I thought this was a cool article!
Robotocists are fascinated by their “swarm intelligence.” Biologists are amazed at their ability to turn grass into energy. Can humans replicate their achievements?
New termite colonies are founded on windless evenings, at dusk, after the rain. Most termites have neither eyes nor wings, but every mature colony has a caste of translucent-winged seeing creatures called alates, which are nurtured by the colony’s workers until they are ready to propagate. When the time comes—given the right temperature and humidity—colonies release thousands of alates into the air, an event called “swarming.” Most of the nutrient-rich alates are eaten by animals as they glide to the ground. The few that survive shed their wings and pair off, male and female. Then they burrow into the earth, future kings and queens. The pair will remain there, alone in a dark hole, for the rest of their lives. They bite off the ends of their antennae, reducing their acute sensitivity; perhaps it’s a means of making more bearable a life wholly given over to procreation. They mate, and the queen begins to lay her eggs. She will lay millions in the course of her decades-long life—the longest life span of any insect. Her translucent white abdomen, constricted by the tight black bands of her exoskeleton, swells to the size of a human thumb, leaving her unable to move. Her tiny head and legs flail while her pulsating body is fed and cleaned by her offspring. The South African naturalist and poet Eugène Marais described the queen’s fate in “The Soul of the White Ant” (first published, in Afrikaans, in 1934): “Although you will apparently be an immobile shapeless mass buried in a living grave, you will actually be a sensitive mainspring. You will become the feeling, the thinking, the seeing, of a life a thousand times greater and more important than yours could ever have become.”
Humans have often looked at insects and seen themselves, or the selves they would like to be. Early-modern European naturalists peered into termite mounds, anthills, and beehives and saw microcosms of well-ordered states: monarchs, soldiers, laborers. (There was no general recognition that bee “kings” were actually female “queens” until the sixteen-seventies, when a Dutch microscopist, Jan Swammerdam, pointed out that bee kings had ovaries.) In 1781, Henry Smeathman wrote a report for the Royal Society celebrating termites as “foremost on the list of the wonders of the creation” for “most closely imitating mankind in provident industry and regular government.” Termites, he wrote, surpassed “all other animals” in the “arts of building” by the same margin that “Europeans excel the least cultivated savages.” According to Smeathman, the “perfect” alate caste “might very appositely be called the nobility or gentry, for they neither labour, or toil, or fight, being quite incapable of either,” but are instead devoted to founding new colonies. (In 1786, Smeathman published a plan for the settling of freed black slaves in a new colony, on the West African coast, where he had done his termite studies.) He viewed the laborers, meanwhile, as “voluntary subjects” who served the “happy pair” of king and queen. Just over a century later, in “Mutual Aid” (1902), the Russian thinker and revolutionary Peter Kropotkin exalted the coöperative habits of termites as a model, and a scientific basis, for Communism. In “Civilization and Its Discontents,” Freud presented the termite mound as an example of the perfect sublimation of the individual will to the demands of the group—a sublimation that, he said, would continue to elude mankind.
Some have seen in termites a darker vision for humanity, a warning rather than a guide. The early-twentieth-century American entomologist William Wheeler began as a believer in the political example of termites and ants, detecting in their colonies a Deweyan ethos, both communitarian and democratic. But, by the late nineteen-twenties, Wheeler had begun to worry that the social insects represented a sort of evolutionary cul-de-sac, which foretold “the eventual state of human society”: “very low intelligence combined with an intense and pugnacious solidarity of the whole.” For Wheeler, the harmony of insect society was made possible by its solution to what he called the “problem of the male.” Males, Wheeler said, are the “antisocial sex,” responsible for the “instability and mutual aggressiveness so conspicuous among the members of our own society.” Termites and ants, with their castes of sterile male workers and soldiers, had done away with the problem of the male. But humans could do so only at the cost of civilization, Wheeler warned, for “all progress . . . is initiated by a relatively small portion of the male population, whose restlessly questing intellects are really driven by the unsocial dominance impulses of their male mammalian constitution and not by any intense desire to improve society.” Among those products of male striving Wheeler counted “sciences, arts, technologies,” along with “philosophies, theologies, social utopias.” He did not appear to worry about what the termite life might mean for women, or about the possibility that the queen was not a queen at all but a slave.
