The Beginning of Infinity: Explanations That Transform the World - David Deutsch (2011)

Chapter 4. Creation

The knowledge in human brains and the knowledge in biological adaptations are both created by evolution in the broad sense: the variation of existing information, alternating with selection. In the case of human knowledge, the variation is by conjecture, and the selection is by criticism and experiment. In the biosphere, the variation consists of mutations (random changes) in genes, and natural selection favours the variants that most improve the ability of their organisms to reproduce, thus causing those variant genes to spread through the population.

That a gene is adapted to a given function means that few, if any, small changes would improve its ability to perform that function. Some changes might make no practical difference to that ability, but most of those that did would make it worse. In other words good adaptations, like good explanations, are distinguished by being hard to vary while still fulfilling their functions.

Human brains and DNA molecules each have many functions, but among other things they are general-purpose information-storage media: they are in principle capable of storing any kind of information. Moreover, the two types of information that they respectively evolved to store have a property of cosmic significance in common: once they are physically embodied in a suitable environment, they tend to cause themselves to remain so. Such information – which I call knowledge – is very unlikely to come into existence other than through the error-correcting processes of evolution or thought.

There are also important differences between those two kinds of knowledge. One is that biological knowledge is non-explanatory, and therefore has limited reach; explanatory human knowledge can have broad or even unlimited reach. Another difference is that mutations are random, while conjectures can be constructed intentionally for a purpose. Nevertheless, the two kinds of knowledge share enough of their underlying logic for the theory of evolution to be highly relevant to human knowledge. In particular, some historic misconceptions about biological evolution have counterparts in misconceptions about human knowledge. So in this chapter I shall describe some of those misconceptions in addition to the actual explanation of biological adaptations, namely modern Darwinian evolutionary theory, sometimes known as ‘neo-Darwinism’.


Creationism is the idea that some supernatural being or beings designed and created all biological adaptations. In other words, ‘the gods did it.’ As I explained in Chapter 1, theories of that form are bad explanations. Unless supplemented by hard-to-vary specifics, they do not even address the problem – just as ‘the laws of physics did it’ will never win you a Nobel prize, and ‘the conjurer did it’ does not solve the mystery of the conjuring trick.

Before a conjuring trick is ever performed, its explanation must be known to the person who invented it. The origin of that knowledge is the origin of the trick. Similarly, the problem of explaining the biosphere is that of explaining how the knowledge embodied in its adaptations could possibly have been created. In particular, a putative designer of any organism must also have created the knowledge of how that organism works. Creationism thus faces an inherent dilemma: is the designer a purely supernatural being – one who was ‘just there’, complete with all that knowledge – or not? A being who was ‘just there’ would serve no explanatory purpose (in regard to the biosphere), since then one could more economically say that the biosphere itself ‘just happened’, complete with that same knowledge, embodied in organisms. On the other hand, to whatever extent a creationist theory provides explanations about how supernatural beings designed and created the biosphere, they are no longer supernatural beings but merely unseen ones. They might, for instance, be an extraterrestrial civilization. But then the theory is not really creationism – unless it proposes that the extraterrestrial designers themselves had supernatural designers.

Moreover, the designer of any adaptation must by definition have had the intention that the adaptation be as it is. But that is hard to reconcile with the designer envisaged in virtually all creationist theories, namely a deity or deities worthy of worship; for the reality is that many biological adaptations have distinctly suboptimal features. For instance, the eyes of vertebrates have their ‘wiring’ and blood supply in front of the retina, where they absorb and scatter incoming light and so degrade the image. There is also a blind spot where the optic nerve passes through the retina on its way to the brain. The eyes of some invertebrates, such as squids, have the same basic design but without those design flaws. The effect of the flaws on the efficiency of the eye is small; but the point is that they are wholly contrary to the eye’s functional purpose, and so conflict with the idea that that purpose was intended by a divine designer. As Charles Darwin put it in The Origin of Species, ‘On the view of each organism with all its separate parts having been specially created, how utterly inexplicable is it that organs bearing the plain stamp of inutility . . . should so frequently occur.’

There are even examples of non-functional design. For instance, most animals have a gene for synthesizing vitamin C, but in primates, including humans, though that gene is recognizably present, it is faulty: it does not do anything. This is very difficult to account for except as a vestigial feature that primates have inherited from non-primate ancestors. One could retreat to the position that all these apparently poor design features do have some undiscovered purpose. But that is a bad explanation: it could be used to claim that any poorly designed or undesigned entity was perfectly designed.

Another assumed characteristic of the designer according to most religions is benevolence. But, as I mentioned in Chapter 3, the biosphere is much less pleasant for its inhabitants than anything that a benevolent, or even halfway decent, human designer would design. In theological contexts this is known as ‘the problem of suffering’ or ‘the problem of evil’, and is frequently used as an argument against the existence of God. But in that role it is easily brushed off. Typical defences are that perhaps morality is different for a supernatural being; or perhaps we are too limited intellectually to be able to understand how moral the biosphere really is. However, here I am concerned not with whether God exists, only with how to explain biological adaptations, and in that regard those defences of creationism have the same fatal flaw as the Haldane–Dawkins argument (Chapter 3): a world that is ‘queerer than we can suppose’ is indistinguishable from a world ‘tricked out with magic’. So all such explanations are bad.

The central flaw of creationism – that its account of how the knowledge in adaptations could possibly be created is either missing, supernatural or illogical – is also the central flaw of pre-Enlightenment, authoritative conceptions of human knowledge. In some versions it is literally the same theory, with certain types of knowledge (such as cosmology or moral knowledge and other rules of behaviour) being spoken to early humans by supernatural beings. In others, parochial features of society (such as the existence of monarchs in government, or indeed the existence of God in the universe) are protected by taboos or taken so uncritically for granted that they are not even recognized as ideas. And I shall discuss the evolution of such ideas and institutions in Chapter 15.

The prospect of the unlimited creation of knowledge in the future conflicts with creationism by undercutting its motivation. For eventually, with the assistance of what we would consider stupendously powerful computers, any child will be capable of designing and implementing a better, more complex, more beautiful, and also far more moral biosphere than the Earth’s, within a video game – perhaps by placing it in such a state by fiat, or perhaps by inventing fictional laws of physics that are more conducive to enlightenments than the actual laws. At that point, a supposed designer of our biosphere will seem not only morally deficient, but intellectually unremarkable. And the latter attribute is not so easy to brush aside. Religions will no longer want to claim the design of the biosphere as one of the achievements of their deities, just as today they no longer bother to claim thunder.

