ENGINES of CREATION | Chapter 2: The Principles of Change
February 21, 2001
- K. Eric Drexler
Think of the design process as involving first the generation of alternatives and then the testing of these alternatives against a whole array of requirements and constraints.
HERBERT A. SIMON
MOLECULAR ASSEMBLERS will bring a revolution without parallel since the development of ribosomes, the primitive assemblers in the cell. The resulting nanotechnologycan help life spread beyond Earth – a step without parallel since life spread beyond the seas. It can help mind emerge in machines – a step without parallel since mind emerged in primates. And it can let our minds renew and remake our bodies – a step without any parallel at all.
These revolutions will bring dangers and opportunities too vast for the human imagination to grasp. Yet the principles of change that have applied to molecules, cells, beasts, minds, and machines should endure even in an age of biotechnology, nanomachines, and artificial minds. The same principles that have applied at sea, on land, and in the air should endure as we spread Earth’s life toward the stars. Understanding the enduring principles of change will help us understand the potential for good and ill in the new technologies.
Order from Chaos
Order can emerge from chaos without anyone’s giving orders: orderly crystals condensed from formless interstellar gas long before Sun, Earth, or life appeared.
Chaos also gives rise to a crystalline order under more familiar circumstances. Imagine a molecule – perhaps regular in form, or perhaps lopsided and knobby like a ginger root. Now imagine a vast number of such molecules moving randomly in a liquid, tumbling and jostling like drunkards in weightlessness in the dark. Imagine the liquid evaporating and cooling, forcing the molecules closer together and slowing them down. Will these randomly moving, oddly shaped molecules simply gather in disordered heaps? Generally not. They will usually settle into a crystalline pattern, each neatly nestled against its neighbors, forming rows and columns as perfect as a checkerboard, though often more complex.
This process involves neither magic nor some special property of molecules and quantum mechanical forces. It does not even require the special matching shapes that enable protein molecules to self-assemble into machines. Marbles of uniform size, if placed in a tray and shaken, also settle into a regular pattern.
Crystals grow by trial and the removal of error, by variation and selection. No tiny hands assemble them. A crystal can begin with a chance clumping of molecules: the molecules wander, bump, and clump at random, but clumps stick best when packed in the right crystalline pattern. Other molecules then strike this first, tiny crystal. Some bump in the wrong position or orientation; they stick poorly and shake loose again. Others happen to bump properly; they stick better and often stay. Layer builds on layer, extending the crystalline pattern. Though the molecules bump at random, they do not stick at random. Order grows from chaos through variation and selection.
In crystal growth, each layer forms a template for the next. Uniform layers accumulate to form a solid block.
In cells, strands of DNA or RNA can serve as templates too, aided by enzymes that act as molecular copying machines. But the subunits of nucleic acid strands can be arranged in many different sequences, and a template strand can separate from its copy. Both strand and copy can then be copied again. Biochemist Sol Spiegelman has used a copying machine (a protein from a virus) in test tube experiments. In a simple, lifeless environment, it duplicates RNA molecules.
Picture a strand of RNA floating in a test tube together with copying machines and RNA subunits. The strand tumbles and writhes until it bumps into a copying machine in the right position to stick. Subunits bump around until one of the right kind meets the copying machine in the right position to match the template strand. As matching subunits chance to fall into position, the machine seizes them and bonds them to the growing copy; though subunits bump randomly, the machine bonds selectively. Finally the machine, the template, and the copy separate.
In the terminology of Oxford zoologist Richard Dawkins, things that give rise to copies of themselves are called replicators. In this environment, RNA molecules qualify: a single molecule soon becomes two, then four, eight, sixteen, thirty-two, and so forth, multiplying exponentially. Later, the replication rate levels off: the fixed stock of protein machines can churn out RNA copies only so fast, no matter how many template molecules vie for their services. Later still, the raw materials for making RNA molecules become scarce and replication starves to a halt. The exploding population of molecules reaches a limit to growth and stops reproducing.
