trojan Horse. But our mechanical horse will be scaled up so that each human engineer is equivalent to one of the robot's DNA molecules in size. Remember that we are thinking of a real horse as a robot built by the genes that ride inside it. The point of the picture is that, if we built a robot horse for ourselves to ride inside, and if our robot horse were as big, relative to us, as a real horse is big, relative to the genes that built it, then our robot horse could bestride the Himalayas (Figure 9.4). A real, living horse is made of trillions of cells. With unimportant exceptions, every one of those cells has a full crew of genes
Figure 9.4 A horse is a robot vehicle for DNA molecules and it is very large in comparison to them. If humans built a Trojan Horse to ride inside, to a similar scale relative to ourselves, it would dwarf the Himalayas. This fantasy was painted by my mother, Jean Dawkins, for one of my Royal Institution Christmas Lectures. {290}
riding inside it although most of them, in any one kind of cell, are sleeping.
A real living body manages to be so big (compared with the genes that built it) because it grows by a very different process from the way a man-made machine grows; quite different from the way this mechanical horse would be built, if it ever were built. The special way of growing that real living things employ is exponential growth. Another way of saying it is that living things grow by local doubling.
We start with a single cell which is very small. Or rather, it is just about the right size for the genes that make it. It is within the range that they can cope with by biochemical manipulation. Their tendrils of influence can reach all corners of a single cell, and they can fashion that single cell to have certain properties. Perhaps the most remarkable property the cell has is the ability to divide into two daughter cells more or less like itself. Being like the parent cell, each daughter cell itself is capable of dividing into two, making four granddaughter cells. Each one of the four, in turn, can double, making eight, and so on. This is exponential growth, or local doubling.
People who are not used to it find the power of exponential growth surprising. As promised, I'll spend a little time on it, because it is important. There are many vivid ways of illustrating it. If you fold a piece of paper once, you have two thicknesses. Fold it again and it is four times as thick. Another fold and you have a wad, eight layers thick. Three more folds is about as many as you can get away with, before the wad becomes too stiff to fold further: sixty-four layers thick. But suppose that this mechanical stiffness were not a problem and that you could go on folding, say fifty times. How thick would the wad of paper be then? The answer is that it would be so thick that it would reach right outside the Earths atmosphere and beyond the orbit of the planet Mars.
In the same way, by local doubling of cells all over the developing body, the number of cells very rapidly gets up into the astronomical range. A blue whale is made of about a hundred thousand trillion (1017) cells. But, such is the power of exponential growth, it would only take about fifty-seven cell generations, under ideal conditions, to produce such a leviathan. By a cell generation, I mean a doubling. {291} Remember that numbers of cells go up 1, 2, 4, 8, 16, 32, etc. So it takes six cell generations to reach thirty-two cells. And, if you go on multiplying by two like that, it takes only fifty-seven cell generations to reach a hundred thousand trillion, the number of cells in a blue whale.
This way of calculating the number
of cell generations is actually unrealistic because it gives only a minimum figure. It assumes that, after every cell generation, all the cells go on to duplicate. In fact, many cell lineages drop out of the doubling game earlier on, when they have finished building a particular part of the body, say the liver. Other cell lineages go on doubling for longer. So a blue whale in fact consists of a number of cell lineages of different length, building different parts of the whale. Some of these lineages go on dividing for more than fifty-seven cell generations. Others stop dividing after fewer than fifty-seven cell generations. In practice there are ‘stem-cells’, subsets of cells that are set aside for the purpose of running off copies of cells like themselves.
You can roughly calculate the minimum number of cell generations it would take, under ideal doubling conditions, to grow any animal, given its weight. You can assume that big animals don't have especially big cells, they just have more of the same kinds of cells as small animals. A naive calculation suggests that it would take a minimum of forty-seven cell-doubling generations to grow an adult human and only about ten more cell generations to grow a blue whale. These figures are certainly underestimates, for the reason I have given. Nevertheless it remains true that, such is the power of exponential growth, you need only make a small change in how long a particular lineage of cells goes on dividing, to get a dramatic change in the final size of the bunch of cells produced. Mutations sometimes do this.
Building these colossal bodies — colossal by the standards of their DNA builders and passengers — can be called gigatechnology. Gigatechnology means the art of building things at least a billion times bigger than you are. The art of gigatechnology is something that our own engineers have no experience of. The biggest vehicles that we build to travel about in — large ships — are not very large {292} relative to their builders and we can walk right round them in a matter of minutes. When we build something like a ship, we do not have the advantage of exponential building. For us, there is nothing for it but to swarm all over the structure, riveting hundreds of prefabricated steel plates together.