For its services Buchnera also gets something in return. By living inside an aphid’s cell, Buchnera floats in a broth of molecules that is rich enough to provide almost all the food the bacterium needs. And not just food: The aphid’s cell walls provide Buchnera with a safe and comfortable home. By dragging Buchnera with them everywhere, aphids insulate them from heat, cold, rain, and other environmental hazards.40 Buchnera needs to take no notice of overgrazed plants, lurking predators, or any other threat. So long as its aphid home survives, the bacterium thrives. Like a vacationer idling its time away in the gently lapping waves of a tepid ocean, Buchnera is sheltered from a hostile world.41
Buchnera’s vacation has been going on for a very long time. The host and the bacterium first cohabited more than a hundred million years ago, and have lived together ever since. Over that much time, one might expect significant changes to any organism. That’s what happened to Buchnera, and those changes reveal much about the relationship between robustness and complexity.
To understand that relationship, it is useful to compare Buchnera to its cousin, Escherichia coli, that marvel of metabolic flexibility, able to survive on dozens of different food sources, and highly robust to changing chemical environments. E. coli’s complex metabolic network harbors more than a thousand chemical reactions, a large collection of skills for surviving in a changeable and uncertain world.
The metabolism of Buchnera’s ancestors once resembled that of E. coli. But no longer. Now its metabolic network has a puny 263 metabolic reactions. Its alliance with aphids started when dinosaurs still walked the earth, and since then it has lost nearly three-quarters of the reactions that E. coli still harbors. A steady stream of DNA copying errors eroded the genes that are needed for these reactions, and many of them disappeared through gene deletions that occur naturally in DNA. Buchnera has survived hundreds of such deletions.42
It takes no genius to see why Buchnera could survive all these deletions and become so much simpler than E. coli. Hundreds of genes and metabolic reactions have become superfluous in Buchnera, because its world has stood still for more than a hundred million years. While its host aphid struggles in an ever-changing environment, Buchnera bathes in a steady if monotonous diet of nutrients. To survive in a simple, unchanging world, a simple metabolism does the trick, and complexity becomes not only superfluous but wasteful.
Buchnera is special but not unique. Many microbes live on or inside other, larger organisms. Some serve their hosts, others exploit them. A well-known human example is the bacterium Mycoplasma pneumoniae, a cause of the “walking pneumonia” that does not tie patients to their beds. This parasite shuttles from body to human body, and relies on human cells to supply its food. Its metabolism is even simpler than that of Buchnera: It needs a mere 189 reactions to survive on the rich molecular buffet that human cells provide. Incredibly, it has even shed part of metabolism’s universal core, the most ancient citric acid cycle. What’s more, its extreme minimalism helps it resist antibiotics that attack the enzymes building the bacterial cell wall: It no longer makes this wall, and even hijacks membrane molecules from our own body to prevent its molecular innards from spilling.43
A corollary of this increase in simplicity is a corresponding decrease in robustness: not just robustness to mutations, but robustness to environmental change—the two are linked. A metabolism that is robust to knocking out enzyme-coding genes will also be robust to changing environments. If E. coli were to live in a single, fixed environment—one, for example, where glucose is the only source of carbon—it could do without 70 percent of the chemical reactions in its complex metabolism.44 But in Buchnera this is no longer the case. Fully 90 percent of its 263 reactions are essential.45 Eliminate one of them and you kill the organism.
Put another way, the metabolic road network of E. coli has many alternate routes, but Buchnera’s doesn’t. It is more like a single-lane road without exits. Block it anywhere and all traffic piles up behind the roadblock—the place where an essential molecule can no longer be made. E. coli is robust, both to the DNA mutations that eliminate metabolic reactions and to changing environments. Buchnera is not robust to either.
E. coli and B. aphidocola are only two specks of dust in a vast library of metabolisms, and what holds for them—an organism viable in a changing environment is more complex and robust—might not be universal. We cannot examine all metabolisms, for there are too many. But we can examine many of them computationally, and thus do what pollsters do for human populations: learn about a very large population from random samples that reflect the population’s properties. By exposing such a random sample of metabolisms to changing environments, we can learn whether E. coli and Buchnera are unusual or typical.