Spontaneous generation

Spontaneous generation is the formation of organisms not as offspring of other organisms, but entirely from non-living precursors – for example, the generation of mice from a pile of rags in a dark corner. The theory that small animals are being spontaneously generated like that all the time (in addition to reproducing in the normal way) was part of unquestioned conventional wisdom for millennia, and was taken seriously until well into the nineteenth century. Its defenders gradually retreated to ever smaller animals as knowledge of zoology grew, until eventually the debate was confined to what are now called micro-organisms – things like fungi and bacteria that grow on nutrient media. For those, it proved remarkably difficult to refute spontaneous generation experimentally. For instance, experiments could not be done in airtight containers in case air was necessary for spontaneous generation. But it was finally refuted by some ingenious experiments conducted by the biologist Louis Pasteur in 1859 – the same year in which Darwin published his theory of evolution.

But experiment should never have been needed to convince scientists that spontaneous generation is a bad theory. A conjuring trick cannot have been performed by real magic – by the magician simply commanding events to happen – but must have been brought about by knowledge that was somehow created beforehand. Similarly, biologists need only have asked: how does the knowledge to construct a mouse get to those rags, and how is it then applied to transform the rags into a mouse?

One attempted explanation of spontaneous generation, which was advocated by the theologian St Augustine of Hippo (354–430), was that all life comes from ‘seeds’, some of which are carried by living organisms and others of which are distributed all over the Earth. Both kinds of seed were created during the original creation of the world. Both could, under the right conditions, develop into new individuals of the appropriate species. Augustine ingeniously suggested that this might explain why Noah’s Ark did not have to carry impossibly large numbers of animals: most species could regenerate after the Flood without Noah’s help. However, under that theory organisms are not being formed purely from non-living raw materials. That distributed kind of seed would be a life form, just as a real seed is: it would contain all the knowledge in its organism’s adaptations. So Augustine’s theory – as he himself stressed – is really just a form of creationism, not spontaneous generation. Some religions regard the universe as an ongoing act of supernatural creation. In such a world, all spontaneous generation would fall under the heading of creationism.

But, if we insist on good explanations, we must rule out creationism, as I have explained. So, in regard to spontaneous generation, that leaves only the possibility that the laws of physics might simply mandate it. For instance, mice might simply form under suitable circumstances, like crystals, rainbows, tornadoes and quasars do.

That seems absurd today, because the actual molecular mechanisms of life are now known. But is there anything wrong with that theory itself, as an explanation? Phenomena such as rainbows have a distinctive appearance that is endlessly repeated without any information having been transmitted from one instance to the next. Crystals even behave in ways that are reminiscent of living things: when placed in a suitable solution, a crystal attracts more molecules of the right kind and arranges them in such a way as to make more of the same crystal. Since crystals and mice both obey the same laws of physics, why is spontaneous generation a good explanation of the former and not of the latter? The answer, ironically, comes from an argument that was originally intended to justify creationism:

The argument from design

The ‘argument from design’ has been used for millennia as one of the classic ‘proofs’ of the existence of God, as follows. Some aspects of the world appear to have been designed, but they were not designed by humans; since ‘design requires a designer’, there must therefore be a God. As I said, that is a bad explanation because it does not address how the knowledge of how to create such designs could possibly have been created. (‘Who designed the designer?’, and so on.) But the argument from design can be used in valid ways too, and indeed its earliest known use, by the ancient Athenian philosopher Socrates, was valid. This issue was: given that the gods have created the world, do they care what happens in it? Socrates’ pupil Aristodemus had argued that they do not. Another pupil, the historian Xenophon, recalled Socrates’ reply:

SOCRATES: Because our eyes are delicate, they have been shuttered with eyelids that open when we have occasion to use them . . . And our foreheads have been fringed with eyebrows to prevent damage from the sweat of the head . . . And the mouth set close to the eyes and nostrils as a portal of ingress for all our supplies, whereas, since matter passing out of the body is unpleasant, the outlets are directed hindwards, as far away from the senses as possible. I ask you, when you see all these things constructed with such show of foresight, can you doubt whether they are products of chance or design?

ARISTODEMUS: Certainly not! Viewed in this light they seem very much like the contrivances of some wise craftsman, full of love for all things living.

SOCRATES: And what of the implanting of the instinct to procreate; and in the mother, the instinct to rear her young; and in the young, the intense desire to live and the fear of death?

ARISTODEMUS: These provisions too seem like the contrivances of someone who has determined that there shall be living creatures.

Socrates was right to point out that the appearance of design in living things is something that needs to be explained. It cannot be the ‘product of chance’. And that is specifically because it signals the presence of knowledge. How was that knowledge created?

However, Socrates never stated what constitutes an appearance of design, and why. Do crystals and rainbows have it? Does the sun, or summer? How are they different from biological adaptations such as eyebrows?

The issue of what exactly needs to be explained in an ‘appearance of design’ was first addressed by the clergyman William Paley, the finest exponent of the argument from design. In 1802, before Darwin was born, he published the following thought experiment in his book Natural Theology. He imagined walking across a heath and finding a stone, or alternatively a watch. In either case, he imagined wondering how the object came to exist. And he explained why the watch would require a wholly different kind of explanation from that of the stone. For all he knew, he said, the stone might have lain there for ever. Today we know more about the history of the Earth, so we should refer instead to supernovae, transmutation and the Earth’s cooling crust. But that would make no difference to Paley’s argument. His point was: that sort of account can explain how the stone came to exist, or the raw materials for the watch, but it could never explain the watch itself. A watch could not have been lying there for ever, nor could it have formed during the solidification of the Earth. Unlike the stone, or a rainbow or a crystal, it could not have assembled itself by spontaneous generation from its raw materials, nor could it be a raw material. But why not, exactly, asked Paley: ‘Why should not this answer serve for the watch as well as for the stone; why is it not as admissible in the second case as in the first?’ And he knew why. Because the watch not only serves a purpose, it is adapted to that purpose:

For this reason, and for no other, viz., that, when we come to inspect the watch, we perceive (what we could not discover in the stone) that its several parts are framed and put together for a purpose, e.g., that they are so formed and adjusted as to produce motion, and that motion so regulated as to point out the hour of the day.