Termites are insects of the infraorder Isoptera. They have bulbous, eyeless heads and teardrop-shaped bodies that are often translucent, exposing a swirl of guts and digesting plant matter. They are eusocial creatures—eusociality being the highest level of animal sociality recognized by sociobiologists, characterized by a division of reproductive labor between fertile and non-fertile castes, and by the collective care of the young. Until 2007, Isoptera was considered its own distinct order, and it had been classified that way for the previous hundred and seventy-five years. But phylogenetic studies confirmed that, despite appearances, termites are a kind of cockroach, and so Isoptera was reclassified under the cockroach order, Blattodea. This demotion has not helped the termite cause. Termites already suffer in the comparison with other eusocial insects: they lack the charisma of bees, with their summery associations and waggle dances, and do not receive the same recognition as ants for their work ethic and load-bearing capacities. They also have a reputation for destruction. In the United States, termites have been estimated to consume somewhere between $1.5 and $20 billion worth of property every year. At times they go straight for the cash: in 2011, termites consumed around ten million rupees’ worth of banknotes in a branch of the State Bank of India in Uttar Pradesh; two years later, termites munched part of the way through the savings of an elderly woman in Guangdong, who had wrapped four hundred thousand yuan in plastic and put it in a drawer. The Australian Mastotermes darwiniensis, the oldest and one of the largest species of termite—most closely resembling the wood-eating cockroach from which termites are thought to have evolved—is reported to have performed legendary feats of chewing, including reducing a house to rubble while its owner was travelling for two weeks.
In fact, only twenty-eight of approximately twenty-six hundred identified species of termite are invasive pests. (If they all were, we would be in big trouble: collectively, termites outweigh humans ten to one.) What’s more, noninvasive termites are ecologically crucial, in irrigating land, protecting against drought, and enriching the soil. They may also have served as a crucial food source for our own australopithecine ancestors. And yet termites are generally unloved. While I was reading Lisa Margonelli’s new book, “Underbug: An Obsessive Tale of Termites and Technology,” I discovered that everyone I knew had an unsavory termite tale. A friend who lives in Los Angeles is disgusted by the piles of black beads she finds near neat holes in her hardwood floors, which I unhelpfully identified as the fecal pellets, or “frass,” of dry-wood termites. Another friend, in Berkeley, swears that she can hear termites chewing when she closes her eyes at night, despite an exterminator’s assurances that her house is not infested. As a small child in suburban New Jersey, I discovered a piece of wood in our back yard that was covered in a maze of delicate etching. I was thrilled with the beauty of it: the smooth, shallow holes and grooves had the look of secret runes—the writing, I imagined, of Druids or fairies. I took it in to my mother. She told me that this was not magic but the sign of a termite infestation, and made me throw it out.
Termites may be hard to love, but they should be easy to admire. Termite mounds are among the largest structures built by any nonhuman animal. They reach as high as thirty feet, which, proportional to the insects’ tiny size, is the equivalent of our building something twice as tall as the 2,722-foot Burj Khalifa, in Dubai. The mounds are also fantastically beautiful, Gaudíesque structures, with rippling, soaring towers, in browns and oranges and reds. The interior of a termite mound is an intricate structure of interweaving tunnels and passageways, radiating chambers, galleries, archways, and spiral staircases. To build a mound, termites move vast quantities of mud and water; in the course of a year, eleven pounds of termites can move about three hundred and sixty-four pounds of dirt (in the form of mud balls) and thirty-three hundred pounds of water (which they suck into their bodies). The point of all this construction is not to have a place to dwell—the colony lives in a nest a metre or two below the mound—but to be able to breathe. A termite colony, which may contain a million bugs, has about the same metabolic rate as a nine-hundred-pound cow, and, like cows (and humans), termites breathe in oxygen and expel carbon dioxide. The mound acts as a lung for the colony, managing the exchange of gases, leveraging small changes in wind speed to inhale and exhale. Also like lungs, a termite mound has a role as a secondary diffusion system, which carries oxygen to and carbon dioxide away from the far reaches of the underground termite nest. The mound functions as a humidifier, too, tightly regulating moisture levels across wet and dry seasons. Some termite species partly outsource their digestion through the practice of fungiculture—the farming of a grass-eating fungus, which they store, tend, and feed in an elaborate garden maze below the mound.