To this end, researchers in my laboratory create hundreds of metabolic networks that contain as few reactions as possible and yet can still sustain life. We call such networks minimal metabolisms. Make them any smaller and you destroy their ability to sustain life. We can create minimal metabolisms that can sustain life in one environment, but also in two, three, and more environments, up to dozens of them, each one differing only in available nutrients.
One lesson from such minimal metabolisms is that living in many environments generally requires complexity. In one study we analyzed environments that differ in their sources of sulfur, a chemical element as essential as carbon. We first identified minimal metabolisms—there are more than one—that can sustain life on only one sulfur source, and found that such metabolisms need fewer than twenty chemical reactions. But to sustain life on five different sulfur sources, a metabolism already needs at least twenty-five reactions. And to be viable in forty different environments, it needs more than sixty chemical reactions. In other words, a metabolism that can sustain life in more and more environments needs more and more reactions. It needs to be more complex.46
The same metabolism also becomes more robust: We can remove more and more reactions from it, while it remains viable in any one environment.47 The more reactions a metabolism contains, the more reactions it can do without in any one environment. These reactions are neutral in one environment, but they can become essential in a different environment. Thus E. coli and Buchnera are not unusual, but special cases of a general principle: Life’s complexity and robustness increase with its exposure to environmental change.48
With this recognition a circle is closing. Environmental change requires complexity, which begets robustness, which begets genotype networks, which enable innovations, the very kind that allow life to cope with environmental change, increase its complexity, and so on, in an ascending spiral of ever-increasing innovability. At the core of this innovability is the self-organized multidimensional fabric of genotype networks, hidden behind life’s visible splendor, but creating this splendor. It is the hidden architecture of life.
CHAPTER SEVEN
From Nature to Technology
YaMoR—an acronym for Yet another Modular Robot—is an experimental robot constructed with hinged components, a bit like the segments of a millipede, that can be arranged in a straight line to undulate like a worm, or in leglike pairs to crawl like an amphibian, or even to walk like an insect. And these segments are smart, equipped with reconfigurable hardware, each segment containing a computer chip that can be rewired a bit like a human brain.1 Built the right way, such robots could learn on the job by rewiring themselves when they need new skills.
Constructed at one of the world’s leading engineering schools, the École Polytechnique Fédérale de Lausanne in Switzerland, YaMoR is a member of a growing family of modular robots created in engineering labs around the world. The family is large but its members lack much resemblance, because their body plans are very different. Some modular robots are dielike cubes, some chains of tetrahedrons. Others are clusters of balls, and still others strings of rotating wheels. Modular robots like YaMoR seem limited only by the constraints of solid geometry.
More than 540 million years before YaMoR took its first steps,
an even more diverse family of body plans arose, in a burst of biological innovation known as the Cambrian explosion. It gave rise to every animal body plan now in use, along with dozens more that are extinct, like the entire phylum of limbless segmented marine animals known as the Vetulicolia. YaMoR’s phylum is much more primitive than the Vetulicolia, but if the warp drives that accelerate nature’s innovation could be put to work in human technology, robotic or otherwise, then the first Cambrian explosion may not have been the last.2
This idea—that analogs of genotype networks could accelerate innovation in technology—is not so far-fetched, as we shall see. A first hint is that innovation in nature and innovation in technology show many parallels.
