The Colony Makes The World
On the emergent intelligence of ant colonies; ecologies of entanglement; a New Release!
Ants are 140 million years old, older than us by far, older even than the dinosaurs. There are some 14,000 known species of them, on every continent save Antarctica. Only 50 have been studied in any real depth. Some farm fungus; some enslave other ants; some weave nests from larval silk; some build leaf castles mortared with aphid spit; some build bridges from their own bodies; some keep other insects as livestock; some work security for trees in exchange for nectar, a home.
Everything ants do, they do without central control—even though individual ants are dumb, mostly blind, and can’t remember anything for more than ten seconds.
When’s the last time you watched an ant? There’s one walking across the café table where I sit and write these words now. This table must be an Everest of steel to her—what an enormous expenditure of energy for a little ant, and all just to wander aimlessly around, stopping here and there to inspect a speck of dust! She seems to have no idea what she’s doing, and I have a hard time imagining that this solitary, bumbling individual will ever find what she seeks. “If you watch ants at all, you end up wanting to help them, because it seems that the ant you’re watching just can’t seem to get it together,” explains Dr. Deborah Gordon. “But in the aggregate, somehow, they get a lot of things done.”
Dr. Gordon, by her own estimation, has “watched more ant colonies for longer than any other scientist, and for longer than most ant colonies have watched each other.” For over 30 years, she’s spent her summers in the Arizona desert, watching the same colonies of Harvester Ants grow older, larger, and wiser. Individual ants come and go, of course: an ant’s lifespan is only about a year. But Gordon has shown that colonies mature, retaining memories individual ants forget.
But back to the ant on the table. There’s nothing here for her. Were I to drop a crumb of cookie onto her path, though, she’d make the most of it. She’d take some and head straight home, laying a trail of pheromones as she went. The next ant to cross that trail in her own aimless wanderings would drop everything and follow it to its end. Finding cookie there, she’d do the same as her predecessor, reinforcing the trail, and once enough ants echoed this task, it’d become a highway, and I’d need to relocate.
This is called a recruitment trail; it’s an example of positive feedback in ant behavior. The more pheromone is laid down, the more ants are drawn to the trail, each adding more pheromone until there are enough ants to make quick work of bringing the food home. But the pheromone is volatile, too, and will evaporate if it’s not reinforced; this secondary mechanism ensures that the shortest trails are always the strongest, and keeps the ants from wasting time on depleted food sources. Efficient.
Many ant species use recruitment trails to siphon workers towards food sources, but the network structure of these trails varies. With the pharaoh ant, an opportunistic species that specializes in making the best of patchy resources, the trail algorithm favors recruitment; its branching structure funnels ants into a recruitment pool, making it easier for those ants who have found food to lure others quickly. Pharaoh ants have three different trail pheromones: a strong attractant, a weak one, and a repellent, which they use to mark dead-ends. With finer chemical control, they can prioritize paths and directly recruit foragers off the main trail to take advantage of an ephemeral food source—say, the sudden appearance of a human picnic.
Turtle ants, however, live in small colonies in the canopies of tropical forests, where vines snake through the branches and plants snap under the footsteps of lizards and birds. The turtle ants must constantly maintain the circuit of trails connecting their nests to nearby food sources lest they be swallowed whole by the green life of the jungle. Here, the algorithm favors coherence; the default, for turtle ants, is to soldier onwards, laying down a slowly-evaporating pheromone at all times. When they reach a dead end, some percentage always explore further, searching to find and heal the broken link in the trail. These variations, Gordon Argues in her book The Ecology of Collective Behavior, demonstrate how the collective behaviors of colonies emerge from ants’ relationship to a dynamic environment. The world makes the colony.
You’ll notice I’m using the word “algorithm.” It’s quite literal—an algorithm is just a sequence of instructions one must follow to achieve a task. A recipe is an algorithm, technically. For the Harvester ants that Gordon studies, that recipe might be something like: wait, until you encounter another ant with odor X three times in the next 30 seconds. If you do, go forage for food. If you find a seed, return it to the nest, then wait again.
