The Science of Discworld IV Judgement Da

FOURTEEN



* * *



A BETTER MOUSETRAP





Rincewind has a soft spot for horses – which unfortunately leave soft spots for everyone else. Even so, he prefers horses to automobiles. You don’t have to make a horse – they make themselves from previous horses.

Cars are made individually, by people. They are designed to serve a purpose, which was present in the designer’s mind before the cars were made, and indeed caused it to be made. Without people, you could leave the Earth on its own for a billion years and it wouldn’t produce a car. But it did produce a horse, without human intervention, in a rather shorter period.

Scientists believe that horses evolved. The proof includes an iconic series of fossils, showing exactly how they evolved, between 54 million and a million years ago. The sequence begins with a horse-like mammal a mere 0.4 metres long. This genus was originally given the poetic name Eohippus (‘dawn horse’), but has now been renamed Hyracotherium because of the rules of taxonomy, which in this case managed to deliver a silly result.fn1 It moves on to Mesohippus, 35 million years old and 0.6 metres long; then Merychippus, 15 million years old and 1 metre long; then Pliohippus, 8 million years old and 1.3 metres long; and finally (so far) Equus, essentially the same as the modern horse, 1 million years old and 1.6 metres long.

Taxonomists can track, in great detail, the sequence of changes that occurred in this lineage of ancient horse ancestors; for example, in the animal’s teeth and hooves. They can also track the timing of these changes, because rocks can be dated. So now evidence from geology can be thrown into the mix. It would take only one fossil species in the wrong stratum of rock to cast doubt on the evolutionary story. The succession of rocks, their ages as determined by a variety of different methods, the evolutionary sequences of fossils, and the DNA of horses and modern relatives – all agree, to a remarkable extent.

There is similar evidence that humans evolved from ape-like ancestors, but the story is not as neat and tidy, with many potentially coexisting species. These ancestral hominins evolved from other mammals, which evolved from reptiles, which evolved from amphibians, which evolved from fish.fn2 Rincewind knows how land animals evolved: he was there. Roundworld’s inhabitants weren’t, which is one reason why they argue about what happened.

Both William Paley, in his Natural Theology of 1802, and modern creationists, believe that horses and humans were designed by God, and that these creatures’ modern forms are exactly those given to them at the moment of their creation. The hypothesis of intelligent design attempts to infer the existence of an unspecified cosmic designer (we all know who, but it would be unscientific to say so …) from the existence of complex structures in living organisms. Darwin argued that design in this sense is neither necessary, nor plausible: instead, living creatures evolved. Almost all biologists agree. Neo-Darwinism underpins those ideas with genes.

Evolved, or designed?

Maybe the difference isn’t as great as most of us think.

When design is presented as an alternative to evolution, there is an unstated assumption that the two are very different. Design is a conscious process carried out by a designer who knows what result he, she, or it is aiming at, and whose purpose is to achieve it. Evolution selects, from a lot of random variants, changes that lead to some kind of improvement in survival prospects; then it makes lots of copies of the successes. It has no aims and no purpose. It is not ‘blind chance’, a description that creationists often use, forgetting (we are being charitable) the crucial element of selection. But the process of evolution is exploratory, not goal-seeking.

On closer examination, however, design and evolution are much more similar than most of us imagine. Technology appears to be designed, but mostly it evolves. Improved technology is selected because it works better, and it then displaces earlier technology. This process is analogous to the way that natural selection causes organisms to evolve, so it is reasonable to speak of technology evolving. (The analogy is loose and should not be taken to extremes. Technical drawings or CAD designs are poor analogues of genes.) Selection of technology may appear to be by human agency, but it is highly constrained. Success is decided by vote, and voters vote with their wallets. The inventor’s intentions are almost irrelevant. Just as in biological evolution, the main constraint is what works.

Because of the difficulties inherent in the simple-minded approach to design that Paley proposed – because designers don’t go straight from idea to design to product – we should look carefully at just how designs appear in technology. That changes our attitude to ‘design’ in nature too.

Most human designs don’t work the first time round. Most jugs still don’t pour well. It’s cheaper to invent a new type of jug, even if it’s no good, than it is to pay licensing fees for one that works. The better mousetrap, even when it is genuinely better, is a minor variation on hundreds of previous mousetraps. Usually.

Mousetrap evolution is a coherent process, not just a succession of unrelated gadgets. The same goes for bicycles, cars, computers, even jugs. Each new capability causes a particular technological path to branch, leading to new roads. Stuart Kauffman, one of the founders of complexity science, introduced the term ‘the adjacent possible’ to mean the possible behaviours of a complex system that are just a short step away from wherever it currently is. The adjacent possible is a list of what potentially might develop. In a sense, it is the system’s potential.

Organic evolution proceeds by invading the adjacent possible. Invasions that fail aren’t invasions at all, and nothing much changes. Successful invasions don’t just change the system that does the invading; they change the adjacent possible of everything. When insects first took to the air, the ones that stayed on the ground were suddenly in danger of predation from above, even though they hadn’t changed at all. Likewise, technology advances by continually invading the adjacent possible. Technological evolution is faster than organic evolution because human minds can use their imaginations to jump into the adjacent possible and see if it works, without actually doing it. They can also copy, which organic evolution does only rarely, aside from reproducing near copies of organisms. These are processes that generate paths and histories, and contexts in which some evolutionary trajectories are viable, but others are not. Only a few select trajectories work. In contrast, thinking in terms of innovations that generate products makes the design process work like magic.