One cannot explain why the watch is as it is without referring to its purpose of keeping accurate time. Like the telescopes that I discussed in Chapter 2, it is a rare configuration of matter. It is not a coincidence that it can keep time accurately, nor that its components are well suited to that task, nor that they were put together in that way rather than another. Hence people must have designed that watch. Paley was of course implying that all this is even more true of a living organism – say, a mouse. Its ‘several parts’ are all constructed (and appear to be designed) for a purpose. For instance, the lenses in its eyes have a purpose similar to that of a telescope, of focusing light to form an image on its retina, which in turn has the purpose of recognizing food, danger and so on.

Actually, Paley did not know the overall purpose of the mouse (though we do now – see ‘Neo-Darwinism’ below). But even a single eye would suffice to make Paley’s triumphant point – namely that the evidence of apparent design for a purpose is not only that the parts all serve that purpose, but that if they were slightly altered they would serve it less well, or not at all. A good design is hard to vary:

If the different parts had been differently shaped from what they are, of a different size from what they are, or placed after any other manner, or in any other order, than that in which they are placed, either no motion at all would have been carried on in the machine, or none which would have answered the use that is now served by it.

Merely being useful for a purpose, without being hard to vary while still serving that purpose, is not a sign of adaptation or design. For instance, one can also use the sun to keep time, but all its features would serve that purpose equally well if slightly (or even massively) altered. Just as we transform many of the Earth’s non-adapted raw materials to meet our purposes, so we also find uses for the sun that it was never designed or adapted to provide. The knowledge, in that case, is entirely in us – and in our sundials – not in the sun. But it is embodied in the watch, and in the mouse.

So, how did all that knowledge come to be embodied in those things? As I said, Paley could conceive of only one explanation. That was his first mistake:

The inference we think is inevitable, that the watch must have had a maker . . . There cannot be design without a designer; contrivance without a contriver; order without choice; arrangement without anything capable of arranging; subserviency and relation to a purpose without that which could intend a purpose; means suitable to an end . . . without the end ever having been contemplated or the means accommodated to it. Arrangement, disposition of parts, subserviency of means to an end, relation of instruments to a use imply the presence of intelligence and mind.

We now know that there can be ‘design without a designer’: knowledge without a person who created it. Some types of knowledge can be created by evolution. I shall come to that shortly. But it is no criticism of Paley that he was unaware of a discovery that had yet to be made – one of the greatest discoveries in the history of science.

However, although Paley was spot on in his understanding of the problem, he somehow did not realize that his proposed solution, creationism, does not solve it, and is even ruled out by his own argument. For the ultimate designer for whose existence Paley was arguing would also be a purposeful and complex entity – certainly no less so than a watch or a living organism. Hence, as many critics have since noticed, if we substitute ‘ultimate designer’ for ‘watch’ in Paley’s text above, we force Paley to ‘the [inevitable] inference . . . that the ultimate designer must have had a maker’. Since that is a contradiction, the argument from design as perfected by Paley rules out the existence of an ultimate designer.

Note that this is not a disproof of the existence of God, any more than the original argument was a proof. But it does show that, in any good explanation of the origin of biological adaptations, God cannot play the role assigned by creationism. Though this is the opposite of what Paley believed he had achieved, none of us can choose what our ideas imply. His argument has universal reach for anything that has, by his criterion, the appearance of design. As an elucidation of the special status of living things, and in setting a benchmark that explanations of knowledge-laden entities must meet if they are to make sense, it is essential to understanding the world.


Before Darwin’s theory of evolution, people had already been wondering whether the biosphere and its adaptations might have come into existence gradually. Darwin’s grandfather Erasmus Darwin (1731–1802), a stalwart of the Enlightenment, was among them. They called that process ‘evolution’, but the meaning of the word then was different from its primary one today. All processes of gradual improvement, regardless of their mechanism, were known as ‘evolution’. (That terminology survives to this day in casual usage and as a technical term in, of all places, theoretical physics, where ‘evolution’ means any sort of continuous change that one is explaining through laws of physics.) Charles Darwin distinguished the process that he discovered by calling it ‘evolution by natural selection’ – though a better name would have been ‘evolution by variation and selection’.

As Paley might well have recognized if he had lived to hear of it, ‘evolution by natural selection’ is a much more substantive mode of explanation than mere ‘evolution’. For the latter does not solve his problem, while the former does. Any theory about improvement raises the question: how is the knowledge of how to make that improvement created? Was it already present at the outset? The theory that it was is creationism. Did it ‘just happen’? The theory that it did is spontaneous generation.

During the early years of the nineteenth century, the naturalist Jean-Baptiste Lamarck proposed an answer that is now known as Lamarckism. Its key idea is that improvements acquired by an organism during its lifetime can be inherited by its offspring. Lamarck was thinking mainly of improvements in the organism’s organs, limbs and so on – such as, for instance, the enlargement and strengthening of muscles that an individual uses heavily, and the weakening of those that it seldom uses. This ‘use-and-disuse’ explanation had also been arrived at independently by Erasmus Darwin. A classic Lamarckian explanation is that giraffes, when eating leaves from trees whose lower-lying leaves were already eaten, stretched their necks to get at the higher ones. This supposedly lengthened their necks slightly, and then their offspring inherited the trait of having slightly longer necks. Thus, over many generations, long-necked giraffes evolved from ancestors with unremarkable necks. In addition, Lamarck proposed that improvements were driven by a tendency, built into the laws of nature, towards ever greater complexity.

The latter is a fudge, for not just any complexity could account for the evolution of adaptations: it has to be knowledge. And so that part of the theory is just invoking spontaneous generation – unexplained knowledge. Lamarck might not have minded that, because, like many thinkers of his day, he took the existence of spontaneous generation for granted. He even incorporated it explicitly into his theory of evolution: he guessed that, as successive generations of organisms are forced by his law of nature to take ever more complex forms, we still see simple creatures because a continuous supply of them is formed spontaneously.

