Cast your mind back 543 million years, to a warm, watery place that will one day be Wales. Imagine yourself as one of the inhabitants of that water world. You are not a big creature - but then, nothing is. Like every other organism you have ever come across in your drifting, you are just a single cell. As far as you're concerned, the single-celled life is the only sort of life that it is possible to lead. All the living things in your world consist of a set of genes, the proteins those genes de-scribe, the chemicals that those proteins can put together and pull apart, and a membrane that wraps the whole shebang up in a neat bundle. The history of life to date has been a matter of refining the make-up of such bundles. After two billion years of this tinkering, you might be for-given for thinking that you, and other single cells like you, are the acme of creation: that you are as good as life gets.
And then your sun is suddenly blotted out. Floating above you is a creature millions of times bigger than you are - a Battle-star Galactica swimming through the water. It is, by your standards, inconceivably vast. Yet on close inspection it turns out to be made up of little cells: bundles of genes and proteins and other stuff just like you. The difference is that there are billions of them acting together. And on top of that they come in a bewildering array of shapes and sizes. Lots of cells; lots of types of cells; all working together. This is life as you haven't known it. This is about to take over your planet. This is the future.
Now imagine yourself as a little computer sitting on a desk a couple of years ago - something more familiar, if far less complex, than an archaic single cell. Like all computers, you consist of a collection of chips, the programs that animate those chips and the information that they generate and use, all wrapped in a sheet of plastic. And you, like an archaic single cell, are the product of a long process of tinkering. So you, too, might feel pretty good about yourself. You are like your ancestors, only better: you're faster, you've got more features. This may be as good as life gets.
Except for the fact that there is a vast new creature slowly becoming visible all around you. A creature that, looked at closely, turns out to be made up of lots of computers, lots of software, all interconnected. A creature that consists of many, many computers working in rough harmony. Some of the computers look strange - smaller than you, bigger than you, specialised for different tasks. Some look like televisions, or telephones, or toasters. Lots of computers, lots of types of computer, and all working together. This is life as you haven't known it. This is about to take over your planet. This is the future.
Firstest with the mostest
Evolution is the basic principle of life - almost its definition. That which lives, evolves. But though the presence of evolution is a constant, its rate can fluctuate wildly.Sometimes a new world of possibility opens up, and evolution explodes into frantic life. The new possibility 543 million years ago was multicellularity. Single cells learned how to develop into complex organisms - metazoans - and life exploded into a whole new realm of possibilities. Palaeontologists call this great blossoming the Cambrian explosion, in honour of the first rocks of that age to be discovered, which were found in Wales.
The blossoming of computer networks and the software that runs on them could be just as dramatic. It is not that the na-ture of information processing changes when, to coin a phrase, the network becomes the computer. The processing is still processing, just as the basics of metabolism are the same in a single-celled creature and the constituent cells of something grander. But just as it would be foolish to say that the life of a fruit bat was simply the life of all its individual cells, so it would be foolish to say that the power of a network is just the power of all its component computers.
When the metabolic underpinning of life is shared among many cells, or when information processing is distributed over many computers, something new emerges. It is that emergence which throws up new possibilities. Today only specialists trouble themselves to think of all the little cells that go to make up a plant or animal; most people appreciate the unique attributes of a creature rather than its shared components. On a similar basis, future computer-users will concern themselves little with the individual bits of hardware and software that make up a network. And they will expect far more from that network than any one machine could be expected to provide. I first came across this analogy between networks of cells and networks of computers through the work of Tom Ray, one of the pioneers of artificial life. Ray plans to create his own net-worked evolution with strictly supervised artificial life forms on to the Internet, but that's another story. For me the real point of Ray's work was to provide a glimpse into the future - across a chasm of revolutionary change. Today, the evolution of the network is a revolution, and the best guide into probable course of that revolution lies in the rocks which record life's own great period of experimentation in the Cambrian era.
The most obvious features of the Cambrian explosion, and those that can clearly be seen in the current explosion of networks, are speed and diversity. These two things have made the Cambrian period a puzzle to palaeontologists ever since Darwin's day. Much older rocks contain only microscopic fossils. Then suddenly, at the beginning of the Cambrian, the rocks start to overflow with life. There are creatures a millimetre across, a centimetre across, a metre across - creatures getting bigger with the same sort of alacrity that microcircuits get smaller. Creepies and crawlies suddenly start creeping, crawling, burrowing and swimming across the shallow seas. By geological standards, it all took place in the blink of an eye; after a couple of billion years of single-celled-ness, the whole incredible burgeoning of life took place in a few million years at most. And the new metazoans emerged in a startling variety of shapes and sizes. In his brilliantly titled book about the contingency of evolution, Wonderful Life, Stephen Jay Gould celebrates the immense diversity that sprang up in the Cambrian. The Cambrian explosion produced founding fathers for all the basic branches of the metazoans' subsequent family tree - not to mention a range of anatomical oddballs that look to us all but incomprehensible.