There are a few useful analogies between technological evolution and organic evolution, and a lot of misleading ones. Comparisons between organic evolution and economics abound in the literature, and nearly all of them are misleading, from social Darwinism to the ‘cost’ of reproduction. Some evolutionary trajectories, however, can usefully be compared to biological ones. Examples include telegraph → telephone, especially international with undersea cables as investment, pens → word processors, and rockets → space elevators, which we’ll come to shortly. These changes get rid of old constraints at each subsequent recursion.

There are biological precedents, in which evolution did not lead to increased complexity (as measured by DNA information), but the reverse. One is the evolution of mammals. Mammals have less DNA than their ‘more primitive’ amphibian ancestors, a trick that can be pulled off because mother mammals control the temperature of their developing embryos by keeping them inside their own bodies. Amphibians need huge quantities of genetic instructions to plan for many different contingencies, as their embryos grow in a pond, subject to the unpredictable vagaries of the weather. Mammals dispense with this excess baggage by investing in temperature control.

With the expanding possibilities of the chemical/physical universe as a substrate, and organic evolution as a model for an emergent phase space, we should be asking ‘What are the constraints on technology, if any?’ rather than ‘What is the pattern of technological advance?’ Sometimes there are persistent patterns. Moore’s Law states that computing power doubles every eighteen months. It has worked for decades, even though (indeed, because) technologies have changed dramatically. Some experts think that the increase in power will shortly have to slow down, but others remain convinced that new ideas, often already visible, will keep it going.

Our culture sometimes seems to follow evolutionary trajectories too. As individuals, we respond to the cultures in which we live, and we are conducted into our technological future as it changes progressively. As far as cultures go, this is an evolutionary process. From a human viewpoint, however, such progressive change looks like the development of a more complex living system, a socio-dynamic. Is technology cancerous, born of mutation as it burst out of its hunter-gatherer background, as it evolves into new forms? Or is it developmental, exploiting new organisations as it invents them but maintaining an adaptable but stable path, just like a developing embryo? An embryo destroys many organised structures, and kills many of its cells, as it develops. It builds scaffolding and throws it away when it is no longer needed.

From the point of view of the individual human, caught up in a technological rat race, this stress is clearly a symptom of social pathology, as Alvin Toffler argued in Future Shock. In contrast, looked at culturally, it is natural development. This difference of viewpoint resembles the two ways to describe a thinking mind: nerve cells and consciousness. More generally, not only can every complex system be described in several non-overlapping ways: it can also be described on several levels … as concrete or as a bridge; as an architectural bridge or as a weak point for an enemy invasion.

Human evolution occurs on two levels: embryological development and cultural development. Neither process is preformational, with all necessary ingredients already present. Neither is a straightforward blueprint: make it like this. In both, evolutionary changes occur through complicity between several programmes, each of which affects the future of the others. As time passes, each programme not only affects its own future by its own internal dynamics: it also changes its future by the changes it causes in the other programmes.

To what extent are those changes predictable or accidental? There is a difference here between two modern viewpoints, one associated with the palaeontologist Simon Conway Morris in Life’s Solution: Inevitable Humans in a Lonely Universe, and the other with the late Stephen Jay Gould in Wonderful Life. This difference is crucial to the issue of design in evolution.

Gould made great play of the variety of the animals represented by the fossils in the Burgess Shale, deposited at the start of the Cambrian period about 570 million years ago. These fossils had been described by previous biologists, but Morris reworked and reconstructed them. He classified these fossils into a wide range of morphological types; in fact, many more basic kinds of animal design (‘phyla’) than had been assigned previously. Gould used this wide range of body designs, only a few of which have descendants among present-day creatures, to argue that life can do almost anything by way of morphology, even in its fundamental or basic structure, and that the organisms that now exist are accidental survivals from the much vaster range that existed at the start of the Cambrian.

Morris, however, has come to believe the opposite, namely: because some of the many themes have converged to produce similar beasts, some specific designs must be winners, no matter how they are realised. Therefore any wide array of different body structures will necessarily evolve to generate much the same spectrum that we observe today, automatically selected because those are the body-plans that work best. The fossil record contains many cases of this kind of convergence:fn3 ichthyosaurs and dolphins have evolved to look like sharks and other carnivorous fishes, because that’s the shape that’s most efficient for a fishy predator. In short, Morris believes that if we were to find living creatures on a similar planet to Earth, or if we were to run Earthly evolution again, then much the same range of animal designs would appear. Aliens on a world like ours would be much like us, even if their biochemistry were totally different.

In contrast, Gould believed, as we do,fn4 that in such a rerun the resulting spectrum of life forms would not resemble the current ones at all. Different designs, fundamentally different body forms, would be just as likely as the ones that happen to exist now. The current body-plans are just a contingent, accidental collection that happened to survive. Aliens, even the highest ones, would most likely be very different in design from us, whatever world they evolved on. Including a reboot of ours.

The old view of the role of genes in Darwinian evolution emphasised mutations: random changes to DNA sequences. However, at least in sexual species, the main source of genetic variability is actually recombination: mix-and-match shuffling of gene variants from the parents. New mutations are not needed to innovate; new combinations of existing genes are sufficient. The diversity of available gene variations can be traced back to much older mutations, but you don’t need a mutation now to change an organism.