The copying machines, however, often miscopy an RNA strand, inserting, deleting, or mismatching a subunit. The resulting mutated strand then differs in length or subunit sequence. Such changes are fairly random, and changes accumulate as miscopied molecules are again miscopied. As the molecules proliferate, they begin to grow different from their ancestors and from each other. This might seem a recipe for chaos.
Biochemists have found that differing RNA molecules replicate at differing rates, depending on their lengths and subunit patterns. Descendants of the swifter replicators naturally grow more common. Indeed, if one kind replicates just 10 percent more rapidly than its siblings, then after one hundred generations, each of the faster kind gives rise to 1,000 times as many descendants. Small differences in exponential growth pile up exponentially.
When a test tube runs out of subunits, an experimenter can sample its RNA and “infect” a fresh tube. The process begins again and the molecules that dominated the first round of competition begin with a head start. More small changes appear, building over time into large changes. Some molecules replicate faster, and their kind dominates the mix. When resources run out, the experimenter can sample the RNA and start again (and again, and again), holding conditions stable.
This experiment reveals a natural process: no matter what RNA sequences the experimenter starts with, the seeming chaos of random errors and biased copying brings forth one kind of RNA molecule (give or take some copying errors). Its typical version has a known, well-defined sequence of 220 subunits. It is the best RNA replicator in this environment, so it crowds out the others and stays.
Prolonged copying, miscopying, and competition always bring about the same result, no matter what the length or pattern of the RNA molecule that starts the process. Though no one could have predicted this winning pattern, anyone can see that change and competition will tend to bring forth a single winner. Little else could happen in so simple a system. If these replicators affected one another strongly (perhaps by selectively attacking or helping one another), then the result could resemble a more complex ecology. As it is, they just compete for a resource.
A variation on this example shows us something else: RNA molecules adapt differently to different environments. A molecular machine called a ribonuclease grabs RNA molecules having certain sequences of exposed subunits and cuts them in two. But RNA molecules, like proteins, fold in patterns that depend on their sequences, and by folding the right way they can protect their vulnerable spots. Experimenters find that RNA molecules evolve to sacrifice swift replication for better protection when ribonuclease is around. Again, a best competitor emerges.
Notice that biological terms have crept into this description: since the molecules replicate, the word “generation” seems right; molecules “descended” from a common “ancestor” are “relatives,” and the words “growth,” “reproduction,” “mutation,” and “competition” also seem right. Why is this? Because these molecules copy themselves with small variations, as do the genes of living organisms. When varying replicators have varying successes, the more successful tend to accumulate. This process, wherever it occurs, is “evolution.”
In this test tube example we can see evolution stripped to its bare essentials, free of the emotional controversy surrounding the evolution of life. The RNA replicators and protein copying machines are well-defined collections of atoms obeying well-understood principles and evolving in repeatable laboratory conditions. Biochemists can make RNA and protein from off-the-shelf chemicals, without help from life.
Biochemists borrow these copying machines from a kind of virus that infects bacteria and uses RNA as its genetic material. These viruses survive by entering a bacterium, getting themselves copied using its resources, and then escaping to infect new bacteria. Miscopying of viral RNA produces mutant viruses, and viruses that replicate more successfully grow more common; this is evolution by natural selection, apparently called “natural” because it involves nonhuman parts of nature. But unlike the test tube RNA, viral RNA must do more than just replicate itself as a bare molecule, Successful viral RNA must also direct bacterial ribosomes to build protein devices that let it first escape from the old bacterium, then survive outside, and finally enter a new one. This additional information makes viral RNA molecules about 4,500 subunits long.
To replicate successfully, the DNA of large organisms must do even more, directing the construction of tens of thousands of different protein machines and the development of complex tissues and organs. This requires thousands of genes coded in millions to billions of DNA subunits. Nevertheless, the essential process of evolution by variation and selection remains the same in the test tube, in viruses, and far beyond.
There are at least three ways to explain the structure of an evolved population of molecular replicators, whether test tube RNA, viral genes, or human genes. The first kind of explanation is a blow-by-blow account of their histories: how specific mutations occurred and how they spread. This is impossible without recording all the molecular events, and such a record would in any event be immensely tedious.