Termites appear to do all this without any centralized planning: there are no architects, engineers, or blueprints. Indeed, the termite mound is not so much a building as a body, a self-regulating organic process that continuously reacts to its changing environment, building and unbuilding itself. Its complex behavior emerges, as if by magic, from its simple constituents. It is generally agreed that individual termites are not particularly intelligent, lacking memory and the ability to learn. Put a few termites into a petri dish and they wander around aimlessly; put in forty and they start stampeding around the dish’s perimeter like a herd. But put enough termites together, in the right conditions, and they will build you a cathedral.
Underbug” is more about humans who are preoccupied with termites than about termites themselves. Specifically, Margonelli is concerned with the sort of human whose interest in termites isn’t confined to wanting to kill them. (About half the scientific papers written about termites from 2000 to 2013 involve their extermination). These entomologists, geneticists, synthetic biologists, mathematical biologists, microbial ecologists, roboticists, computer scientists, and physicists are drawn to termites for a variety of reasons, not all of which are compatible. Some of these scientists, the minority, simply appear to be seduced by termites, and want to understand how they do what they do. One such is J. Scott Turner, a physiologist who, before turning to termites, placed alligators in wind tunnels in order to understand how they regulate their body temperature. By pumping propane gas down termite mounds, he was able to show that they function as lungs, not as chimneys that allow hot air to escape, which had been the previous assumption. (Putting things into a mound and seeing what happens is a favored mode of termite experimentation; Turner and his team have also experimented with plastic beads and molten aluminum. One convenience of working with termites is that there are few regulations concerning their treatment.)
Turner is a proponent of what he calls the “extended organism” thesis. (It’s meant as a variant of, and ultimately as an alternative to, Richard Dawkins’s “extended phenotype” model.) In Turner’s view, the physical termite mound—with its mud tunnels and walls, digested wood and grass and fungus—is part of the termite, rather than part of the environment on which the termite acts. The entire mound—insects plus structure—is thus a living thing: a self-regulating physiological and cognitive system, with a sense of its own boundaries, a memory, and a kind of collective intentionality.
The extended-organism hypothesis also recalls an older idea: that the termite, bee, or ant colony is a “superorganism.” This term was coined by William Wheeler in 1911, though the idea dates back to Darwin, who saw the superorganism as a solution to the “problem” of eusociality. The problem is this: if natural selection favors those organisms which are best at reproducing, then how do castes of nonreproductive insects ever evolve? One way to address the problem is to regard the colony as a whole as the unit of selection. The sterile worker should be thought of not as an individual organism but as a “well-flavored vegetable,” in Darwin’s phrase, produced by the queen.
Today, most evolutionary theorists favor the “inclusive fitness” explanation of eusociality, a theory developed by W. D. Hamilton in the early nineteen-sixties. Hamilton showed mathematically that altruism can be a beneficial reproductive strategy for an organism, so long as the altruistic act benefits another organism to which it is sufficiently genetically similar. As a human being, the obvious way for me to reproduce my genes is to have biological children, who will inherit half of my genes. But I can also reproduce my genes by helping my sister, who shares on average half of my genetic material, nurture and protect her own children, who will share a quarter. If sacrificing my life will enable my sister to have more than twice as many children as I would have had, my sacrifice is “worth it,” from the perspective of my selfish genes. E. O. Wilson, though an early evangelist for Hamilton’s theory, has recently argued for a return to the superorganism as a solution to Darwin’s problem. In this, Wilson is very much in the minority; Richard Dawkins has called his criticisms of inclusive fitness “downright perverse.”