All biologists now agree that the body-plans of organisms are not built up piece by piece, mutation by mutation, but have been selected by recombination. Instead of mutations to new genetic variants, we find recombinations of many ancient mutations. These are sorted from kits of compatible parts in every generation, not put together higgledy-piggledy and expected to work. If, as seems plausible, only a few developmental trajectories can lead to larvae that can feed and grow into working adults, compared to the huge number that can’t, then it is to be expected that the successful designs are all separated, without intermediate forms bridging the gaps. ‘Missing links’ need not be missing – or links – because continuous variation is not required in a discontinuous process.

By looking at so-called r-strategists, animals like plaice and oysters whose larvae comprise only a few developmentally competent ones among a majority that aren’t, we can see how this is achieved today. What it does not tell us – what distinguishes the Morris and Gould views – is whether the successful designs are out there in some Platonic organism-space, waiting to be found, or whether the organisms have all invented their own, unpredictable, designs as they went along. Morris, a Christian, believes the former: the appearance of design is the revelation of transcendental attractors in God’s design-space of possible organisms. We, however, believe that there are so many possible ways of being a successful organism, so many effective designs, that the drunkard’s walk of evolution keeps finding them, even though they are sparsely embedded in the vast majority of failures.

In particular, we think that intelligent design focuses too narrowly on the evolution of specific structures found today, such as the precise molecular configuration of haemoglobin or the bacterial motor. In retrospect, these structures seem highly improbable; if nature were to aim for them again, it would almost certainly miss. But evolution selected these structures when it encountered them. What matters is how likely it is that evolution could find some such structure, not that specific one. If there are many suitable structures, then a process that automatically homes in on anything that seems to be an improvement has a good chance of finding one of them.

Think how improbable you are. If two genomes had not combined just so, if that egg and that sperm had not come together, if your father hadn’t met your mother at the dance, if the wartime bomb in the harbour had hit your grandfather instead of being a hundred metres away, if Napoleon had won the Battle of Waterloo, or if victory had gone the other way in the American War of Independence, if the nascent Earth had not acquired an ocean, or the ripples in the Big Bang had been slightly different … you wouldn’t be here.

The odds that you exist are infinitesimal.

No. The odds that you exist are certain, because you do.

The processes that led to you are robust, and at each stage would have led to something similar, albeit different, if run again. No complex process ever produces the same result twice. But if it produces a similar result instead, that makes its consequences certain, not utterly unlikely. Only fine details will be different, second time around. The lottery of life is quite different when seen through the eyes of the eventual winners, rather than those of a random competitor before the contest has happened.

It’s tempting to assume that the evolution of technology can tell us about organic evolution, or vice versa, but these processes lead to apparent design in very different ways. However, there is a grand overall similarity in how we think about both systems, particularly how our thinking has changed over the last few years. The appearance of design is the most dramatic element in both systems. Although its provenance is different in the two cases, we are no longer surprised by it. We have realised that the universe is not doomed by increasing entropy to an eventual ‘heat death’, a traditional but somewhat misleading term which actually means that the universe will end up as a structureless lukewarm soup. Instead, the universe ‘makes it up as it goes along’, and what it makes up are designs. In that sense, at least, the appearance of new design in both technical and organic systems can be considered comparable. But it’s important not to stretch the metaphor too far.

Cultures can also be seen as evolving. In many ways, cultural evolution sits between organic and technological evolution. Advanced human societies make their members different and varied. All societies produce numerous distinct roles, from those limited by sex and age, such as childbearing or going to school, to those that seem to be chosen by the individual: warrior, accountant, thief. There is a division among sociologists that is comparable to that between Morris and Gould. Some believe that the roles are in some sense transcendent or universal; they look for proto-accountants in ‘primitive’ hunter-gatherer societies. Carl Gustav Jung’s theory of archetypes, such as the persona, the shadow, and the self. In his view, these were extremely ancient common images derived from humanity’s collective subconscious, which affected how we interpret the world. Others, however, believe that some roles in different societies, even though they look similar and the names translate similarly, can be fundamentally different: a Japanese car worker has a different worldview from that of his English equivalent, and occupies a different societal slot.

Both sociological viewpoints can provide useful insights: different societies, like different ecologies and different cultures, provide diverse roles for their members. The cultural invention of generic occupations is comparable to the organic invention of things like chordates, trilobites, muscles and nests. It is also comparable to the technological inventions of – say – bicycles, the internal combustion engine, wheat and rope. Money in human societies is usefully analogous to the way cells produce and exchange energy, using the molecules ADP and ATP (adenosine di- and tri-phosphate). Indeed, ATP is often called the unit of molecular currency. The appearance of new designs in organic evolution, in cultures, in technology, and even in language, can usefully be compared. Even so, such comparisons must be made very carefully and not pushed beyond reasonable limits.

The idea that technology evolves is not the orthodox view, wherein design and evolution are considered to be opposites. Design in technology is usually seen as being invented, not as having evolved. This assumption lies at the core of Paley’s famous analogy between a living creature and a watch. Watches are intricate devices, designed and made by an intelligent agency. Therefore, if you find something equally intricate in living creatures, it must also be designed, and the creatures must have been made by an intelligent agency. Therefore there must have been a cosmic designer, QED. The same assumption motivates the current hypothesis of intelligent design, which is basically Paley’s argument restated using examples from modern biochemistry.