The second kind of explanation resorts to a somewhat misleading word: purpose. In detail, the molecules simply change haphazardly and replicate selectively. Yet stepping back from the process, one could describe the outcome by imagining that the surviving molecules have changed to “achieve the goal” of replication. Why do RNA molecules that evolved under the threat of ribonuclease fold as they do? Because of a long and detailed history, of course, but the idea that “they want to avoid attack and survive to replicate” would predict the same result. The language of purpose makes useful shorthand (try discussing human action without it!), but the appearance of purpose need not result from the action of a mind. The RNA example shows this quite neatly.
The third (and often best) kind of explanation – in terms of evolution – says that order emerges through the variation and selection of replicators. A molecule folds in a particular way because it resembles ancestors that multiplied more successfully (by avoiding attack, etc.), and left descendants including itself. As Richard Dawkins points out, the language of purpose (if used carefully) can be translated into the language of evolution.
Evolution attributes patterns of success to the elimination of unsuccessful changes. It thus explains a positive as the result of a double negative – an explanation of a sort that seems slightly difficult to grasp. Worse, it explains something visible (successful, purposeful entities) in terms of something invisible (unsuccessful entities that have vanished). Because only successful beasts have littered the landscape with the bones of their descendants, the malformed failures of the past haven’t even left many fossils.
The human mind tends to focus on the visible, seeking positive causes for positive results, an ordering force behind orderly results. Yet through reflection we can see that this great principle has changed our past and will shape our future: Evolution proceeds by the variation and selection of replicators.
The history of life is the history of an arms race based on molecular machinery. Today, as this race approaches a new and swifter phase, we need to be sure we understand just how deeply rooted evolution is. In a time when the idea of biological evolution is often slighted in the schools and sometimes attacked, we should remember that the supporting evidence is as solid as rock and as common as cells.
In pages of stone, the Earth itself has recorded the history of life. On lake bottoms and seabed, shells, bones, and silt have piled, layer on layer. Sometimes a shifting current or a geological upheaval has washed layers away; otherwise they have simply deepened. Early layers, buried deep, have been crushed, baked, soaked in mineral waters, and turned to stone.
For centuries, geologists have studied rocks to read Earth’s past. Long ago, they found seashells high in the crushed and crumpled rock of mountain ranges. By 1785 – seventy-four years before Darwin’s detested book – James Hutton had concluded that seabed mud had been pressed to stone and raised skyward by forces not yet understood. What else could geologists think, unless nature itself had lied?
They saw that fossil bones and shells differed from layer to layer. They saw that shells in layers here matched shells in layers there, though the layers might lie deep beneath the land between. They named layers (A,B,C,D…, or Osagian, Meramecian, Lower Chesterian, Upper Chesterian, . .), and used characteristic fossils to trace rock layers. The churning of Earth’s crust has nowhere left a complete sequence of layers exposed, yet geologists finding A,B,C,D,E in one place, C,D,E,F,G,H,I,J in another and J,K,L somewhere else could see that A preceded L. Petroleum geologists (even those who care nothing for evolution or its implications) still use such fossils to date rock layers and to trace layers from one drill site to another.
Scientists came to the obvious conclusion. Just as sea species today live in broad areas, so did species in years gone by. Just as layer piles on top of layer today, so did they then. Similar shells in similar layers mark sediments laid down in the same age. Shells change from layer to layer because species changed from age to age. This is what geologists found written in shells and bones on pages of stone.
The uppermost layers of rock contain bones of recent animals, deeper layers contain bones of animals now extinct. Still earlier layers show no trace of any modern species. Below mammal bones lie dinosaur bones; in older layers lie amphibian bones, then shells and fish bones, and then no bones or shells at all. The oldest fossil-bearing rocks bear the microscopic traces of single cells.
Radioactive dating shows these oldest traces to be several billion years old. Cells more complex than bacteria date to little more than one billion years ago. The history of worms, fish, amphibians, reptiles, and mammals spans hundreds of millions of years. Human-like bones date back several million years. The remains of civilizations date back several thousand.