Most of the other scientists Margonelli follows are interested in termites as a means to human ends, and aim at simplifying their complexity to something replicable. Consider termites’ ability to convert dead plant matter into energy. They do this with the help of the hundreds, sometimes thousands, of species of microbes—bacteria and protists—that live in their guts, ninety per cent of which are found nowhere else on earth. Some of these microbes are themselves, like the termite superorganism, composite animals. The protist Trichonympha, found in some termite guts, is itself host to colonies of symbiotic bacteria. Termites and their gut microbes are thought to have coevolved between two hundred and fifty million and a hundred and fifty-five million years ago, when some cockroaches ingested wood-eating microbes, and then began sharing what entomologists politely call “woodshake”—a mixture of feces, microbes, and plant matter—among themselves, mouth to mouth, and mouth to anus. This practice, known as “trophallaxis” (another of William Wheeler’s coinages), allows a communal pooling of digestive capacity, which can then be handed down from one generation to the next. (With the rise of fecal transplants to cure C. difficile infection and other gastrointestinal disorders, trophallaxis is gaining popularity among humans; the F.D.A. has, since 2013, officially classified human feces as a drug.) The Department of Energy says that the U.S. can produce 1.3 billion tons of dry biomass—from harvested trees, cornstalks, high-energy grasses, and the like—without taking anything away from regular agricultural uses. If humans can crack the code to termite digestion, the U.S. could turn the stuff into nearly a hundred billion gallons of biofuel a year—what’s sometimes called “grassoline”—and thereby reduce automobile emissions by eighty-six per cent.
The search for a termite-inspired grassoline is a major goal of the emerging field of synthetic biology, in which biological systems—metabolic pathways, cells, organisms—are reëngineered to produce things humans want, including biofuels and precursors of drugs. One of the field’s leaders is Jay Keasling, who runs the Department of Energy’s Joint BioEnergy Institute, or J.B.E.I. Keasling imagines a fully modular system of synthetic biology, with different companies producing different off-the-shelf parts—empty cell “bags,” the chromosomes with which to program them, the molecules to “boot” them up—that can readily be assembled to produce the desired chemical output. Manufacturing a termite biofuel would require identifying the genes for wood eating from the termite’s microbe colony and inserting them into a cellular bag. The first challenge is overcoming the fickleness of microbes: less than one per cent of them can be isolated and grown in a petri dish. This used to mean that it was nearly impossible to map the genomes of the termite’s wood-eating microbes. But in 2004 a team led by the Berkeley earth scientist Jill Banfield came up with “metagenomics,” a process of sequencing the genes of an entire microbial community at once. In 2007, Nature published a metagenomic analysis of gut microbes from a Costa Rican termite; puzzle-piecing together fifty-four million base pairs of DNA, researchers identified more than a thousand genes that might be for digesting wood. A termite biofuel seemed not far off.
Yet the synthetic biologists at J.B.E.I. still have not produced a grassoline that can compete with ordinary fossil fuels. (They have turned their attention to the production of other biofuels, including those in demand by the military.) Margonelli suggests two reasons for this failure. First, the termite’s gut turned out to be too complex to understand, let alone imitate. Phil Hugenholtz, one of the researchers who helped sequence the gut microbes of the Costa Rican termite, jokes that “you might as well go and hook your car to a bunch of termites.” Second, the biology itself seems to resist being reëngineered in the way that synthetic biologists would like. “What we’re doing,” Héctor García Martín, a physicist who works with Keasling, says, “is taking a bug”—like E. coli—“with no interest in producing biofuels and forcing it to produce them.” García Martín goes on to cite the microbiologist Carl Woese, who observed that, unlike electrons, cells have a history—something like memories of what they have metabolized in the past. These “memories” are encoded not in the cells’ DNA but somewhere else in their chemistry, so it may be misguided to think in terms of swapping genetic programs in and out of cell “bags.” The willingness, on the part of a physicist like García Martín, to talk about the “memories” and “interests” of biological systems is surprising. But it reflects a larger shift among synthetic biologists away from a belief in the fundamentally mechanical nature of life.