But even a simple algorithm, running on hundreds of ants, each interacting with their environment and updating their tasks as the information they receive changes, can produce complex ripple effects. When a colony’s circumstances change—say, there is an abundance of food—it’s able to instantly reallocate its workers to different tasks. And in a larger colony, where the rate of interaction between ants is higher, the same algorithm might lead to different behaviors than in a smaller, younger colony. To understand these dynamics, myrmecologists like Gordon work with computer scientists and engineers to create quantitative models of how ant colonies work, porting their algorithms from sandy anthills, to symbols, to silicon.
The results are, often, humbling. Gordon has shown that the algorithm Harvester ants use to regulate their foraging behavior across the desert is uncannily similar to the Transmission Control Protocol used to regulate data traffic on the internet, for example. Meaning that ants beat us to network design by a hundred million years.
Are there other algorithms regulating the “anternet” that might help us to build more efficient networks of our own? Already, algorithms inspired by ant recruitment trails are used to solve combinatorial optimization problems in computer science, and one researcher has proposed that living “ant computers” are theoretically possible—an unsettling vision, but one familiar to anyone who has read the British science fiction novel Children of Time, in which hyper-intelligent alien spiders use sophisticated chemistry to turn neighboring ant colonies into server farms.
In one sense, this isn’t new. Edward O. Wilson, probably the most famous scientist to ever study social insects, often wrote about ants in mechanistic terms, comparing colonies to factories. In doing so, he drew heavy inspiration from cybernetics, the science of communication and control. For most of its history, the study of ants—and more broadly, biology as a whole—has been in thrall to this vision of behavior. Ants were merely automatons, going about pre-programmed roles following chemical cues. The division of labor in a colony was as inflexible as a caste system: the moment they emerged from their pupa, some ants patrolled, some foraged, some piled the bodies of the dead, some kept the tunnels clean, some fed the brood, each a piece of a clockwork, enslaved from egg to midden-pile.
Only in the last few decades has the true flexibility of ant society come to light, thanks to Gordon and a new generation of systems-minded scientists. Even in species where ants vary widely in size, any ant’s role is determined by the dynamic, ever-changing needs of the colony—and, just as importantly, by an environment that is both constantly changing and constantly being changed by the ants. As Gordon observes, the world makes the colony.
But the colony, too, makes the world.
🐜
I’ve been at MacDowell for a little over a week. My little studio—which has sheltered, over the years, Audre Lorde, Celine Song, Michael Chabon, and, in an extraordinary coincidence, my college mentor, the poet Martha Ronk—is already strewn with print-outs, library books, and jotted notes. Every day I ride my bicycle around the property, huffing forest, braking for wild turkeys and deer. Every artist deserves this.
Finally, some recent news from the outside world:
My band, YACHT, has a new album coming at the end of August. In an experimental move, we launched a “reverse-preorder” campaign in July, selling New Release on vinyl two months before its streaming premiere; the first edition sold out immediately, but a second wave is live now. We talked to Yancey Strickler about staying willfully independent, making art under capitalism, and transcending the sinking feeling that “your house is a store and your life is a job.”
I’m extremely pleased to be featured in Ecologies of Entanglement, a new interview series from Are.na (truly the ant colony of my mind) and Willa Köerner’s wonderful newsletter Dark Properties. The series will draw connections between natural and technological networks. My interview is here—read on if you’re interested in natural systems, virtual worms, speculation, and slime mold.
My friend Kriss Knapp is launching a series of “Object + Word” collaborations, pairing poster prints with commissioned texts from writer friends. My contribution is a little ecological poem; the print, with an illustration by Kriss, comes out on the 6th. If you subscribe to her Substack you’ll see it there first.
xo
Claire
this was a lovely read, thank you