Societal collapse

Why do complex societies sometimes collapse rather than simply shrink?

If a society is in trouble, the intuitive expectation is a gentle decline: less trade, a smaller army, fewer monuments, a slide back down the slope it climbed. What the archaeological record sometimes shows instead is a fall off a step — writing stops, cities empty, long-distance goods vanish, the whole administrative apparatus gone within a couple of generations. Why should the way down not simply retrace the way up?

First, be careful what we are asking about. Define collapse the way the archaeologist Joseph Tainter did: a rapid, substantial loss of an established level of sociopolitical complexity. Not everyone dying — fewer distinct social roles, less specialisation, less central control, less information moving, shorter trade routes, smaller units. People may survive perfectly well; what ends is the elaborate arrangement they were living inside.

Now ask what that complexity was for. Tainter's answer, worth taking seriously as one influential account rather than as settled fact, is that complexity is a problem-solving strategy that costs energy to run. A society facing a shortfall builds irrigation, or a bureaucracy, or a standing army, or a road network, and each buys real returns.

But which problems get solved first? The cheap and obvious ones. The best land is farmed before the marginal land; the easy administrative fix comes before the elaborate one. So each further increment costs more and returns less: you begin by hiring administrators and end by hiring administrators to administer the administrators. The returns curve bends over — and a society can keep investing long past the point where the investment pays.

Here is the step that answers the original question. When marginal returns are high, the arrangement is worth supporting, and shocks — a bad harvest, a lost battle, an epidemic — come out of the surplus. When returns have flattened, two things are true at once: there is less reserve to absorb a shock, and participation has become a poor deal for the parts. A province paying heavily and receiving little has an incentive to stop paying.

And complexity is interlocked. The specialists depend on the surplus, the surplus on the trade, the trade on the security, the security on the tax, and the tax on the specialists. Withdraw support from one layer and you do not get a slightly smaller version of the same structure; you remove what other layers were standing on. That is why the descent can be fast: the system does not slide down its own curve, it drops to a far simpler arrangement that is self-supporting without the missing layers. In Tainter's framing, collapse is less a catastrophe befalling a society than an economising choice made in pieces, by many hands.

Two honest cautions. The first is about evidence: the markers of complexity in the record — writing, monumental building, imported goods, coinage — are exactly what disappears first and most visibly, so the record itself exaggerates abruptness. The second is about the word. Archaeologists have pushed back hard on popular collapse stories, arguing that what collapsed was usually a state and its elite, while rural populations, languages and practices continued. Maya cities in the southern lowlands were largely abandoned across the eighth and ninth centuries, but Maya people, and cities further north, did not vanish. The Easter Island "ecocide" narrative in particular is heavily contested: rat predation on palm seeds, and above all European contact, slave raiding and introduced disease, now carry far more weight than a story of islanders felling their last tree.

Nor is Tainter the only account on offer. Climate shocks, epidemic disease, conquest, and models of elite overproduction all have serious advocates, and specific cases usually look like several at once.

The reasoning connects up like this: complexity is bought, not free; what is bought cheaply first is bought dearly later; and a structure sustained by the participation of its parts fails suddenly when participation stops being worth it. The abruptness comes from interdependence, not from the severity of the shock — which is why the same drought or invasion that a society shrugs off in one century ends it in another.

Two common readings need correcting. Collapse is not a verdict about decadence; nothing in the account requires anyone to have behaved badly. And it is not extinction: a simpler society is often perfectly viable, and for many ordinary people the loss of a demanding central state was not obviously a loss at all.

• Why some societies survive shocks that end others: the resilience literature on redundancy and loose coupling.
• What the "transformation, not collapse" school does with the Roman west and the Maya lowlands.
• Whether modern globally-coupled systems make collapse less likely or merely larger in scale.

Every term the collection defines is gathered in the glossary.