In 2014, Keasling and three other prominent synthetic biologists published a paper in the journal Cell, in which they declared it an “open question . . . whether biology is genuinely modular in an engineering sense”—that is, a predictable aggregation of rudimentary components—“or whether modularity is only a human construct that helps us understand biology.” But the spectre raised by termites, microbes, and other organisms that are at once simple and devilishly complex is that the very metaphor of modularity might be misleading: that, as long as we think of living systems as machines, we are guaranteed not to understand them.
Another reason termites interest engineers is that they are a paradigm of “swarm intelligence”—highly complex behavior that emerges from the interaction of individual units in the absence of a centralized command. Each termite is presumed to be governed by a set of simple rules, which dictate particular actions—crawl, turn, dig, stack a mud ball—in response to specific triggers from the environment or from other termites. But it’s unclear precisely what mechanism produces termites’ group intelligence—which chemical or physical signals trigger which actions, and by what rules.
Since 1959, the dominant theory has been “stigmergy,” first developed by the French biologist Pierre-Paul Grassé. The term comes from the Greek stigma (mark or sign) and ergon (work or action); the idea is that a trace left behind in the environment by one agent triggers further action by other agents, creating a positive-feedback loop. Stigmergy seeks to explain how extremely simple creatures, with no capacity for communication, can achieve the appearance of joint decision-making. In the case of termites (stigmergy has also been used to explain the complex emergent behavior of other simple creatures, such as multicellular bacteria) scientists speculate that the action-triggering “trace” is found in their saliva. A termite picks up a mud ball, gets some of its saliva on it, and drops it, presumably at random; other termites, triggered by the saliva scent, start stacking mud-and-saliva balls on top of the first ball, strengthening the signal; eventually, the mud balls turn into a wall or a pillar.
In the nineteen-nineties, computer scientists began programming virtual termites that built “walls” via the principles of stigmergy. These virtual termites could build two-dimensional shapes, but they could not produce anything like the complex three-dimensional architecture of real termites. And though stigmergy might explain how termites build, it does not readily explain why they so often unbuild, dismantling and modifying their work as they go. Recent studies suggest that some individual termites have a tendency to lead, while others have a tendency to follow—meaning that what gets the stigmergic process going is not a random action but something more systematic. It also appears that termites are not so much industrious drones as they are denizens of a post-capitalist Utopia: in a petri dish of twenty-five termites, only five appear to work at any one time. It seems likely that stigmergy is, at best, just one of several mechanisms that produce the complex group behavior of termites. For many researchers, identifying these mechanisms is the key to the future of robotics and A.I.: not one smart machine but a hyper-smart flock of thousands of small, cheap, dumb machines.
In 2014, an issue of Science featured, on its cover, a piece on termes, a termite-inspired robot created by the computer scientist Radhika Nagpal and her team at Harvard’s Wyss Institute for Biologically Inspired Engineering. termes are adorable, semicircular, tissue-box-size robots that move on four “whegs” (short for “wheel legs,” a feature inspired by a cockroach’s climbing behavior) and lift and move blocks with their clawlike heads. Each termes robot is programmed with an algorithm that tells it what basic action (move forward, turn, pick up block, place block) to perform next, based on the input its sensors receive about its environment. By following a sequence of a hundred or so programmed steps, each robot can construct a preordained structure: a wall, a staircase, or a four-sided building. What is more, a group of termes, each programmed with the same set of individual instructions, will collectively build the same structure, without any centralized command or inter-robot communication; if one robot detects another in its way, it simply pauses until it stops sensing the other robot, and then gets back to its regular programming. The robots are built on the principle of what Nagpal and her team call “extended stigmergy”: the embedding of design information in the robots’ environment, rather than in the robots. Each building block, for example, can be given a unique label, allowing the robots to use the blocks as landmarks. In some versions of the termes system, the robots themselves tag the blocks as they build.