In three billion years, life evolved from single cells able to soak up chemicals to collections of cells embodying minds able to soak up ideas. Within the last century, technology has evolved from the steam locomotive and electric light to the spaceship and the electronic computer – and computers are already being taught to read and write. With mind and technology, the rate of evolution has jumped a millionfold or more.
Another Route Back
The book of stone records the forms of long-dead organisms, yet living cells also carry records, genetic texts only now being read. As with the ideas of geology, the essential ideas of evolution were known before Darwin had set pen to paper.
In lamp-lit temples and monasteries, generations of scribes copied and recopied manuscripts. Sometimes they miscopied words and sentences – whether by accident, by perversity, or by order of the local ruler – and as the manuscripts replicated, aided by these human copying machines, errors accumulated. The worst errors might be caught and removed, and famous passages might survive unchanged, but differences grew.
Ancient books seldom exist in their original versions. The oldest copies are often centuries younger than the lost originals. Nonetheless, from differing copies with differing errors, scholars can reconstruct versions closer to the original.
They compare texts. They can trace lines of descent from common ancestors because unique patterns of errors betray copying from a common source. (Schoolteachers know this: identical right answers aren’t a tipoff – unless on an essay test – but woe to students sitting side by side who turn in tests with identical mistakes!) Where all surviving copies agree, scholars can assume that the original copy (or at least the last shared ancestor of the survivors) held the same words. Where survivors differ, scholars study copies that descended separately from a distant ancestor, because areas of agreement then indicate a common origin in the ancestral version.
Genes resemble manuscripts written in a four-letter alphabet. Much as a message can take many forms in ordinary language (restating an idea using entirely different words is no great strain), so different genetic wording can direct the construction of identical protein molecules. Moreover, protein molecules with different design details can serve identical functions. A collection of genes in a cell is like a whole book, and genes – like old manuscripts – have been copied and recopied by inaccurate scribes.
Like scholars studying ancient texts, biologists generally work with modern copies of their material (with, alas, no biological Dead Sea Scrolls from the early days of life). They compare organisms with similar appearances (lions and tigers, horses and zebras, rats and mice) and find that they give similar answers to the essay questions in their genes and proteins. The more two organisms differ (lions and lizards, humans and sunflowers), the more these answers differ, even among molecular machines serving identical functions. More telling still, similar animals make the same mistakes – all primates, for example, lack enzymes for making vitamin C, an omission shared by only two other known mammals, the guinea pig and the fruit bat. This suggests that we primates have copied our genetic answers from a shared source, long ago.
The same principle that shows the lines of descent of ancient texts (and that helps correct their copying errors) thus also reveals the lines of descent of modern life. Indeed, it indicates that all known life shares a common ancestor.
The Rise of the Replicators
The first replicators on Earth evolved abilities beyond those possible to RNA molecules replicating in test tubes. By the time they reached the bacterial stage, they had developed the “modern” system of using DNA, RNA, and ribosomes to construct protein. Mutations then changed not only the replicating DNA itself, but protein machines and the living structures they build and shape.
Teams of genes shaped ever more elaborate cells, then guided the cellular cooperation that formed complex organisms. Variation and selection favored teams of genes that shaped beasts with protective skins and hungry mouths, animated by nerve and muscle, guided by eye and brain. As Richard Dawkins puts it, genes built ever more elaborate survival machines to aid their own replication.
When dog genes replicate, they often shuffle with those of other dogs that have been selected by people, who then select which puppies to keep and breed. Over the millennia, people have molded wolf-like beasts into greyhounds, toy poodles, dachshunds, and Saint Bernards. By selecting which genes survive, people have reshaped dogs in both body and temperament. Human desires have defined success for dog genes; other pressures have defined success for wolf genes.
Mutation and selection of genes has, through long ages, filled the world with grass and trees, with insects, fish, and people. More recently, other things have appeared and multiplied – tools, houses, aircraft, and computers. And like the lifeless RNA molecules, this hardware has evolved.