There was Wiwaxia, which looks like nothing more than a Belgian chocolate armed to fight Mad Max. There was Hallucigenia, so called because the geologist who first identified it thought it looked like one. It was so odd, with its spikes and tentacles, that at first no one could tell its arse from its elbow, quite literally: the spines now thought to go on top were originally imagined running along the belly. And then there was Anomalocaris, the Cambrian's largest predator. With its circular mouth and grasping protojaws, its long, rippled, ray-like body, Anomalocaris was so unlike any fossils seen before that the first researchers to look at it thought it a whole set of separate creatures that simply liked to hang out together. They mistook its mouth for a jellyfish, its jaws for shrimp.
Gould has been criticised - with some merit - for overstating the Cambrian's diversity. But such overstatement is like exaggerating height when discussing the Himalayas. Whatever you might say, the Himalayas are still pretty bloody tall - and in the Cambrian, evolution produced an amazingly diverse range of creatures compared to what had come before. Since ecology and evolution are just two different ways of looking at the relationship between creatures and their environment, this diversity shows the Cambrian to have been a period of new ecological richness. The diversity of form reflects a new diversity of behaviour and ways of life.
The speed of change and the diversity created are two sides of the same coin. Each drove the other. Already, network applications are following a similarly self-accelerating path of change. New network software breeds niches for yet more network software, speeding overall development. The Web supports Web crawlers, which in turn provide niches for bookmark managers, and so on. The more things networks do, the more interstices there are in which new things can be done. This sort of positive feedback process - typical of life itself and especially of its bursts of creativity - drives exponential growth in innovation.
Through the linked forces of speed and diversity, the parallels between the emergence of networks of cells and the emergence of networks of computers is reasonably plain. Where the metaphor becomes slipperier, however, is when one opens up the organisms and networks to see what happens to the parts as they create wholes greater than the sums of themselves. It turns out that networks are not like creatures; they are more like limits on what sorts of creatures can exist.
The landscape of life
Looking at the fossils, as Gould does, it is diversity without that catches the eye. But the diversity within is just as wonderful - and a necessary condition for everything else that happened. A metazoan is far more than a lot of self-sufficient single cells stuck together. Multicellular creatures embody the principle of division of labour. They have cells for passing messages - nerve cells - and for passing oxygen - blood cells. They have osteoblasts that make bone, and Langerhans cells that make insulin. They have cells of hundreds of different specialist types, all working together, linked in any number of ways - by juxtaposition, by hormones, by common descent, by neurotransmitters.Similarly, a network can contain all sorts of entities specialised for different tasks, but capable of working together for a common purpose. It can have clients and servers, PDAs and Crays. Household appliances can join in; so can telephones. A network has diversity - and it is arranged into patterns.
Here, though, things get tricky. The metazoans owe the division of labour - the differentiation of their cells, and the patterns of tissue and organ that such differentiation creates - to the process of development. The original single cell, the fertilised egg, divides, and divides again, and again. In a smooth unfolding of form, this growth produces the specific patterns of differentiation that make the egg into a man, or a mahseer, or a marmoset. All the cells are descendants of the first cell; they all share the same genes. But some use the genes that make them to become nerve cells, some become muscle, and so on. It is the pattern of activ- ity that changes, not the genetic complement.
So the development of form in a metazoan depends on how differentiation is governed during development; the process is historical, repeated with slight differences whenever the right set of genes turns up in the right sort of environment. A computer network, though, has no such historical process. Computer networks do not breed; computers do not copy themselves. They are engineered and arbitrary.
Whereas in a living creature the developmental process means that the links between the cells are largely preordained and difficult to alter, in a computer network the links can, in principle, be more or less whatever the users decide they should be. New programs, new software, can reconfigure the relationships between processors at a single stroke of the keyboard.
The ability to re-engineer networks makes them different from organisms. Because it can be changed so easily from outside, a network is more like a way of simulating any number of organisms that have roughly the same degree of complexity. The Internet - a network of networks - is not an organism; it is a possibility space for information organisms, a vast realm of potential mechanisms for information processing. When it expands rapidly, so too do its possibilities. You can look at the Cambrian explosion as just such an increase in possibilities, a rapid unfolding of new vistas in the landscape of life.
The Internet is not like a single creature, some vast confusing Anomalocaris; it is instead a way of thinking about all of the possible creatures that people might create - as well as a platform for creating them if need and opportunity should arise. The Internet is not like one of biology's living objects; it is instead like one of the biologists' intellectual constructs. It is like a fitness landscape: the theoretical space in which all possible ways of being a creature are arrayed.
In modern Darwinism, the hilly terrain of the fitness landscape is the battleground of evolution. This terrain is a surface that represents the possibilities of life; the ups in its endless unevenness correspond to more fitness, the downs to less. All the other dimensions (you can picture as many as your mind allows, but I tend to stick to two) correspond to the characteris- tics of possible creatures. So the peaks on the landscape correspond to creatures whose characteristics make them fitter than neighbouring - and thus similar - possibilities. Natural selection pushes all creatures towards such local peaks.