However, analysis of the history of nearly all inventions shows them either to be developments of previous technology, that is to say adaptations, or perversions of some technology in a different sphere. (A few do seem to come out of thin air, with no significant precursors.) The biological term for such things is ‘exaptations’, a word introduced by Gould and Elizabet Vrba in the 1980s. It refers to an organic or a technological development that arises from an entirely different structure or function. An example is the use of feathers for flight. Feathers first appeared in dinosaurs, but their skeletal structure shows that the early feathered dinosaurs didn’t use their feathers to fly. We can’t be certain what they did use them for, but the most plausible functions are for warmth or for sexual display. It may well have been both. Later, feathers turned out to be useful for wings and flight, and birds evolved. Nature is an opportunist. A technological example of exaptation is the use of disc-recorded sound for music. Edison originally developed the phonograph for a more serious purpose, to record for posterity the last words of famous men and the historical speeches of politicians. He greatly deplored its use for frivolities like music, but accepted payments gracefully, nevertheless.

Exaptation is one of the less obvious tricks that evolution has up its sleeve, and is often the solution to evolutionary puzzles, in which a particular function can occur only when several interrelated structures apparently have to appear simultaneously, but none of them can perform that function on its own. Although it’s tempting to deduce that such structures can’t evolve at all, they can if exaptations occur. Then the structures concerned initially perform different functions.

A classic instance is the bacterial flagellum, a structure that proponents of intelligent design argue cannot possibly have evolved by any conceivable route. The flagellum allows some bacteria to move of their own volition. Its most important component is a tiny molecular motor, which causes the flagellum to rotate, much as the motor of a boat turns the propeller. The bacterial motorfn5 is made from a large number of different protein molecules. Until recently, evolutionary biologists could offer no convincing explanation of the origin of such a complex structure by natural selection.

In 1978 Robert MacNab wrote: ‘One can only marvel at the intricacy, in a simple bacterium, of the total motor and sensory system … What advantage could derive … from a “preflagellum” [meaning a subset of its components], and yet what is the probability of “simultaneous” development?’ In 1996 Michael Behe, a biochemist and leading proponent of intelligent design, repeated MacNab’s worries in Darwin’s Black Box, together with several similar evolutionary puzzles. He concluded that while many, indeed most, features of living creatures have evolved, some cannot possibly have done so because they are irreducibly complex: if you remove any component, they cease to function.

It’s a genuine puzzle, but before invoking some unspecified genie-of-the-lamp, without independent evidence that it exists, we ought to make sure that conventional evolutionary processes definitely can’t hack it. Intelligent design doesn’t just argue that some specific evolutionary route is wrong: it claims a proof that in principle no such route can exist. If you’re going to invoke a general principle of this kind to assert the existence of a supernatural being or a highly advanced cosmic designer, you need to close any loopholes in your logic. Otherwise your entire philosophy will be built on sand, whatever actually happened. The Book of Genesis could be true in every detail, but your supposed proof would still be nonsense if its logic were defective.

In response to intelligent design, biochemists have taken a closer look at the proteins in the bacterial motor and the associated genes. The most prominent components of these motors are rings of proteins, which are very common in evolution. What use is a ring? It has a hole. Holes are amazingly useful to a bacterium or a cell, because they can function as pores or sockets. Pores let in molecules from the outside world, or expel molecules into the outside world. Different-sized pores deal with different-sized molecules. That’s something that natural selection can work on: a mutation in the DNA that codes for the protein can lead to one with a similar, but slightly different, shape or size. As soon as a pore does a useful job, evolution can find a pore that is better at doing that job, if there is one.

Sockets allow bacteria or cells to attach new structures, either inside or outside the cell membrane. Many different molecules can fit into the same socket, and again, evolution has plenty of opportunities to work with. What began as a pore can become a socket if something happens to fit into it. When the two modules come together, their function may change. Exaptation demolishes irreducible complexity as an obstacle to evolution. You don’t even have to prove exactly how a given structure evolved, because irreducible complexity allegedly rules out not just the actual route, but any conceivable one.

So let’s conceive.

A number of biologists have attempted to deduce a plausible or likely evolutionary route to a bacterial motor, from DNA and other biochemical evidence. This turns out not to be especially difficult. Many details are still provisional, as is all science, but the story is now sufficiently complete to disprove the contention that the motor exhibits a type of complexity that rules out all evolutionary explanations. Agreed, that doesn’t prove that the current evolutionary explanation is correct. That must be confirmed, or denied, by further scientific investigations. But it’s quite different from asking whether, in principle, any such explanation can exist.

The most fully developed synthesis of these proposals, put together by Nicholas Matzke, starts with a general-purpose pore. This evolves into a pore with more specific functions. At this early stage, the structure is not a motor, but it already has a very useful, entirely different, function: it can transport molecules out of the cell. In fact, it is recognisable as a primitive version of so-called Type III Export Apparatus, which exists in modern bacteria, and DNA sequences support this. Further changes, in which the pore’s function is successively improved, or changed by exaptation, provide an entirely plausible evolutionary route to the bacterial motor, increasingly supported by DNA evidence.fn6

Yes, if you take away enough parts of the bacterial motor, then it might not be a very good motor any more. But evolution didn’t know it was supposed to be making a motor.

So ‘design’ isn’t what it is often thought to be, even for human technology, let alone biology. Each innovative step may be driven by human intentions, but what works, and what passes on to later technology, evolves. To some extent, cars evolved from horse-drawn carriages, and a ballpoint pen is the lineal descendant of a quill made from a feather. We can legitimately compare these developments to mammals evolving from a Devonian fish that came out of the water onto land, or to our little middle-ear bones being the lineal descendants of bony gill structures in that fish.