Some have considered this a pretty vision. But it bears hardly any resemblance to the facts. Its most glaring mismatch is that, in reality, evolutionary adaptations are of a wholly different character from the changes that take place in an individual during its lifetime. The former involve the creation of new knowledge; the latter happen only when there is already an adaptation for making that change. For instance, the tendency of muscles to become stronger or weaker with use and disuse is controlled by a sophisticated (knowledge-laden) set of genes. The animal’s distant ancestors did not have those genes. Lamarckism cannot possibly explain how the knowledge in them was created.

If you were starved of vitamin C, your defective vitamin-C-synthesis gene would not thereby be caused to improve – unless, perhaps, you are a genetic engineer. If a tiger is placed in a habitat in which its colouration makes it stand out more instead of less, it takes no action to change the colour of its fur, nor would that change be inherited if it did. That is because nothing in the tiger ‘knows’ what the stripes are for. So how would any Lamarckian mechanism have ‘known’ that having fur that was a tiny bit more striped would slightly improve the animal’s food supply? And how would it have ‘known’ how to synthesize pigments, and to secrete them into the fur, in such a way as to produce stripes of a suitable design?

The fundamental error being made by Lamarck has the same logic as inductivism. Both assume that new knowledge (adaptations and scientific theories respectively) is somehow already present in experience, or can be derived mechanically from experience. But the truth is always that knowledge must be first conjectured and then tested. That is what Darwin’s theory says: first, random mutations happen (they do not take account of what problem is being solved); then natural selection discards the variant genes that are less good at causing themselves to be present again in future generations.


The central idea of neo-Darwinism is that evolution favours the genes that spread best through the population. There is much more to this idea than meets the eye, as I shall explain.

A common misconception about Darwinian evolution is that it maximizes ‘the good of the species’. That provides a plausible, but false, explanation of apparently altruistic behaviour in nature, such as parents risking their lives to protect their young, or the strongest animals going to the perimeter of a herd under attack – thereby decreasing their own chances of having a long and pleasant life or further offspring. Thus, it is said, evolution optimizes the good of the species, not the individual. But, in reality, evolution optimizes neither.

To see why, consider this thought experiment. Imagine an island on which the total number of birds of a particular species would be maximized if they nested at, say, the beginning of April. The explanation for why a particular date is optimal will refer to various trade-offs involving factors such as temperature, the prevalence of predators, the availability of food and nesting materials, and so on. Suppose that initially the whole population has genes that cause them to nest at that optimum time. That would mean that those genes were well adapted to maximizing the number of birds in the population – which one might call ‘maximizing the good of the species’.

Now suppose that this equilibrium is disturbed by the advent of a mutant gene in a single bird which causes it to nest slightly earlier – say, at the end of March. Assume that when a bird has built a nest, the species’ other behavioural genes are such that it automatically gets whatever cooperation it needs from a mate. That pair of birds would then be guaranteed the best nesting site on the island – an advantage which, in terms of the survival of their offspring, might well outweigh all the slight disadvantages of nesting earlier. In that case, in the following generation, there will be more March-nesting birds, and, again, all of them will find excellent nesting sites. That means that a smaller proportion than usual of the April-nesting variety will find good sites: the best sites will have been taken by the time they start looking. In subsequent generations, the balance of the population will keep shifting towards the March-nesting variants. If the relative advantage of having the best nesting sites is large enough, the April-nesting variant could even become extinct. If it arises again as a mutation, its holder will have no offspring, because all sites will have been taken by the time it tries to nest.

Thus the original situation that we imagined – with genes that were optimally adapted to maximizing the population (‘benefiting the species’) – is unstable. There will be evolutionary pressure to make the genes become less well adapted to that function.

This change has harmed the species, in the sense of reducing its total population (because the birds are no longer nesting at the optimum time). It may thereby also have harmed it by increasing the risk of extinction, making it less likely to spread to other habitats, and so on. So an optimally adapted species may in this way evolve into one that is less ‘well off’ by any measure.

If a further mutant gene then appears, causing nesting still earlier in March, the same process may be repeated, with the earlier-nesting genes taking over and the total population falling again. Evolution will thus drive the nesting time ever earlier, and the population lower. A new equilibrium would be reached only when the advantage to an individual bird’s offspring of getting the very best nesting site was finally outweighed by the disadvantages of slightly earlier nesting. That equilibrium might be very far from what was optimal for the species.

A related misconception is that evolution is always adaptive – that it always constitutes progress, or at least some sort of improvement in useful functionality which it then acts to optimize. This is often summed up in a phrase due to the philosopher Herbert Spencer, and unfortunately taken up by Darwin himself: ‘the survival of the fittest’. But, as the above thought experiment illustrates, that is not the case either. Not only has the species been harmed by this evolutionary change, every individual bird has been harmed as well: the birds using any particular site now have a harsher life than before, because they are using it earlier in the year.

Thus, although the existence of progress in the biosphere is what the theory of evolution is there to explain, not all evolution constitutes progress, and no (genetic) evolution optimizes progress.

What exactly has the evolution of those birds achieved during that period? It has optimized not the functional adaptation of a variant gene to its environment – the attribute that would have impressed Paley – but the relative ability of the surviving variant to spread through the population. An April-nesting gene is no longer able to propagate itself to the next generation, even though it is functionally the best variant. The early-nesting gene that replaced it may still be tolerably functional, but it is fittest for nothing except preventing variants of itself from procreating. From the point of view of both the species and all its members, the change brought about by this period of its evolution has been a disaster. But evolution does not ‘care’ about that. It favours only the genes that spread best through the population.

Evolution can even favour genes that are not just suboptimal, but wholly harmful to the species and all its individuals. A famous example is the peacock’s large, colourful tail, which is believed to diminish the bird’s viability by making it harder to evade predators, and to have no useful function at all. Genes for prominent tails dominate simply because peahens tend to choose prominent-tailed males as mates. Why was there selection pressure in favour of such preferences? One reason is that, when females mated with prominent-tailed males, their male offspring, having more prominent tails, found more mates. Another may be that an individual able to grow a large, colourful tail is more likely to be healthy. In any case, the net effect of all the selection pressures was to spread genes for large, colourful tails, and genes for preferring such tails, through the population. The species and the individuals just had to suffer the consequences.