Trial and error, for one thing. Thomas Edison, an archetype of inventive genius, “tested no fewer than 6,000 vegetable growths, and ransacked the world for the most suitable filament material” until he eventually stumbled upon bamboo as the best solution for the fickle filaments of his incandescent light bulb.3 One of the dozens of quotations attributed to him sums it up: “I have not failed. I have just found ten thousand ways that do not work.” The quote reminds everyone that trial and error—especially error—is as crucial to technological innovations as to biological ones. And it is no less crucial today than in Edison’s time. John Backus, a co-creator of the highly successful computer programming language Fortran, which helps scientists simulate and understand the universe, from the motions of atoms to those of galaxies, said that “you need the willingness to fail all the time. You have to generate many ideas and then you have to work very hard only to discover that they don’t work. And you keep doing that over and over until you find one that does work.”4
To be sure, failure has different consequences in evolution than it does for an inventor with a failed light bulb or for a scientist with a disproved theory. A bar-headed goose with a mutant hemoglobin produced by nature’s tinkering with DNA is a living experiment.5 If the mutation improves its ability to scavenge oxygen from thin air, great. But woe unto the bird whose mutant globin can no longer bind oxygen. Its lights go out permanently.
Failure in science and technology does not usually mean bodily death, but this does not mean that ideas die easily. Sir Fred Hoyle, one of the world’s most prominent astronomers and astrophysicists, went to his death in 2001 not only denying the Big Bang theory but defending the belief that flu epidemics arise when a lull in the solar wind allows extraterrestrial flu viruses to enter the atmosphere. The nineteenth century’s Lord Kelvin used the laws of thermodynamics—and his Christian faith—to underestimate the age of the earth more than a hundredfold.6 The historical battlefields of science and technology are littered with brilliant minds who took wrong beliefs to their graves. Max Planck, a father of quantum theory, observed that “a new scientific truth does not triumph by convincing its opponents and making them see the light, but rather because its opponents eventually die, and a new generation grows up that is familiar with it.” Science, like nature, advances one funeral at a time.7
One of nature’s antidotes against catastrophic failure has been accidentally co-opted by technological innovators: populations. Great inventions aren’t the work of solitary geniuses any more often than nature’s libraries are explored by single organisms. Despite the cliché of human innovators conjuring unimagined worlds from the depths of their minds—from Archimedes in his bathtub to Einstein in his patent office—the truth is that technological innovation depends on the same kind of crowdsourcing that biological innovation uses when its chemical libraries are explored by armies of browsers. A team of people developed Fortran, and Edison used dozens of assistants to create and test new designs for light bulbs, telephones, and telegraphs. The nineteenth-century Industrial Revolution was made possible by the rise of an entire new class—highly educated artisans far more numerous than aristocratic gentlemen scientists—who needed to make money from their work and created a flurry of inventions from steam engines to automatic looms.8 And today, any new technology, from new cell phones to new drugs and new energy carriers, requires armies of scientists and engineers, intense competition, and myriad failures before finding success. Given the importance of trial and error, it’s hard to see how it could be otherwise: The more explorers, the more solutions can be explored, with correspondingly greater odds of success.
And when the armies of technological innovators advance, they do so on many fronts simultaneously—again like nature. The American sociologist Robert Merton, who also coined the terms “role model” and “self-fulfilling prophecy,” is well remembered in the history of sciences for documenting the prevalence of inventions with multiple origins—he simply called them “multiples.”9 The list is nearly endless: The relation between the heat and pressure of a gas is known as Boyle’s law in English-speaking countries, and as Mariotte’s law to Francophones, since the same phenomenon was independently derived by Robert Boyle and Edme Mariotte. Robert Fulton, the Marquis de Jouffroy, and James Rumsey are all “inventors” of the first steamboats. The thermometer has at least six different inventors, and calculus, as is well known, was created almost simultaneously by both Isaac Newton and Gottfried Wilhelm Leibniz.10 Elisha Gray filed for a patent on a working telephone the same day as Alexander Graham Bell (though Bell won the subsequent legal wrangle over priority).11
Multiple origins are possible because the problems of technology—like those of biology—have multiple solutions. Among the best-documented biological examples are the innovations that remove carbon dioxide from the atmosphere, a process called carbon fixation by biologists and carbon scrubbing by engineers. The premier biological carbon scrubber is the one used by plants, an enzyme that attaches carbon dioxide to a sugar called ribulose-1,5-bisphosphate. It then modifies this carrier further, such that the carbon dioxide eventually becomes part of the plant’s body. This is not only how plants grow and how carbon gets into the fossil fuels we burn. It is also the way carbon dioxide is fed back into the cycle of life. But plants are not the only organisms that fix carbon: Some microbes attach it to the carrier molecule acetyl-CoA, and others add it to molecules from the ancient citric acid cycle.12 Environmental engineers—they seek to solve the same problem to forestall catastrophic climate change—have already come up with several further carbon-scrubbing technologies that use molecules like monoethanolamine and sodium hydroxide.