As the stone of Earth records the emergence of ever more complex and capable forms of life, so the relics and writings of humanity record the emergence of ever more complex and capable forms of hardware. Our oldest surviving hardware is itself stone, buried with the fossils of our ancestors; our newest hardware orbits overhead.
Consider for a moment the hybrid ancestry of the space shuttle. On its aircraft side, it descends from the aluminum jets of the sixties, which themselves sprang from a line stretching back through the aluminum prop planes of World War II, to the wood-and-cloth biplanes of World War I, to the motorized gliders of the Wright brothers, to toy gliders and kites. On its rocket side, the shuttle traces back to Moon rockets, to military missiles, to last century’s artillery rockets (“and the rocket’s red glare…”), and finally to fireworks and toys. This aircraft/rocket hybrid flies, and by varying components and designs, aerospace engineers will evolve still better ones.
Engineers speak of “generations” of technology; Japan’s “fifth generation” computer project shows how swiftly some technologies grow and spawn. Engineers speak of “hybrids,” of “competing technologies,” and of their “proliferation.” IBM Director of Research Ralph E. Gomory emphasizes the evolutionary nature of technology, writing that “technology development is much more evolutionary and much less revolutionary or breakthrough-oriented than most people imagine.” (Indeed, even breakthroughs as important as molecular assemblers will develop through many small steps.) In the quote that heads this chapter, Professor Herbert A. Simon of Carnegie-Mellon University urges us to “think of the design process as involving first the generation of alternatives and then the testing of these alternatives against a whole array of requirements and constraints.” Generation and testing of alternatives is synonymous with variation and selection.
Sometimes various alternatives already exist. In “One Highly Evolved Toolbox,” in The Next Whole Earth Catalog, J. Baldwin writes: “Our portable shop has been evolving for about twenty years now. There’s nothing really very special about it except that a continuing process of removing obsolete or inadequate tools and replacing them with more suitable ones has resulted in a collection that has become a thing-making system rather than a pile of hardware.”
Baldwin uses the term “evolving” accurately. Invention and manufacture have for millennia generated variations in tool designs, and Baldwin has winnowed the current crop by competitive selection, keeping those that work best with his other tools to serve his needs. Through years of variation and selection, his system evolved – a process he highly recommends. Indeed, he urges that one never try to plan out the purchase of a complete set of tools. Instead, he urges buying the tools one often borrows, tools selected not by theory but by experience.
Technological variations are often deliberate, in the sense that engineers are paid to invent and test. Still, some novelties are sheer accident, like the discovery of a crude form of Teflon in a cylinder supposedly full of tetrafluoroethylene gas: with its valve open, it remained heavy; when it was sawed open, it revealed a strange, waxy solid. Other novelties have come from systematic blundering. Edison tried carbonizing everything from paper to bamboo to spider webs when he was seeking a good light-bulb filament. Charles Goodyear messed around in a kitchen for years, trying to convert gummy natural rubber into a durable substance, until at last he chanced to drop sulfurized rubber on a hot stove, performing the first crude vulcanization.
In engineering, enlightened trial and error, not the planning of flawless intellects, has brought most advances; this is why engineers build prototypes. Peters and Waterman in their book In Search of Excellence show that the same holds true of advances in corporate products and policies. This is why excellent companies create “an environment and a set of attitudes that encourage experimentation,” and why they evolve “in a very Darwinian way.”
Factories bring order through variation and selection. Crude quality-control systems test and discard faulty parts before assembling products, and sophisticated quality-control systems use statistical methods to track defects to their sources, helping engineers change the manufacturing process to minimize defects. Japanese engineers, building on W. Edwards Deming’s work in statistical quality control, have made such variation and selection of industrial processes a pillar of their country’s economic success. Assembler-based systems will likewise need to measure results to eliminate flaws.
Quality control is a sort of evolution, aiming not at change but at eliminating harmful variations. But just as Darwinian evolution can preserve and spread favorable mutations, so good quality control systems can help managers and workers to preserve and spread more effective processes, whether they appear by accident or by design.