If the best-spreading genes impose sufficiently large disadvantages on the species, the species becomes extinct. Nothing in biological evolution prevents that. It has presumably happened many times in the history of life on Earth, to species less lucky than the peacock. Dawkins named his tour-de-force account of neo-Darwinism The Selfish Gene because he wanted to stress that evolution does not especially promote the ‘welfare’ of species or individual organisms. But, as he also explained, it does not promote the ‘welfare’ of genes either: it adapts them not for survival in larger numbers, nor indeed for survival at all, but only for spreading through the population at the expense of rival genes, particularly slight variants of themselves.

Is it sheer luck, then, that most genes do usually confer some, albeit less than optimal, functional benefits on their species, and on their individual holders? No. Organisms are the slaves, or tools, that genes use to achieve their ‘purpose’ of spreading themselves through the population. (That is the ‘purpose’ that Paley and even Darwin never guessed.) Genes gain advantages over each other in part by keeping their slaves alive and healthy, just as human slave owners did. Slave owners were not working for the benefit of their workforces, nor for the benefit of individual slaves: it was solely to achieve their own objectives that they fed and housed their slaves, and indeed forced them to reproduce. Genes do much the same thing.

In addition, there is the phenomenon of reach: when the knowledge in a gene happens to have reach, it will help the individual to help itself in a wider range of circumstances, and by more, than the spreading of the gene strictly requires. That is why mules stay alive even though they are sterile. So it is not surprising that genes usually confer some benefits on their species and its members, and do often succeed in increasing their own absolute numbers. Nor should it be surprising that they sometimes do the opposite. But what genes are adapted to – what they do better than almost any variant of themselves – has nothing to do with the species or the individuals or even their own survival in the long run. It is getting themselves replicated more than rival genes.

Neo-Darwinism and knowledge

Neo-Darwinism does not refer, at its fundamental level, to anything biological. It is based on the idea of a replicator (anything that contributes causally to its own copying).* For instance, a gene conferring the ability to digest a certain type of food causes the organism to remain healthy in some situations where it would otherwise weaken or die. Hence it increases the organism’s chances of having offspring in the future, and those offspring would inherit, and spread, copies of the gene.

Ideas can be replicators too. For example, a good joke is a replicator: when lodged in a person’s mind, it has a tendency to cause that person to tell it to other people, thus copying it into their minds. Dawkins coined the term memes (rhymes with ‘dreams’) for ideas that are replicators. Most ideas are not replicators: they do not cause us to convey them to other people. Nearly all long-lasting ideas, however, such as languages, scientific theories and religious beliefs, and the ineffable states of mind that constitute cultures such as being British, or the skill of performing classical music, are memes (or ‘memeplexes’ – collections of interacting memes). I shall say more about memes in Chapter 15.

The most general way of stating the central assertion of the neo-Darwinian theory of evolution is that a population of replicators subject to variation (for instance by imperfect copying) will be taken over by those variants that are better than their rivals at causing themselves to be replicated. This is a surprisingly deep truth which is commonly criticized either for being too obvious to be worth stating or for being false. The reason, I think, is that, although it is self-evidently true, it is not self-evidently the explanation of specific adaptations. Our intuition prefers explanations in terms of function or purpose: what does a gene do for its holder, or for its species? But we have just seen that the genes generally do not optimize such functionality.

So the knowledge embodied in genes is knowledge of how to get themselves replicated at the expense of their rivals. Genes often do this by imparting useful functionality to their organism, and in those cases their knowledge incidentally includes knowledge about that functionality. Functionality, in turn, is achieved by encoding, into genes, regularities in the environment and sometimes even rule-of-thumb approximations to laws of nature, in which case the genes are incidentally encoding that knowledge too. But the core of the explanation for the presence of a gene is always that it got itself replicated more than its rival genes.

Non-explanatory human knowledge can also evolve in an analogous way: rules of thumb are not passed on perfectly to the next generation of users, and the ones that survive in the long run are not necessarily the ones that optimize the ostensible function. For instance, a rule that is expressed in an elegant rhyme may be remembered, and repeated, better than one that is more accurate but expressed in ungainly prose. Also, no human knowledge is entirely non-explanatory. There is always at least a background of assumptions about reality against which the meaning of a rule of thumb is understood, and that background can make some false rules of thumb seem plausible.

Explanatory theories evolve through a more complicated mechanism. Accidental errors in transmission and memory still play a role, but a much smaller one. That is because good explanations are hard to vary even without being tested, and hence random errors in the transmission of a good explanation are easier for the receiver to detect and correct. The most important source of variation in explanatory theories is creativity. For instance, when people are trying to understand an idea that they hear from others, they typically understand it to mean what makes most sense to them, or what they are most expecting to hear, or what they fear to hear, and so on. Those meanings are conjectured by the listener or reader, and may differ from what the speaker or writer intended. In addition, people often try to improve explanations even when they have received them accurately: they make creative amendments, spurred by their own criticism. If they then pass the explanation on to others, they usually try to pass on what they consider to be the improved version.

Unlike genes, many memes take different physical forms every time they are replicated. People rarely express ideas in exactly the same words in which they heard them. They also translate from one language to another, and between spoken and written language, and so on. Yet we rightly call what is transmitted the same idea – the same meme – throughout. Thus, in the case of most memes, the real replicator is abstract: it is the knowledge itself. This is in principle true of genes as well: biotechnology routinely transcribes genes into the memories of computers, where they are stored in a different physical form. Those records could be translated back into DNA strands and implanted in different animals. The only reason this is not yet a common practice is that it is easier to copy the original gene. But one day the genes of a rare species could survive its extinction by causing themselves to be stored on a computer and then implanted into a cell of a different species. I say ‘causing themselves to be stored’ because the biotechnologists would not be recording information indiscriminately, but only information that met a criterion such as ‘gene of an endangered species’. The ability to interest biotechnologists in this way would then be part of the reach of the knowledge in those genes.

So, both human knowledge and biological adaptations are abstract replicators: forms of information which, once they are embodied in a suitable physical system, tend to remain so while most variants of them do not.