Other examples of Merton’s multiples are legion. An automobile’s engine can use reciprocating pistons or an eccentric rotor, and it can trigger fuel combustion with the spark plug of a gasoline engine or through the heat of compression in a diesel engine. Organisms can detect light waves using a flexible single lens, or the rigid compound eye of a fly.13 Antifreeze proteins of Arctic and Antarctic fish, transparent crystallins that originate from different enzymes, and highly diverse oxygen-binding globins are all multiple solutions to similar problems.
Another commonality of both technological and biological innovations is that they endow old things with new life. The history of technological innovation is practically saturated in examples. In the words of the writer Stephen Johnson, Johannes Gutenberg borrowed “a machine designed to get people drunk”—a screw-driven wine press—“and turned it into an engine of mass communication.”14 Microwave ovens heat food with a technology originally developed for radar—a radar engineer discovered its heating powers when it melted a chocolate bar in his pocket. (The first commercial version was called the “Radarange.”) The lightweight synthetic Kevlar, originally developed to replace steel in racing tires, has been co-opted for bulletproof vests and steel helmets. The same principle is at work even in mundane contraptions that barely deserve the label “innovation.” A door placed on two sawhorses can make a great desk. Boots can serve as low-tech doorstops. Milk crates can make wonderful filing cabinets, and so on.15 Edison said it well: “To invent, you need a good imagination and a pile of junk.”16
In 1982, the paleontologists Stephen Jay Gould and Elizabeth Vrba baptized the biological versi
on of this phenomenon “exaptation.”17 (Once again, Darwin had gotten there first—he reminded the Origin’s reader that “an organ originally constructed for one purpose . . . may be converted into one for a widely different purpose.”)18 A classic example of exaptation is the bird’s feather, a complicated assemblage of fibrous proteins called keratins, the same proteins that make up the scales of reptiles. The first feathers were most likely involved in insulation or waterproofing, and were only later co-opted—“exapted”—for flying.19 Such exaptations also abound in molecules, including those regulators that help make feathers. One of them is a protein called “Sonic hedgehog”—yes, it is named for the character in the eponymous video game—that helps control the growth of both fingers and spinal cord in your body, but was co-opted in birds to sculpt feathers.20 Likewise, a regulator protein that helps shape legs in some organisms was co-opted to paint eyespots in others, and metabolic enzymes were co-opted as the crystallins of lenses.
Such co-option is a special case of a final parallel between nature and technology: Innovation is combinatorial. It combines old things to make the new. We encountered combinatorial innovations first in metabolism, where new combinations of old reactions turned toxins like pentachlorophenol into fuels to manufacture biomass, and allowed our ancestors to detox their bodies by synthesizing urea. In proteins they are new combinations of old amino acids, and in regulation circuits they are new combinations of interacting regulators. But a technological innovation like the aviation-transforming jet engine is just as combinatorial.21 Its three principal parts are a compressor that increases the air’s pressure, a combustion chamber that mixes air with fuel and ignites the mixture, and a rotating turbine that generates thrust from the expanding mixture. None of these three elements were new when half a dozen British, German, Hungarian, and Italian engineers—remember those multiples—were building the first jet engines in the twentieth century. The earliest compressors were bellows used to run forges more than two thousand years ago, and have been used in industrial blowers ever since. Combustion chambers had been central to both steam locomotives and internal combustion engines. Archimedes had invented a screw turbine in the third century BC, and the first gas turbine was patented in England in the late eighteenth century.
Arrival of the Fittest: Solving Evolution's Greatest Puzzle Page 20