All this tinkering by engineers and manufacturers prepares products for their ultimate test. Out in the market, endless varieties of wrench, car, sock, and computer compete for the favor of buyers. When informed buyers are free to choose, products that do too little or cost too much eventually fail to be re-produced. As in nature, competitive testing makes yesterday’s best competitor into tomorrow’s fossil. “Ecology” and “economy” share more than linguistic roots.
Both in the marketplace and on real and imaginary battlefields, global competition drives organizations to invent, buy, beg, and steal ever more capable technologies. Some organizations compete chiefly to serve people with superior goods, others compete chiefly to intimidate them with superior weapons. The pressures of evolution drive both.
The global technology race has been accelerating for billions of years. The earthworm’s blindness could not block the development of sharp-eyed birds. The bird’s small brain and clumsy wings could not block the development of human hands, minds, and shotguns. Likewise, local prohibitions cannot block advances in military and commercial technology. It seems that we must guide the technology race or die, yet the force of technological evolution makes a mockery of anti-technology movements: democratic movements for local restraint can only restrain the world’s democracies, not the world as a whole. The history of life and the potential of new technology suggest some solutions, but this is a matter for Part Three.
The Evolution of Design
It might seem that design offers an alternative to evolution, but design involves evolution in two distinct ways. First, design practice itself evolves. Not only do engineers accumulate designs that work, they accumulate design methods that work. These range from handbook standards for choosing pipes to management systems for organizing research and development. And as Alfred North Whitehead stated, “The greatest invention of the nineteenth century was the invention of the method of invention.”
Second, design itself proceeds by variation and selection, Engineers often use mathematical laws evolved to describe (for example) heat flow and elasticity to test simulated designs before building them. They thus evolve plans through a cycle of design, calculation, criticism, and redesign, avoiding the expense of cutting metal. The creation of designs thus proceeds through a nonmaterial form of evolution.
Hooke’s law, for example, describes how metal bends and stretches: deformation is proportional to the applied stress; twice the pull, twice the stretch. Though only roughly correct, it remains fairly accurate until the metal’s springiness finally yields to stress. Engineers can use a form of Hooke’s law to design a bar of metal that can support a load without bending too far – and then make it just a bit thicker to allow for inaccuracies in the law and in their design calculations. They can also use a form of Hooke’s law to describe the bending and twisting of aircraft wings, tennis rackets, and automobile frames. But simple mathematical equations don’t wrap smoothly around such convoluted structures. Engineers have to fit the equations to simpler shapes (to pieces of the design), and then assemble these partial solutions to describe the flexing of the whole. It is a method (called “finite element analysis”) that typically requires immense calculations, and without computers it would be impractical. With them, it has grown common.
Such simulations extend an ancient trend. We have always imagined consequences, in hope and fear, when we have needed to select a course of action. Simpler mental models (whether inborn or learned) undoubtedly guide animals as well. When based on accurate mental models, thought experiments can replace more costly (or even deadly) physical experiments – a development evolution has favored. Engineering simulations simply extend this ability to imagine consequences, to make our mistakes in thought rather than deed.
In “One Highly Evolved Toolbox,” J. Baldwin discusses how tools and thought mesh in job-shop work: “You begin to build your tool capability into the way you think about making things. As anyone who makes a lot of stuff will tell you, the tools soon become sort of an automatic part of the design process . . . But tools can’t become part of your design process if you don’t know what is available and what the various tools do.”
Having a feel for tool capabilities is essential when planning a jobshop project for delivery next Wednesday; it is equally essential when shaping a strategy for handling the breakthroughs of the coming decades. The better our feel for the future’s tools, the sounder will be our plans for surviving and prospering.
A craftsman in a job shop can keep tools in plain sight; working with them every day makes them familiar to his eyes, hands, and mind. He gets to know their abilities naturally, and can put this knowledge to immediate creative use. But people – like us – who have to understand the future face a greater challenge, because the future’s tools exist now only as ideas and as possibilities implicit in natural law. These tools neither hang on the wall nor impress themselves on the mind through sight and sound and touch – nor will they, until they exist as hardware. In the coming years of preparation only study, imagination, and thought can make their abilities real to the mind.
What Are the New Replicators?