This may make the process sound short-lived. The landscape is created, the kings of the castles occupy its peaks, the dirty rascals languishing at lower levels of fitness die out: "end of stoareh," as Louis Armstrong once remarked.
But this story does not end so quickly - in fact it never ends. The landscape is not still; it shifts and shakes in tectonic hyperactivity. This makes the creatures keep moving - and the fact that they do so keeps the landscape in flux. Fitness, after all, depends largely on the competition; it is a relative, not an absolute, concept. The landscape is not independent of its inhabitants but is their con- stant creation.
The Internet is just such a landscape, such a space of possibilities. The contours of fitness, and therefore of success, are constantly changing within the Netspace which provides the boundaries, and the ground rules, of a burgeoning evolutionary struggle. Hopeful monsters
How were the expanded possibilities of the Cambrian ful-filled? To use the terms of Jim Valentine, an American palaeontologist, the answer seems to lie in loose genes. The creatures of the Cambrian appear to have had the ability to evolve much more radically than the creatures of today. If this ability allowed them to find the peaks in the great new fitness landscape, then it is a trick that those who create networks might like to study.
The Cambrian fauna that Gould chronicles show a re-markably loose approach to the niceties of anatomy. Modern creatures have a relatively small number of stylised body plans available - only a few basic ways of being a living thing. Acci-dents of history have eliminated huge realms of developmental possibility on the you-can't-get-there-from-here basis. Pigs don't have ancestors with wings, and no amount of mutation can create them.
But in the Cambrian, these limits had not yet been set. The unfolding of form as egg became creature seems to have followed simpler, looser genetic rules. The creatures of the Cambrian seem to have added and shed legs, body segments, mandibles, fin-like things and more or less any other sort of appendage with a rash abandon. They played with body parts and survival strategies with a wild and innocent freedom that their more complex descendants will never recapture.
Looked at in terms of the fitness landscape, this morphological mayhem has dramatic consequences. If change is incremental and subtle, then creatures will always evolve uphill towards the nearest scalable peak. If there is a peak of far greater fitness elsewhere, but reaching it would require a preliminary trip down into a valley, then that peak is effectively unreachable. Natural selection makes no exceptions on the basis of some possible long- term pay off; it punishes the less fit in every generation. In the Cambrian, though, loose genes may have made it possible for creatures to leap across the fitness landscape where today's creatures can only shuffle. Those evolutionary theorists who believe such things possible call such expeditionaries "hopeful monsters".
Most of their hopeful leaps across the Cambrian fitness landscape must have failed; most of the monsters will have vanished into local crevasses and sinkholes of unfitness. But some, by blind luck, must have leapt from peak to higher peak. And each successful leap would have had its effects elsewhere throughout the landscape, as the creatures constantly adapted to new ecological conditions.
This is the model that creators of new network software - new possibilities in the Netspace - should always keep in mind. They have even looser genes than those of the Cambrian at their disposal - genes so loose they barely deserve the name. They can rewrite these genes whenever they want to, if they can then get those new bits of programming arranged around the network the right way. Their only constraint is getting it to work - and getting people to believe that it does work.
That might seem unmitigatedly good news. In fact, maximising looseness may not be the best policy; the loosest genes may collapse embarrassingly around the ankles. The secret of Cambrian success appears to have been the fact that loose control over body plan was matched by quite tight local controls. Loose master genes changed overall plans - but tighter local controls helped make sure that legs were still legs, gills still gills.
For this reason, the creatures of the Cambrian built new bodies from a mix-and-match collection of ready-made parts. But, while wringing changes on a small number of widely accepted sub-systems seems to have worked spectacularly, concentrating on the tried-and- true rules out whole areas of the fitness landscape. With more creatures competing for the limited space available, the landscape effectively becomes more jagged, changing from rolling hills to daunting cliffs and pinnacles.
Something like this happened in the Cambrian. Competition must have become harder as time went on, and diversity was pruned. And as life became more complex, mutations in the basic developmental program of the sort that produce hopeful monsters became that much more more difficult. Organisms became captive to the complexity of their developmental histories, and the explosion of diversity ended. Many creatures and types of creature became extinct, closing off realms of possibility that only their five-eyed, three-mouthed descendants could have filled. The descendants of the survivors produced a world filled with magnificently different variations on limited themes; boa constrictors and boobies, whales and wombats.
In the Netspace, that need not happen. Directed evolution will ensure that peaks of fitness are sharper, and as the complexity of the systems in the network increase, their diversity may begin to drop off. But our ability to spot completely new niches will remain - and with it our ability, as creators of net creatures, to guide a hopeful leap from one peak to another. While competition in providing those services that are already identified will be harsh, we'll always have the possibility of finding something monstrously new, and prospering from it. The limitations of the metazoan genome limited that freedom in the Cambrian - but in the Internet and its successors, no such limitations need hold. Playing god does have practical advantages.
The way that life works - the rules that control metazoan growth - meant that the Cambrian explosion had to come to an end. The way that software works, and foresight of its creators, mean that the network revolution may not have to. On the Net, there will always be hope for monsters.
Oliver Morton (oem@wired.co.uk) is the editor of Wired.