Evolution is not efficient. It throws an awful lot of things away. Innumerable land vertebrate species have gone extinct. Similarly, most human designs don’t work. From the enormous number on offer, only a few develop into sophisticated structure/function niches. We are all bound by tradition, as well as by functional constraints that require any new development to fulfil the same functions as its ancestor. There’s a classic example: Apollo rockets were moved to their launching-pads on rails that were much too close together for stability, because the gauge of America’s railways came from mine railways that were two horses wide. So the Moon project was jeopardised by horse’s asses.

To be specific, let’s think about better mousetraps. Mousetrap evolution is a process, not just a succession of models; it branches into the future. The pattern that has a metal bar coming down and (one hopes) breaking the mouse’s neck, has expanded into dozens of different models, some computer-controlled. Those that trap the mouse in a metal tube, or a cage, are more like descendants of lobster-pots, but these too have performed what biologists would call an adaptive radiation: we found seven different kinds, with sprung doors or elastic apertures for entry.

The same goes for bicycles, cars or computers: they all adaptively radiate into the future. Each new ability, such as computer control – a logic chip – on a particular technical road branches into new roads. Think of the familiar cat flap, now available in versions that allow your own cat, wearing its magnetic collar, in or out, but exclude foreign cats. Or fancy electronic ones that verify your cat’s ID. Full-body scanners to detect terrorist cats carrying exploding mice cannot be far away. Just as in organic evolution, the adjacent possible is continually being invaded: possibilities just one step away from current practice are tried, rather unoriginally.

We usually think of this as technical development, not innovation, unless it is in an unexpected direction: Teflon used for non-stick frying-pans, or penguins’ wings used for swimming. Most aquatic vertebrates, unlike these birds that have become secondarily aquatic, use their tails, not their fins, for propulsion. Such more original changes of direction are best thought of as exaptations rather than adaptations. Or, to use a less biological term, genuine innovations.

Among those who accept evolution as a reasonable metaphor for many examples of progress in technology, it used to be thought that the major difference between technical and organic evolution is that technological evolution is Lamarckian – named after the French naturalist Jean-Baptiste Lamarck, a contemporary of Darwin – whereas organic evolution is Darwinian. In Lamarckian evolution, acquired characteristics can be inherited – if a blacksmith acquires strong arms because of his work, his sons should also have strong arms. In Darwinian evolution, that’s not possible. Neo-Darwinism illuminates the difference: heritable characteristics are those that are determined by genes.

Lately, this distinction has become a bit blurred, and each mechanism has acquired features that were thought to be characteristic of the other. Technical development has borrowed a trick from evolution to construct so-called genetic algorithms for the development of new products. Digitised designs are shuffled, by analogy with recombination, the way biological reproduction shuffles gene variants from both parents. The next technological generation to survive this process combines the more useful features of previous generations. Sometimes it has new emergent properties, which are selected if they prove useful, and are retained. Often the final design is incomprehensible to a human engineer. Evolution need not obey human narrativium.

The phenomenon of genetic assimilation, which is entirely Darwinian, can look very Lamarckian. Changing a population progressively by selecting genetic combinations that work can change the thresholds at which particular capabilities come into play. As a result, effects that originally depended on some environmental stimulus can happen without that stimulus in later generations. For example, the skin on the soles of our feet gets thicker when we walk regularly, an acquired characteristic; however, genetic recombinations that provide babies with thicker skin on their feet from the start make this process more effective, and so are selected for. Any new feature, acquired or not, that works – that improves the chances of surviving to reproduce – reveals a feature that Darwinian evolution could blunder into and exploit. Genetic assimilation may indeed be the usual way that originally responsive adaptations get built in to the developmental schedule.

In particular, the old distinction between Lamarck and Darwin has lost its power to distinguish technical from organic evolution. But that doesn’t imply that there are no significant differences. It’s tempting to think that one obvious aspect of technological evolution surely can’t apply to Darwinian evolution: imagining a possibility before designing a technique or gadget to implement it. Human technology is born in the imagination of a series of inventors or discoverers: ‘What would happen if …?’ is a theoretical exploration of Kauffman’s adjacent possible. Much of the time, imagining possibilities leads to hypothetical new inventions being rejected without bothering to make them or test them: they wouldn’t work because … or no one could use them because … or they would be too expensive … or they wouldn’t perform well enough to displace the widget that already does the job very well.

It doesn’t seem possible that this imaginative process could have an organic analogue – but it does. In 1896 the psychologist James Mark Baldwin wondered whether animals carrying out behavioural experiments might be drawn into the evolutionary process, in effect by imagining what would happen if they could perform some new task that was actually beyond them. For instance, an okapi is like a giraffe, but its neck and legs are of normal length. Suppose that an adventurous okapi, for example, kept reaching up in an attempt to browse on the lowest branches of trees, despite repeated failure. Because it failed, this would be analogous to imagination. But occasional success could favour okapi with slightly longer necks and legs, leading to a giraffe. This process is often called the Baldwin effect.