History shows us that hardware evolves. Test tube RNA, viruses, and dogs all show how evolution proceeds by the modification and testing of replicators. But hardware (today) cannot reproduce itself – so where are the replicators behind the evolution of technology? What are the machine genes?
Of course, we need not actually identify replicators in order to recognize evolution. Darwin described evolution before Mendel discovered genes, and geneticists learned much about heredity before Watson and Crick discovered the structure of DNA. Darwin needed no knowledge of molecular genetics to see that organisms varied and that some left more descendants.
A replicator is a pattern that can get copies of itself made. It may need help; without protein machines to copy it, DNA could not replicate. But by this standard, some machines are replicators! Companies often make machines that fall into the hands of a competitor; the competitor then learns their secrets and builds copies. Just as genes “use” protein machines to replicate, so such machines “use” human minds and hands to replicate. With nanocomputers directing assemblers and disassemblers, the replication of hardware could even be automated.
The human mind, though, is a far subtler engine of imitation than any mere protein machine or assembler. Voice, writing, and drawing can transmit designs from mind to mind before they take form as hardware. The ideas behind methods of design are subtler yet: more abstract than hardware, they replicate and function exclusively in the world of minds and symbol systems.
Where genes have evolved over generations and eons, mental replicators now evolve over days and decades. Like genes, ideas split, combine, and take multiple forms (genes can be transcribed from DNA to RNA and back again; ideas can be translated from language to language). Science cannot yet describe the neural patterns that embody ideas in brains, but anyone can see that ideas mutate, replicate, and compete. Ideas evolve.
Richard Dawkins calls bits of replicating mental patterns “memes” (meme rhymes with cream). He says “examples of memes are tunes, ideas, catch-phrases, clothes fashions, ways of making pots or of building arches. Just as genes propagate themselves in the gene pool by leaping from body to body [generation to generation] via sperms or eggs, so memes propagate themselves in the meme pool by leaping from brain to brain via a process which, in the broad sense, can be called imitation.”
The Creatures of the Mind
Memes replicate because people both learn and teach. They vary because people create the new and misunderstand the old. They are selected (in part) because people don’t believe or repeat everything they hear. As test tube RNA molecules compete for scarce copying machines and subunits, so memes must compete for a scarce resource – human attention and effort. Since memes shape behavior, their success or failure is a deadly serious matter.
Since ancient times, mental models and patterns of behavior have passed from parent to child. Meme patterns that aid survival and reproduction have tended to spread. (Eat this root only after cooking; don’t eat those berries, their evil spirits will twist your guts.) Year by year, people varied their actions with varying results. Year by year, some died while others found new tricks of survival and passed them on. Genes built brains skilled at imitation because the patterns imitated were, on the whole, of value – their bearers, after all, had survived to spread them.
Memes themselves, though, face their own matters of “life” and “death”: as replicators, they evolve solely to survive and spread. Like viruses, they can replicate without aiding their host’s survival or well-being. Indeed, the meme for martyrdom-in-a-cause can spread itself through the very act of killing its host.
Genes, like memes, survive by many strategies. Some duck genes have spread themselves by encouraging ducks to pair off to care for their gene-bearing eggs and young. Some duck genes have spread themselves (when in male ducks) by encouraging rape, and some (when in female ducks) by encouraging the planting of eggs in other ducks’ nests. Still other genes found in ducks are virus genes, able to spread without making more ducks. Protecting eggs helps the duck species (and the individual duck genes) survive; rape helps one set of duck genes at the expense of others; infection helps viral genes at the expense of duck genes in general. As Richard Dawkins points out, genes “care” only about their own replication: they appear selfish.
But selfish motives can encourage cooperation. People seeking money and recognition for themselves cooperate to build corporations that serve other people’s wants. Selfish genes cooperate to build organisms that themselves often cooperate. Even so, to imagine that genes automatically serve some greater good (- of their chromosome? – their cell? – their body? – their species?) is to mistake a common effect for an underlying cause. To ignore the selfishness of replicators is to be lulled by a dangerous illusion.