The fact that the principles of neo-Darwinist theory are, from a certain perspective, self-evident has itself been used as a criticism of the theory. For instance, if the theory must be true, how can it be testable? One reply, often attributed to Haldane, is that the whole theory would be refuted by the discovery of a single fossilized rabbit in a stratum of Cambrian rock. However, that is misleading. The import of such an observation would depend on what explanations were available under the given circumstances. For instance, misidentifications of fossils, and of strata, have sometimes been made and would have to be ruled out by good explanations before one could call the discovery ‘a fossilized rabbit in Cambrian rock’.

Even given such explanations, what would have been ruled out by the rabbit would be not the theory of evolution itself, but only the prevailing theory of the history of life and geological processes on Earth. Suppose, for instance, that there was a prehistoric continent, isolated from the others, on which evolution happened several times as fast as elsewhere, and that, by convergent evolution, a rabbit-like creature evolved there during the Cambrian era; and suppose that the continents were later connected by a catastrophe that obliterated most of the life forms on that continent and submerged their fossils. The rabbit-like creature was a rare survivor which became extinct soon afterwards. Given the supposed evidence, that is still an infinitely better explanation than, for instance, creationism or Lamarckism, neither of which gives any account of the origin of the apparent knowledge in the rabbit.

So what would refute the Darwinian theory of evolution? Evidence which, in the light of the best available explanation, implies that knowledge came into existence in a different way. For instance, if an organism was observed to undergo only (or mainly) favourable mutations, as predicted by Lamarckism or spontaneous generation, then Darwinism’s ‘random variation’ postulate would be refuted. If organisms were observed to be born with new, complex adaptations – for anything – of which there were no precursors in their parents, then the gradual-change prediction would be refuted and so would Darwinism’s mechanism of knowledge-creation. If an organism was born with a complex adaptation that has survival value today, yet was not favoured by selection pressure in its ancestry (say, an ability to detect and use internet weather forecasts to decide when to hibernate), then Darwinism would again be refuted. A fundamentally new explanation would be needed. Facing more or less the same unsolved problem that Paley and Darwin faced, we should have to set about finding an explanation that worked.


The physicist Brandon Carter calculated in 1974 that if the strength of the interaction between charged particles were a few per cent smaller, no planets would ever have formed and the only condensed objects in the universe would be stars; and if it were a few per cent greater, then no stars would ever explode, and so no elements other than hydrogen and helium would exist outside them. In either case there would be no complex chemistry and hence presumably no life.

Another example: if the initial expansion rate of the universe at the Big Bang had been slightly higher, no stars would have formed and there would be nothing in the universe but hydrogen – at an extremely low and ever-decreasing density. If it had been slightly lower, the universe would have recollapsed soon after the Big Bang. Similar results have been since obtained for other constants of physics that are not determined by any known theory. For most, if not all of them, it seems that if they had been slightly different, there would have been no possibility for life to exist.

This is a remarkable fact which has even been cited as evidence that those constants were intentionally fine-tuned, i.e. designed, by a supernatural being. This is a new version of creationism, and of the design argument, now based on the appearance of design in the laws of physics. (Ironically, given the history of this controversy, the new argument is that the laws of physics must have been designed to create a biosphere by Darwinian evolution.) It even persuaded the philosopher Antony Flew – formerly an enthusiastic advocate of atheism – of the existence of a supernatural designer. But it should not have. As I shall explain in a moment, it is not even clear that this fine-tuning constitutes an appearance of design in Paley’s sense; but, even if it does, that does not alter the fact that invoking the supernatural makes for a bad explanation. And, in any case, arguing for supernatural explanations on the grounds that a current scientific explanation is flawed or lacking is just a mistake. As we carved in stone in Chapter 3, problems are inevitable – there are always unsolved problems. But they get solved. Science continues to make progress even, or especially, after making great discoveries, because the discoveries themselves reveal further problems. Therefore the existence of an unsolved problem in physics is no more evidence for a supernatural explanation than the existence of an unsolved crime is evidence that a ghost committed it.

A simple objection to the idea that fine-tuning requires an explanation at all is that we have no good explanation implying that planets are essential to the formation of life, or that chemistry is. The physicist Robert Forward wrote a superb science-fiction story, Dragon’s Egg, based on the premise that information could be stored and processed – and life and intelligence could evolve – through the interactions between neutrons on the surface of a neutron star (a star that has collapsed gravitationally to a diameter of only a few kilometres, making it so dense that most of its matter has been transmuted into neutrons). It is not known whether this hypothetical neutron analogue of chemistry exists – nor whether it could exist if the laws of physics were slightly different. Nor do we have any idea what other sorts of environment permitting the emergence of life would exist under those variant laws. (The idea that similar laws of physics can be expected to give rise to similar environments is undermined by the very existence of fine-tuning.)

Nevertheless, regardless of whether the fine-tuning constitutes an appearance of design or not, it does constitute a legitimate and significant scientific problem, for the following reason. If the truth is that the constants of nature are not fine-tuned to produce life after all, because most slight variations in them do still permit life and intelligence to evolve somehow, though in dramatically different types of environment, then this would be an unexplained regularity in nature and hence a problem for science to address.

If the laws of physics are fine-tuned, as they seem to be, then there are two possibilities: either those laws are the only ones to be instantiated in reality (as universes) or there are other regions of reality – parallel universes* – with different laws. In the former case, we must expect there to be an explanation of why the laws are as they are. It would either refer to the existence of life or not. If it did, that would take us back to Paley’s problem: it would mean that the laws had the ‘appearance of design’ for creating life, but had not evolved. Or the explanation would not refer to the existence of life, in which case it would leave unexplained why, if the laws are as they are for non-life-related reasons, they are fine-tuned to create life.

If there are many parallel universes, each with its own laws of physics, most of which do not permit life, then the idea would be that the observed fine-tuning is only a matter of parochial perspective. It is only in the universes that contain astrophysicists that anyone ever wonders why the constants seem fine-tuned. This type of explanation is known as ‘anthropic reasoning’. It is said to follow from a principle known as the ‘weak anthropic principle’, though really no principle is required: it is just logic. (The qualifier ‘weak’ is there because several other anthropic principles have been proposed, which are more than just logic, but they need not concern us here.)

However, on closer examination, anthropic arguments never quite finish the explanatory job. To see why, consider an argument due to the physicist Dennis Sciama.

