Why Civilizations Collapse: A Society Shedding Complexity Past a Tipping Point

When a civilization falls, we reach for stories: it grew decadent, or an enemy destroyed it. There is a more structural way to see it, and it travels better across cases. Joseph Tainter (1988) defined collapse not as decline and not as conquest, but as a rapid, large loss of an established level of sociopolitical complexity. A society builds up administration, infrastructure, specialists, all to solve problems, and that complexity pays off, at first. The trouble is that the payoff shrinks. Each new layer of complexity solves a little less and costs a little more, until the society is straining to maintain an elaborate apparatus that no longer earns its keep. Then a shock that a leaner society would have shrugged off can instead trigger a fast shedding of complexity, down to a simpler level it can actually sustain. The suddenness is not a puzzle; it is the point.

Is collapse sudden or gradual?

That suddenness has a familiar shape in other sciences. Other systems do this too, a lake, a climate, a market. They look steady, hold steady, then flip fast at a tipping point. Scientists call it a critical transition. Marten Scheffer and colleagues (2009) showed these transitions are everywhere, and that they often announce themselves: as the system nears the edge, it recovers more slowly from small disturbances and its fluctuations grow. Put next to Tainter, a collapsing society looks like a critical transition, with the loss of complexity as the flip and diminishing returns as what drive the system toward the edge. The reading does come with a condition. A true tipping point needs the high-complexity state and the collapsed state to be genuinely separate stable states, so that the system jumps between them rather than sliding smoothly down one slope; the early-warning signs are meaningful only in that case. One more caution: those warning signs are generic, shared by many kinds of tipping points, so on their own they show that something tipped, not that diminishing returns were the cause. What pins the cause is the falling return on complexity arriving alongside, and ideally ahead of, the warning signs. No virtue or villainy required, just dynamics.

The same shape, a structure propped up by ever-larger investment under falling returns until a shock tips it into rapid simplification, is tempting to see in other over-extended systems. I am only gesturing at that, and nothing here rests on it. The solid claim is about societies: collapse is a loss of complexity, with the dynamics of a threshold crossing.

It is testable, and the test looks at the run-up. Two things should show before a collapse. The returns on added complexity should be visibly falling, the society paying more for less; you could track this by watching output per person against the size of the administrative and specialist apparatus, and asking whether each added layer buys less than the last. And the generic early-warning signs of an approaching transition should appear, slower recovery from shocks and rising variability in key measures, before the fall itself. These two signs test different things. If the returns are not falling, the diminishing-returns story is wrong. If there is no slowing-down signature beyond normal noise, only the tipping-point framing fails, and the plainer simplification story still stands. The account holds where both signs reliably precede the shift, with the falling returns arriving first.