Some genes in cells are out-and-out parasites. Like herpes genes inserted in human chromosomes, they exploit cells and harm their hosts. Yet if genes can be parasites, why not memes as well?
In The Extended Phenotype, Richard Dawkins describes a worm that parasitizes bees and completes its life cycle in water. It gets from bee to water by making the host bee dive to its death. Similarly, ant brainworms must enter a sheep to complete their life cycle. To accomplish this, they burrow into the host ant’s brain, somehow causing changes that make the ant “want” to climb to the top of a grass stem and wait, eventually to be eaten by a sheep.
As worms enter other organisms and use them to survive and replicate, so do memes. Indeed, the absence of memes exploiting people for their own selfish ends would be amazing, a sign of some powerful – indeed, nearly perfect – mental immune system. But parasitic memes clearly do exist. Just as viruses evolve to stimulate cells to make viruses, so rumors evolve to sound plausible and juicy, stimulating repetition. Ask not whether a rumor is true, ask instead how it spreads. Experience shows that ideas evolved to be successful replicators need have little to do with the truth.
At best, chain letters, spurious rumors, fashionable lunacies, and other mental parasites harm people by wasting their time. At worst, they implant deadly misconceptions. These meme systems exploit human ignorance and vulnerability. Spreading them is like having a cold and sneezing on a friend. Though some memes act much like viruses, infectiousness isn’t necessarily bad (think of an infectious grin, or infectious good nature). If a package of ideas has merit, then its infectiousness simply increases its merit – and indeed, the best ethical teachings also teach us to teach ethics. Good publications may entertain, enrich understanding, aid judgment – and advertise gift subscriptions. Spreading useful meme systems is like offering useful seeds to a friend with a garden.
Parasites have forced organisms to evolve immune systems, such as the enzymes that bacteria use to cut up invading viruses, or the roving white blood cells our bodies use to destroy bacteria. Parasitic memes have forced minds down a similar path, evolving meme systems that serve as mental immune systems.
The oldest and simplest mental immune system simply commands “believe the old, reject the new.” Something like this system generally kept tribes from abandoning old, tested ways in favor of wild new notions – such as the notion that obeying alleged ghostly orders to destroy all the tribe’s cattle and grain would somehow bring forth a miraculous abundance of food and armies of ancestors to drive out foreigners. (This meme package infected the Xhosa people of southern Africa in 1856; by the next year 68,000 had died, chiefly of starvation.)
Your body’s immune system follows a similar rule: it generally accepts all the cell types present in early life and reject such as potential cancer cells and invading bacteria, as foreign and dangerous. This simple reject-the-new system once worked well, yet in this era of organ transplantation it can kill. Similarly, in an era when science and technology regularly present facts that are both new and trustworthy, a rigid mental immune system becomes a dangerous handicap.
For all its shortcomings, though, the reject-the-new principle is simple and offers real advantages. Tradition holds much that is tried and true (or if not true, then at least workable). Change is risky: just as most mutations are bad, so most new ideas are wrong. Even reason can be dangerous: if a tradition links sound practices to a fear of ghosts, then overconfident rational thought may throw out the good with the bogus. Unfortunately, traditions evolved to be good may have less appeal than ideas evolved to sound good – when first questioned, the soundest tradition may be displaced by worse ideas that better appeal to the rational mind.
Yet memes that seal the mind against new ideas protect themselves in a suspiciously self-serving way. While protecting valuable traditions from clumsy editing, they may also shield parasitic claptrap from the test of truth. In times of swift change they can make minds dangerously rigid.
Much of the history of philosophy and science may be seen as a search for better mental immune systems, for better ways to reject the false, the worthless, and the damaging. The best systems respect tradition, yet encourage experiment. They suggest standards for judging memes, helping the mind distinguish between parasites and tools.
The principles of evolution provide a way to view change, whether in molecules, organisms, technologies, minds, or cultures. The same basic questions keep arising: What are the replicators? How do they vary? What determines their success? How do they defend against invaders? These questions will arise again when we consider the consequences of the assembler revolution, and yet again when we consider how society might deal with those consequences.