Imagine that, at some time in the future, theoreticians have calculated, for one of those constants of physics, the range of its values for which there would be a reasonable probability that astrophysicists (of a suitable kind) would emerge. Say that range is from 137 to 138. (No doubt the real values will not be whole numbers, but let us keep it simple.) They also calculate that the highest probability of astrophysicists occurs at the midpoint of the range – when the constant is 137.5.

Next, experimentalists set out to measure the value of that constant directly – in laboratories, or by astronomical observation, say. What should they predict? Curiously enough, one immediate prediction from the anthropic explanation is that the value will not be exactly 137.5. For suppose that it were. By analogy, imagine that the bull’s-eye of a dartboard represents the values that can produce astrophysicists. It would be a mistake to predict that a typical dart that strikes the bull’s eye will strike it at the exact centre. Likewise, in the overwhelming majority of universes in which the measurement could take place (because they contain astrophysicists), the constant would not take the exactly optimal value for producing astrophysicists, nor be extremely close to it, compared with the size of the bull’s-eye.

So Sciama concludes that, if we did measure one of those constants of physics, and found that it was extremely close to the optimum value for producing astrophysicists, that would statistically refute, not corroborate, the anthropic explanation for its value. Of course that value might still be a coincidence, but if we were willing to accept astronomically unlikely coincidences as explanations we should not be puzzled by the fine-tuning in the first place – and we should tell Paley that the watch on the heath might just have been formed by chance.

Furthermore, astrophysicists should be relatively unlikely in universes whose conditions are so hostile that they barely permit astrophysicists at all. So, if we imagine all the values consistent with the emergence of astrophysicists arrayed on a line, then the anthropic explanation leads us to expect the measured value to fall at some typical point, not too close to the middle or to either end.

However – and here we are reaching Sciama’s main conclusion – that prediction changes radically if there are several constants to explain. For although any one constant is unlikely to be near the edge of its range, the more constants there are, the more likely it is that at least one of them will be. This can be illustrated pictorially as follows, with our bull’s-eye replaced by a line segment, a square, a cube . . . and we can imagine this sequence continuing for as many dimensions as there are fine-tuned constants in nature. Arbitrarily define ‘near the edge’ as meaning ‘within 10 per cent of the whole range from it’. Then in the case of one constant, as shown in the diagram, 20 per cent of its possible values are near one of the two edges of the range, and 80 per cent are ‘away from the edge’. But with two constants a pair of values has to satisfy two constraints in order to be ‘away from the edge’. Only 64 per cent of them do so. Hence 36 per cent are near the edge. With three constants, nearly half the possible choices are near the edge. With 100 constants, over 99.9999999 per cent of them are.


Whatever anthropic reasoning predicts about the values of multiple constants, it predicts will only just happen.

So, the more constants are involved, the closer to having no astrophysicists a typical universe-with-astrophysicists is. It is not known how many constants are involved, but it seems to be several, in which case the overwhelming majority of universes in the anthropically selected region would be close to its edge. Hence, Sciama concluded, the anthropic explanation predicts that the universe is only just capable of producing astrophysicists – almost the opposite prediction from the one that it makes in the case of one constant.

On the face of it, this might in turn seem to explain another great unsolved scientific mystery, known as ‘Fermi’s problem’, named after the physicist Enrico Fermi, who is said to have asked, ‘Where are they?’ Where are the extraterrestrial civilizations? Given the Principle of Mediocrity, or even just what we know of the galaxy and the universe, there is no reason to believe that the phenomenon of astrophysicists is unique to our planet. Similar conditions presumably exist in many other solar systems, so why would some of them not produce similar outcomes? Moreover, given the timescales on which stars and galaxies develop, it is overwhelmingly unlikely that any given extraterrestrial civilization is currently at a similar state of technological development to ours: it is likely to be millions of years younger (i.e. non-existent) or older. The older civilizations have had plenty of time to explore the galaxy – or at least to send robot space probes or signals. Fermi’s problem is that we do not see any such civilizations, probes or signals.

Many candidate explanations have been proposed, and none of them, so far, are very good. The anthropic explanation of fine-tuning, in the light of Sciama’s argument, might seem to solve the problem neatly: if the constants of physics in our universe are only just capable of producing astrophysicists, then it is not surprising that this event has happened only once, since its happening twice independently in the same universe would be vanishingly unlikely.

Unfortunately, that turns out to be a bad explanation too, because focusing on fundamental constants is parochial: there is no relevant difference between (1) ‘the same’ laws of physics with different constants and (2) different laws of physics. And there are infinitely many logically possible laws of physics. If they were all instantiated in real universes – as has been suggested by some cosmologists, such as Max Tegmark – it would be statistically certain that our universe is exactly on the edge of the astrophysicist-producing class of universes.

We know that that cannot be so from an argument due to Feynman (which he applied to a slightly different problem). Consider the class of all possible universes that contain astrophysicists, and consider what elsemost of them contain. In particular, consider a sphere just large enough to contain your own brain. If you are interested in explaining fine-tuning, your brain in its current state counts as an ‘astrophysicist’ for these purposes. In the class of all universes that contain astrophysicists, there are many that contain a sphere whose interior is perfectly identical to the interior of your sphere, including every detail of your brain. But in the vast majority of those universes there is chaos outside the sphere: almost a random state, since almost-random states are by far the most numerous. A typical such state is not only amorphous but hot. So in most such universes the very next thing that is going to happen is that the chaotic radiation emanating from outside the sphere will kill you instantly. At any given instant, the theory that we are going to be killed a picosecond hence is refuted by observation a picosecond later. Whereupon another such theory presents itself. So it is a very bad explanation – an extreme version of the gambler’s hunches.

The same holds for purely anthropic explanations of all other fine-tunings involving more than a handful of constants: such explanations predict that it is overwhelmingly likely that we are in a universe in which astrophysicists are only just possible and will cease to exist in an instant. So they are bad explanations.

On the other hand, if the laws of physics exist in only one form, with only the values of a few constants differing from one universe to another, then the very fact that laws with different forms are not instantiated is a piece of fine-tuning that that anthropic explanation leaves unexplained.