When the Science piece came out, there was a brief media frenzy, with some journalists predicting that termes would end up colonizing Mars, and others warning of the coming robot apocalypse. Still, termes are limited: they can build only on a black-and-white floor, in quiet rooms, and with magnetized blocks. Indeed, these are features of extended stigmergy: termes rely heavily on the orderliness of their environment to be able to build. Real termites, by contrast, are masters at responding to the novel and the unpredictable. “I don’t really know how to do that,” Nagpal says. What is not clear is whether termes ever will be termites—whether a more sophisticated version of stigmergy will eventually allow robots to mimic their biological models, or whether stigmergy, like modularity, is a framework that can take engineers only so far.
The Wyss Institute’s most famous robot is the RoboBee, a mechanical bee, smaller than a paper clip, that can take off, fly, and land. Although research for the RoboBee was funded by the National Science Foundation, its creator, Robert Wood, has previously been funded by darpa and the Air Force. (J. Scott Turner, of the extended-organism thesis, has also been funded by the military.) An influential paper published by the Center for a New American Security, “Robotics on the Battlefield Part II: The Coming Swarm,” cites the RoboBee as evidence of the possibility of 3-D-printed, less-than-a-dollar-apiece drones that, in vast quantities, could “flood” civilian and combat areas as “smart clouds.” As Margonelli writes, “Everything termites do, the military would like to do, too.” The military would like to have weapons that are at once tiny (like termites) and massive (like swarms)—weapons that are easy to maneuver and hard to detect, but also smart and lethal. One researcher in Nagpal’s lab tells Margonelli, “We can’t stop the technology because it might be used for bad.”
Indeed, synthetic swarm intelligence is already with us. A few years ago, the U.S. Navy began testing swarms of autonomous, self-organizing robotic speedboats. In 2012, Human Rights Watch and the Harvard Law School’s International Human Rights Clinic called for a preëmptive, international ban on the development of fully autonomous weapons. The same year, the Department of Defense issued a directive that stopped far short of banning autonomous weapons, requiring only that a human be somehow involved whenever they are used to deliver lethal force. Mark Hagerott, the former deputy director of the Center for Cyber Security Studies at the Naval Academy, favors stringent restrictions on the development of swarming weapons, including limits on size (no smaller than a human), fuel sources, and numbers. He worries that, with both semi-autonomous and autonomous weapons, it is increasingly difficult to identify the crucial place where finger meets trigger. This matters, Hagerott says, because this is the place where empathy is exercised, when it is exercised, during warfare.
What is less often mentioned by critics of autonomous weapons is that there is something valuable in the high casualty rate of conventional warfare. If war costs states nothing but money, what is there to hold them back? What will stop a bellicose government from pursuing its foreign projects, if there are no body bags to focus its citizens’ outrage?
The termite is no longer what it was to earlier observers: a model of what humans could be—more coöperative and harmonious, less competitive and aggressive. Instead, it has become a resource to be harnessed for the achievement of our own, already established, ends.<
This article by Professor Amia Srinivasan (University of Oxforf) first appeared in The New Yorker’s September 17, 2018 print edition issue under the headline “Busy Bodies.” (https://www.newyorker.com/magazine/2018/09/17/what-termites-can-teach-us?mbid=nl_Daily%20091118&CNDID=46256997&utm_source=Silverpop&utm_medium=email&utm_campaign=Daily%20091118&utm_content=&spMailingID=14227848&spUserID=MTcwMjEwOTU5NDMyS0&spJobID=1480908966&spReportId=MTQ4MDkwODk2NgS2