A few years ago, we observed some animal behaviour that could well become the root of such an evolutionary trajectory – an exaptation in the making. Plecostomid catfish (‘plecs’) are common scavengers in larger aquariums, cleaning algae off the glass with their sucker-like mouths. In the wild, they can hold tight to smooth rocks as they glean the algal film; they also have effective armour with barbed bony supports in their dorsal and pectoral fins. In aquaria, these characteristics give them an entirely different ability, which we saw a plec exploiting in the Mathematics Institute common room at the University of Warwick. This plec’s natural abilities made it much better than other fish at garnering floating pellets of food. It did this using a method quite alien to wild plecs: it turned on its back and used its sucker-mouth to take in soggy pellets, while its spiky fins kept off the competition. So a catfish mouth, adapted for taking food from rocks, can be exapted to take food pellets from the water surface, especially if the fish concerned has effective defences, and the food is soft.

Future genetic assimilation could easily build this kind of exapted behaviour into the genes of the plec population. It could be selected for, and then adapted along an evolutionary trajectory, so that a plec would take food from the surface normally in just this way.

In fact, something of the kind has probably happened already – though not in descendants of the Mathematics Institute plec, which had none. The fish in question is the upside-down catfish Synodontis nigriventris,fn7 which takes insects from the surface of the water in the wild using a similar technique. We have, then, both ends of a plausible evolutionary trajectory. It starts, perhaps, with a hungry catfish alerted to a food mass on the surface, near it in shallow water; perhaps a rotting, floating insect carcass. The catfish turns over in its attempts to get its mouth near to the tempting morsel, and even if it mostly makes a hash of this, any occasional success is rewarded. It will now be sensitised to this source of food, and might haunt the shallows for more of them. Its offspring, growing up in the same environment, are then more likely to be selected for similar behaviour if genetic changes can make it more effective.

This scenario contradicts Stephen Jay Gould’s assertion in The Flamingo’s Smile that adaptations like the upside-down feeding of the flamingo, scooping up crustaceans from saline lakes, must involve a single radical departure from the normal use of the beak. Animals can try out little behavioural experiments, and if they are rewarded, these can become built into their subsequent behaviour. Then, if the reward is as important as a new source of food or novel access to mates, natural selection can improve it.

Technical evolution can avoid such time-wasting, progeny-wasting, and new-function-wasting aspects of organic evolution in two ways. The first, we have discussed already: human minds can attempt to jump into the adjacent possible and see if it works ‘in the imagination’. Can we imagine an aeroplane ten times the size, and what would need to be changed for it to work? If we exaggerated the length of a bicycle frame and had the cyclist lean back, how could he see the road? Do we then want him on his front? Both have been tried, and are excellent examples: technical results of our imagination playing in the adjacent possible.

The other trick that minds can do to improve technology is to copy: to take a technical trick used in one invention and to spread its use to others. That trick, except for a few cases where genes have achieved horizontal transfer between species, is impossible for organic evolution: each lineage must invent for itself. A recent spread of this kind has been the use of digital switches in a variety of machines from toasters and children’s toys to automobiles. The big one before that was the use of plastics to replace metals in the nursery, kitchen and laboratory. Before that, transparent plastics, mostly acrylics, had been used to replace glass in many applications. The progressive use of semiconductor technology is giving us solar panels, tiny refrigerating or heating elements, and a new family of very efficient light bulbs: white-light LEDs. Banks of coloured LEDs can now be tuned to give different lighting conditions; bright white light is not conducive to sleep and can be replaced with softer tones. Flexible television/computer screens, which can be rolled up like paper, already exist in the laboratory, and are not far from commercial production. An entire book has just been encoded in DNA, and a human face has been printed on a human hair.

In biological evolution, it used to be thought that environmental ‘niches’, such as predatory behaviour, were already available and waiting to be occupied, rather as though some cosmic script had already written down all the possible things that an organism might do. Now it is thought that organisms construct niches as they evolve; for instance, you can’t occupy the dog-flea niche until there are dogs.

Even taking copying into account, the analogous questions of competition and niche-construction in technology are as important as they are in the natural world, and they too force the evolution of new products. A good example was the colonisation of the marketplace in the 1970s by VHS videotapes, even though its rival Betamax was much better in several respects. As in natural ecologies, it often happens that a less-adapted, often foreign, invader exploits the ecosystem more effectively, forcing the demise of well-established local species. The grey squirrel, for example, carries a disease that decimates indigenous red squirrels, much as the Spanish invaded South America and destroyed Inca and Maya empires. The red squirrel was better-adapted to its original environment, but the arrival of the invading grey squirrel changed that environment; in particular, it now included grey squirrels and their disease organisms. The change was sudden, biological warfare rather than the usual sedate pace of natural selection in a slowly changing environment.

In the technical world, then, there do exist processes resembling those of organic ecosystems. Many of them are recursive, affecting their own development: supermarkets make their own ecosystem of consumers, just as dogs create a new niche for dog-fleas. This makes questions about the design of technology much more difficult, because there are few real innovations, but many exaptations, copyings and adaptational trajectories. Only a few really novel tricks can be claimed to have a human designer in a non-evolutionary sense.

There is a trajectory of development for a technological product: a car starting from carriages and an engine, steam or internal-combustion; a radio starting from a crystal set and headphones; a bicycle starting as a penny-farthing and evolving through the sit-up-and-beg still seen all over China and India to the mountain bikes and lie-down versions of the latest adaptive radiation.