The theory that all logically possible laws of physics are instantiated as universes has a further severe problem as an explanation. As I shall explain in Chapter 8, when considering infinite sets such as these, there is often no objective way to ‘count’ or ‘measure’ how many of them have one attribute rather than another. On the other hand, in the class of all logically possible entities, those that can understand themselves, as the physical reality that we are in does, are surely, in any reasonable sense, a tiny minority. The idea that one of them ‘just happened’, without explanation, is surely just a spontaneous-generation theory.

In addition, almost all the ‘universes’ described by those logically possible laws of physics are radically different from ours – so different that they do not properly fit into the argument. For instance, infinitely many of them contain nothing other than one bison, in various poses, and last for exactly 42 seconds. Infinitely many others contain a bison and an astrophysicist. But what is an astrophysicist in a universe that contains no stars, no scientific instruments and almost no evidence? What is a scientist, or any sort of thinking person, in a universe in which only bad explanations are true?

Almost all logically possible universes that contain astrophysicists are governed by laws of physics that are bad explanations. So should we predict that our universe, too, is inexplicable? Or has some high but unknowable probability to be? Thus, again, anthropic arguments based on ‘all possible laws’ are ruled out for being bad explanations.

For these reasons I conclude that, while anthropic reasoning may well be part of the explanation for apparent fine-tuning and other observations, it can never be the whole explanation for why we observe something that would otherwise look too purposeful to be explicable as coincidence. Specific explanation, in terms of specific laws of nature, is needed.

The reader may have noticed that all the bad explanations that I have discussed in this chapter are ultimately connected with each other. Expect too much from anthropic reasoning, or wonder too carefully how Lamarckism could work, and you get to spontaneous generation. Take spontaneous generation too seriously, and you get to creationism – and so on. That is because they all address the same underlying problem, and are all easily variable. They are easily interchangeable with each other or with variants of themselves, and they are ‘too easy’ as explanations: they could equally well explain anything. But neo-Darwinism was not easy to come by, and it is not easy to tweak. Try to tweak it – even as far as Darwin’s own misconceptions – and you will get an explanation that doesn’t work nearly as well. Try to account for something non-Darwinian with it – such as a new, complex adaptation of which there were no precursors in the organism’s parents – and you will not be able to think of a variant with that feature.

Anthropic explanations are attempting to account for purposeful structure (such as the fine-tuned constants) in terms of a single act of selection. That is unlike evolution, and it cannot work. The solution of the fine-tuning puzzle is going to be in terms of an explanation that will specifically explain what we observe. It will be, as Wheeler put it, ‘an idea so simple . . . that . . . we will all say to each other, how could it have been otherwise?’ In other words, the problem has been not that the world is so complex that we cannot understand why it looks as it does, but it is that it is so simple that we cannot yet understand it. But this will be noticeable only with hindsight.

All those bad explanations of the biosphere either fail to address the problem of how the knowledge in adaptations is created or they explain it badly. That is to say, they all underrate creation – and, ironically, the theory that underrates creation most of all is creationism. Consider this: if a supernatural creator were to have created the universe at the moment when Einstein or Darwin or any great scientist (appeared to have) just completed their major discovery, then the true creator of that discovery (and of all earlier discoveries) would have been not that scientist but the supernatural being. So such a theory would deny the existence of the only creation that really did take place in the genesis of that scientist’s discoveries.

And it really is creation. Before a discovery is made, no predictive process could reveal the content or the consequences of that discovery. For if it could, it would be that discovery. So scientific discovery is profoundly unpredictable, despite the fact that it is determined by the laws of physics. I shall say more about this curious fact in the next chapter; in short, it is due to the existence of ‘emergent’ levels of explanation. In this case, the upshot is that what science – and creative thought in general – achieves is unpredictable creation ex nihilo. So does biological evolution. No other process does.

Creationism, therefore, is misleadingly named. It is not a theory explaining knowledge as being due to creation, but the opposite: it is denying that creation happened in reality, by placing the origin of the knowledge in an explanationless realm. Creationism is really creation denial – and so are all those other false explanations.

The puzzle of understanding what living things are and how they came about has given rise to a strange history of misconceptions, near-misses and ironies. The last of the ironies is that the neo-Darwinian theory, like the Popperian theory of knowledge, really does describe creation, while their rivals, beginning with creationism, never could.


Evolution (Darwinian)   Creation of knowledge through alternating variation and selection.

Replicator   An entity that contributes causally to its own copying.

Neo-Darwinism   Darwinism as a theory of replicators, without various misconceptions such as ‘survival of the fittest’.

Meme   An idea that is a replicator.

Memeplex   A group of memes that help to cause each other’s replication.

Spontaneous generation   Formation of organisms from non-living precursors.

Lamarckism   A mistaken evolutionary theory based on the idea that biological adaptations are improvements acquired by an organism during its lifetime and then inherited by its descendants.

Fine-tuning   If the constants or laws of physics were slightly different, there would be no life.

Anthropic explanation   ‘It is only in universes that contain intelligent observers that anyone wonders why the phenomenon in question happens.’


– Evolution.

– More generally, the creation of knowledge.


The evolution of biological adaptations and the creation of human knowledge share deep similarities, but also some important differences. The main similarities: genes and ideas are both replicators; knowledge and adaptations are both hard to vary. The main difference: human knowledge can be explanatory and can have great reach; adaptations are never explanatory and rarely have much reach beyond the situations in which they evolved. False explanations of biological evolution have counterparts in false explanations of the growth of human knowledge. For instance, Lamarckism is the counterpart of inductivism. William Paley’s version of the argument from design clarified what does or does not have the ‘appearance of design’ and hence what cannot be explained as the outcome of chance alone – namely hard-to-vary adaptation to a purpose. The origin of this must be the creation of knowledge. Biological evolution does not optimize benefits to the species, the group, the individual or even the gene, but only the ability of the gene to spread through the population. Such benefits can nevertheless happen because of the universality of laws of nature and the reach of some of the knowledge that is created. The ‘fine-tuning’ of the laws or constants of physics has been used as a modern form of the argument from design. For the usual reasons, it is not a good argument for a supernatural cause. But ‘anthropic’ theories that try to account for it as a pure selection effect from an infinite number of different universes are, by themselves, bad explanations too – in part because most logically possible laws are themselves bad explanations.