What Is Coevolution?

The word looks like it belongs in a biology textbook. For a long time it did. Coevolution was coined in 1964 by Paul Ehrlich and Peter Raven to describe what they saw between butterflies and the plants those butterflies fed on: each was steering the other's evolution, generation by generation, in a slow reciprocal dance. Neither side could be understood in isolation.

Sixty years later, the word has quietly escaped the lab. Economists use it to describe technologies and institutions. Anthropologists use it to describe humans and their tools. Designers use it to describe products and their users. We use it as the name of this publication. This is what it means, where it comes from, and why we think it is the most useful single word for the era we are living through.

The one-sentence definition

Coevolution is what happens when two or more things change in response to each other, so that change in one becomes a selection pressure on the other, and the cycle repeats.

That is the whole idea. Everything else is examples.

Coevolution in biology — the classic examples

The textbook case is an orchid and a moth. In 1862 Charles Darwin received a specimen of Angraecum sesquipedale, a Madagascan orchid with a nectar tube nearly thirty centimetres long. He predicted that somewhere on the island a moth must exist with a tongue long enough to reach the bottom. Forty years later, after his death, entomologists found it: Xanthopan morganii praedicta, the predicted hawk moth. Neither orchid nor moth makes sense on its own. Each is the answer to a question the other has been asking for millions of years.

Predator and prey do the same dance, less elegantly. Cheetahs run faster, gazelles run faster, cheetahs run faster again. Biologists call this an arms race. Neither side ever wins; both sides keep changing.

Leafcutter ants and the fungi they farm are a third case, and the most intimate. The ants do not eat the leaves they carry. They feed them to a fungus that lives only inside their nests, and they eat the fungus. Neither the ant nor the fungus can survive without the other. They have been coevolving for roughly fifty million years, which is longer than humans have existed as a species.

Mimicry is the strangest case. Some butterflies look almost identical to other, more poisonous butterflies. Birds learn to avoid the poisonous ones, and the harmless mimics get a free ride — until enough mimics appear that birds stop being deterred, and the original poisonous species has to evolve a new warning pattern. The signal coevolves with the response to the signal.

The mechanism, in plain language

What unites these examples is a feedback loop. Trait A appears in species 1. It changes the environment of species 2 — what species 2 has to eat, escape, attract, or imitate. Species 2 evolves trait B in response. Trait B changes the environment of species 1, which evolves trait A′. The two histories become inseparable.

Coevolution is not cooperation, and it is not competition. It is what happens when two systems become each other's environment.

This is worth holding onto, because it is the part that travels best out of biology. Coevolution does not require intent. It does not require partnership. It does not even require that both sides benefit. It only requires that two systems are coupled tightly enough that change in one reliably changes the other.

Why the word escaped biology

The first people to take coevolution out of biology were economists and historians of technology. Brian Arthur, at the Santa Fe Institute, argued that technologies evolve by combining and recombining existing technologies, and that each new technology changes the landscape in which the next one will appear. Kevin Kelly, in What Technology Wants, called the whole accumulating system the technium and described it as a seventh kingdom of life — not alive in the biological sense, but evolving by analogous rules and entangled with us at every step.

Anthropologists made a parallel move. Cooking, they argued, did not just feed our ancestors; it selected for smaller jaws and bigger brains. Writing did not just store information; it rewired memory and attention. Cities did not just house people; they selected for the kinds of people who thrive in cities. The tool and the toolmaker shape each other. Marshall McLuhan put it most bluntly: we shape our tools, and thereafter our tools shape us.

Once you start looking, the pattern is everywhere. Roads and cars coevolve. Markets and regulators coevolve. Platforms and the behaviour of their users coevolve. The word is a lens, and the lens fits a lot of things.

Coevolution today — humans and machines

Which brings us to now.

The smartphone is fifteen years old. In that time it has changed what attention feels like, what waiting feels like, what loneliness feels like, what a friendship looks like across a continent. We built the device to fit our pockets and our hands. The device, in turn, has reshaped what our hands and minds reach for when they are idle. That is coevolution, running fast.

Large language models are about three years into their public life. They are already changing how people write — emails, code, memos, school essays, this sentence. They are changing how people read, because more of what is read was generated. And they are changing how people think, because the cognitive labour of drafting, summarising, and explaining is being offloaded to a system that learned from our previous drafts, summaries, and explanations. The next generation of these models will be trained partly on text shaped by the current generation. Trait A, trait B, trait A prime.

Recommendation algorithms and the people they recommend to are coevolving in a tighter loop than almost anything in biological history. The feed learns what you click; you learn what the feed shows; what you click changes; the feed changes again. A few iterations of that loop is a year of human life. A few hundred is a generation of cultural change. Nobody designed it as coevolution. It became coevolution because the coupling was tight enough.

None of this is good or bad on its own. Coevolution is morally neutral, the way gravity is morally neutral. But it has consequences, and the consequences arrive faster than the institutions built to govern them. That is the territory this publication tries to cover.

Why we are called The Coevolution

We chose the name because most of the conversation about technology still treats it as something that happens to us — a wave to brace for, a tool to deploy, a risk to mitigate. That framing is a hundred years old and it is no longer accurate. The systems we are building are now building us back. Pretending otherwise produces bad journalism and worse policy.

Calling the publication The Coevolution is a small act of insistence: that AI, work, governance, and human development are not four separate beats but one story told from four sides. That the right unit of analysis is the loop, not the link. That writing about technology without writing about the people it is shaping — and writing about people without writing about the technology shaping them — leaves the most important part out.

If that framing is useful to you, the work lives in our sections, our latest issue, and the Sunday newsletter. Start anywhere. The loop will find you.