These are paths through our cultural history, and they make their own contexts as they evolve. The car creates vast and important areas of our cities where cars are built, where auto workers live, where the suppliers of parts have some of their factories and warehouses. When we give little Johnnie a bicycle on his seventh birthday, we introduce him to a new world that has grazed knees, gears, punctures, comparison with Fred’s bike … When the transistor radio erupted into Western culture in the 1960s, it changed the relationships of teenagers to each other and to pop stars, though nothing like as much as the mobile phone has changed all of our lives in the last few years. Alexander Graham Bell, on a promotional tour of his invention the telephone, so impressed one city’s mayor that he is said to have declared: ‘What a wonderful invention; every town should have one.’

Artefacts evolve, and the functions they perform get better, wider, cheaper. But they also change the society around them, so that their ‘improved’ next generation already has the ground prepared for it. The Ford Model T would not have been viable without gas stations, which had appeared to service the much more expensive previous generation of automobiles. In turn, the Model T and other similar affordable automobiles with privacy in the back seat changed much of the sex life of the young men and women who had access to them. Society’s rules change as the Ford Model T, the transistor radio, central heating, subway travel and mobile phones affect their context, and the context in turn constrains or directs the further evolution of the product.

Nearly all inventions don’t follow that kind of successful path; like nearly all species of organisms, they prosper for a little while but then die out. The few that do survive find a trajectory that takes them into the future. Frequently they move into a whole new phase space of possibilities, where their original design is effectively useless, but the new world now has an improved design. Like a genuine Stone-Age axe that’s had its handle and blade changed several times, we find a new world with a new artefact and a new function.

In The Science of Discworld III we described how apparently rigid limitations on the energy needed to put a person or cargo in orbit around the Earth could, in principle, be overcome by changing the context. If you use a rocket, the amount of energy needed to get a 100-kilogramme man up to synchronous orbit can be calculated using Newton’s laws of motion. It is the difference in potential energy caused by the planet’s gravity well. You can’t change that, so at first the limitation seems foolproof.

In the mid-1970s, however, a wholly new suggestion was made: the space bolas. Essentially this is a giant Ferris wheel in orbit. The traveller gets into the cabin as it swings past the upper atmosphere, and gets out again when it approaches the furthest point from Earth. A succession of such gadgets can deposit him in synchronous orbit a few weeks later.

A third step in the ladder of technology, not practical yet but already being discussed by engineers, is the space elevator. The science fiction writer and futurologist Arthur C. Clarke was one of several people who had this idea: take a ‘rope’ up to synchronous orbit and let it down to an equatorial landing-strip. The result would be a material link from a point in synchronous orbit to the ground. Once this is set up, a system of cabins and pulleys-and-weights like those used in skyscraper elevators could take a person up to orbit very efficiently. Counterweights, or another man coming down, would reduce the cost to that of the energy required to override friction.

The point is not whether we can do this yet. We can’t; even carbon-fibre ‘rope’ is too weak. But the space elevator shows how a design trajectory can take a function away from its earliest, primitive constraints, so that a whole new set of rules applies, and the old limitations become … not invalid, but irrelevant.

More familiar examples of this ‘transcendent’ process are writing and telecommunication. The first attempts at writing probably involved scratches on rock or bark, and these matured in two directions – pictorial and phonetic writing. Pictorial writing, such as ancient Egyptian hieroglyphs and modern Chinese, has found it difficult to move up the technological ladder. They are not even at the rocket stage; fireworks, perhaps. Phonetic writing was more suitable for printing – the space bolas stage of the technology. This was improved as far as the great newspaper printing presses of the twentieth century and the electric typewriter. Then came the space elevator stage, word processing by computer. Ironically, this may just have saved Chinese ideograms, now easily typeset by computer, from oblivion. A further stage is starting to appear with eBooks and iPads. Eventually, all writing might be virtual, encoded in physically tiny memories until it needs only to be actualised on screens and in minds.

Communication at a distance started with semaphore and chains of watch-fires on hilltops. Navies developed coded systems of flags for communication between ships. Discworld inventors developed the clacks, a mechanical telegraph with repeaters at limit-of-sight, aided by telescopes, while we used a signal-box and mechanical linkages to signal to trains miles from the box. With electricity came the ability to send signals via cables, and the telegraph was born. Several different coding systems for commercial transactions, and a primitive fax machine, were in commercial use before 1900. All these were rocket-ships. Then came the telephone, which uses sound waves to modulate an electrical signal. Much capital investment went into wiring the countryside and undersea cables to connect the continents. These heroic ventures were comparable, in technical difficulty, with putting up a space bolas now. Meanwhile ‘wireless’ began to be used: radio, and later television. With mobile phone technology, depending upon billions of pounds of investment in immensely sophisticated base stations and in research to improve and develop the handsets, we are now beginning the space elevator stage of telephone technology.

We can compare these technical innovations to developments in organic evolution. We analyse the development of mammals on two scales, to show how the evolutionary process outgrows its initial constraints and achieves new properties and functions as the trajectory changes direction. We choose the two scales to emphasise that this is not a description of what actually happens during organic evolution.

We have already met the question of the extent of diversity in the animals of the Burgess Shale, and the differences of opinion between Gould and Morris. Among these animals from the Cambrian explosion, several were early chordates, ancestral to our own group of animals, including today’s fishes, amphibians, reptiles, birds and mammals, as well as a diversity of modern oddities like sea squirts and lampreys. The Burgess Shale fossil Pikaia is the best-known early chordate, but there are others in similar Australian and Chinese fossil beds.

The early chordates produced a great adaptive radiation, firstly of jawless armoured fishes, then of a substantial number of jawed forms, including sharks, rays and bony fishes. Some of the latter, in the Devonian period, came out onto the land as early amphibians. These aquatic forms are/were the rocket-ship phase of chordate existence. The amphibians, and their diverse reptilian descendants, such as dinosaurs, birds and mammal-like reptiles that included our ancestors, constitute the next step up, the chordate space bolas. The third stage was achieved separately, and rather differently, by birds and mammals. Birds specialised in warm-bloodedness and efficient lung ventilation for flight, so that they had to provide food for their young, caring for them in nests until they could adopt the very demanding lifestyle of their parents. Mammals became turbo-powered by maintaining a stable high body temperature, and invaded many more habitats than birds, from burrowing and swimming to flying. Which they now do nearly as well as birds, but without flow-through lung ventilation. From a wide-screen chordate viewpoint, mammalian design is their space elevator.

Within that last step, we can find a similar series of invasions of the adjacent possible, in which terrestrial ecosystems were themselves changed by the presence of large land animals. Grassland such as savannah and steppes, arctic birch, lichen and moss tundra are all maintained by continuing interactions with large herbivorous mammals. Vast numbers of small rodents – mice, rats, voles, lemmings, hamsters – live in and under these grasslands. They eat more of the vegetation than their larger cousins do, and they contribute more to those ecosystems. Some interactions between mammals and their environment are familiar: rabbits making warrens, badgers excavating setts, deer ringing trees. We have to visit zoos to see the full adaptive radiation, including those strange rodents of the South American pampas: pacas, capybaras and cavies (guinea-pigs). And bats. And porpoises, dolphins, toothed whales and filter-feeding baleen whales. And all of the primates, including us. So mammals, like insects among the invertebrates, are the big terrestrial success story.

In terms of our space-exploration analogy, the mammal-like reptiles of four hundred million years ago, and today’s monotremes (egg-laying oddities like the echidna and the duck-billed platypus) are the rocket-ships. The marsupial mammals – kangaroos, potoroos and opossums – are the space bolas. The placental mammals – most of today’s mammals, including cows, pigs, cats, dogs, hippos, elephants, monkeys, apes and humans – are the space elevator.

Any evolutionary series can be presented as a ladder of emergent properties, new ways of being that obey new rules and have effectively discarded the old constraints. This vision is as appropriate to mammals as it is to writing tools or radio receivers. It is a general property of our self-complexifying planet in its self-complexifying universe. As time passes, more different things happen in more ways, with new rules and new functions.

That vision, of the multifarious universe knotting itself into patterns that themselves build upon previous patterns, is almost perfectly opposite to the twentieth-century view of ever-increasing entropy leading to heat death. Can this self-complication continue infinitely? We don’t know, but it is as sensible a view as its opposite, and there is considerable evidence for it. Does that mean that anything not possible now will necessarily be possible in future? Of course not. At each step upwards, there is selection among possibilities.

This selection process is what mathematicians call symmetry-breaking: more possibilities seem to be available beforehand than are actualised at the next stage, yet paradoxically there are more possibilities afterwards than before. If advancement is the rule, and it seems to be, then contingency and selection are making up the future by evolving from rocket-ships into future space elevators, almost everywhere. We should perhaps be surprised that Moore’s Law has worked for so long, but when we examine the changes in computing technology over the last decades we see that, just as in the recent mammal story, the improvements were always inconceivable at the earlier step.

This is why blinkered applications of laws of nature, such as conservation of energy or the second law of thermodynamics, can be misleading. As well as content, laws have contexts. A law of nature may appear to pose an insuperable barrier, but if you have applied the law in an inappropriate context, you may have left a way for nature to sneak round it. And it will.

fn1 In 1841 Richard Owen, a leading palaeontologist, found an incomplete fossil that he thought was a hyrax (because of its teeth) and assigned it to a new genus, Hyracotherium. In 1876 Othniel Marsh discovered a complete skeleton, obviously horse-like, and assigned it to another new genus, Eohippus (dawn horse). Later it became clear that the two fossils belonged to the same genus, and by the rules of taxonomy, the name that was the first to be published won. So the evocative ‘dawn horse’ was lost, and a scientific misconception was preserved.

fn2 Since 1881, fossil discoveries have inserted a whole series of intermediates between fish and amphibians: Osteolepis, Eusthenopteron, Panderichthys, Tiktaalik, Elginerpeton, Obruchevichthys, Ventastega, Acanthostega, Ichthyostega, Hynerpeton, Tulerpeton, Pederpes, Eryops.

fn3 Jack recalls a bright Irish student who, in an exam question about convergent evolution, defined it as ‘where the organs of two descendants are more alike than they were in the common ancestor’.

fn4 See Jack Cohen and Ian Stewart, What Does a Martian Look Like?

fn5 We say ‘the’ motor, because everyone does, but different bacteria have different motors. Darwin was puzzled why the deity would design hundreds of very similar barnacles, all of different species; we may similarly wonder why an intelligent designer would intervene in the normal process of evolution to equip dozens of bacteria with individually designed motors.

fn6 N.J. Matzke, Evolution in (Brownian) space: a model for the origin of the bacterial flagellum, www.talkdesign.org/faqs/flagellum.html.

fn7 The name means ‘dark belly’, because when it swims upside down, its back has become light like most fishes’ bellies, and its belly dark.