16.1 — The Patterns That Repeat

Mark Twain is supposed to have said that history does not repeat itself, but it rhymes. There is no evidence he said it.

Which is a suitable opening for this chapter, because the first thing to establish is how much weight a historical pattern can actually carry.

What a pattern in history can and cannot be

It cannot be a law. History has no controlled experiments, one run of the tape, and every event has thousands of causes. Anyone who claims a historical law that predicts the future is selling something, and the twentieth century's two great attempts to build politics on such laws — the Marxist stages of history and the racial theories of Chapter 11.6 — both produced catastrophe.

What it can be is a mechanism. A causal process that has operated repeatedly in identifiable circumstances, which tells you what to look for and what is likely rather than what will happen.

The test for whether a proposed pattern is real: does it name a mechanism rather than a resemblance; does it have counter-examples and does it explain them; and was it stated before the outcome or fitted afterwards.

Everything below meets that test as well as this subject allows, and each is stated with its limits.

1. Institutions outlive the conditions that created them

The mechanism. An arrangement is built to solve a problem. It works. The problem changes. The arrangement persists, because the people who benefit from it are organised and those who would benefit from changing it are not.

The Roman republic's constitution worked for a city-state and could not govern a Mediterranean empire, and the attempt destroyed it (Chapter 5.3).

The Mughal jagir system worked while the empire was expanding and produced a crisis when it stopped (Chapter 6.12).

The Licence Raj was a defensible response to 1950 and cost India four decades (Chapter 6.28).

And the Security Council's composition reflects 1945 (Chapter 13.7).

What it predicts. When an institution is defended by describing what it achieved rather than what it does, it is in this phase.

2. The winning asset becomes the losing one

The mechanism. A capability that produces dominance attracts investment, becomes identity, and is retained past the point where it works, because abandoning it means abandoning the people and the status built around it.

War elephants, effective for two thousand years and a liability the moment artillery arrived (Chapter 6.15).

The Maratha adoption of European-style infantry, which abandoned the doctrine that had beaten Aurangzeb (Chapter 6.13).

British naval and industrial supremacy in 1900, and an economy still organised around it in 1930.

And Kodak, Nokia and the standard corporate cases.

What it predicts. The more comprehensively an organisation won with something, the longer it will keep it.

3. Efficiency buys fragility

The mechanism. Removing redundancy raises performance under normal conditions and reduces the capacity to absorb a shock. Every removal is individually rational.

The Bronze Age collapse: a tightly coupled trading system with no slack, hit by drought (Chapter 4.6).

Buddhism concentrated in a few great monasteries, which did not survive their destruction while household traditions did (Chapter 6.8).

Financial leverage in 2008 (Chapter 12.8).

And single-source supply chains in 2020 (Chapter 12.10).

What it predicts. A system optimised to a single objective is a system that will fail when conditions move outside the range it was optimised for.

4. The trap that locks behind you

The mechanism. A change that makes each individual worse off but supports more people cannot be reversed, because reversing it requires the additional people to not exist.

Agriculture (Chapter 3.7): more calories per hectare, fewer per hour of work, and no way back once the population had grown.

Irrigation and salinisation in Mesopotamia (Chapter 4.1).

Fossil fuels, which now sustain a population that could not be fed without industrial fertiliser (Chapter 14.6).

And urbanisation.

What it predicts. A technology that raises carrying capacity will be adopted regardless of its effect on individual welfare, and cannot be abandoned without a substitute.

5. Terror is cheap and brittle; co-option is expensive and durable

The mechanism. Rule by fear requires no consent and generates no loyalty, so it must be maintained continuously and collapses completely at the first serious defeat, because everyone defects at once.

Assyria, which vanished within a few years of losing a battle (Chapter 4.7). Persia, which lasted two centuries by co-opting local elites.

Rome's grant of citizenship (Chapter 5.3). Akbar's Rajput settlement against Aurangzeb's reversal (Chapters 6.11 and 6.12).

And the Nazi occupation of the Soviet Union, which turned a population that might have welcomed liberation into partisans within months (Chapter 11.7).

What it predicts. A regime that has no constituency with a stake in its survival will fall faster than its apparent strength suggests.

6. A coalition loses to a unified command

The mechanism. A coalition's members have their own interests and calculate what happens after the battle as well as during it, so defection is individually rational even when it is collectively catastrophic.

Chapter 6.15 built an entire argument on this — Tarain, Talikota, Panipat, Plassey, the Anglo-Maratha and Anglo-Sikh wars.

And it is not an Indian phenomenon. The Greek city-states against Macedon. The Italian states against France in 1494. The German states before 1871.

What it predicts. Where an attacker can make credible separate offers, a defensive coalition will fragment.

7. The decision's speed matters more than its content

The mechanism. Under time pressure, the option that is already prepared is the one taken, and the alternatives are not generated.

July 1914, where mobilisation timetables gave diplomats hours (Chapter 11.2).

Partition's ten-week timetable, which multiplied the deaths from a decision taken over years (Chapter 6.23).

The dissolution of the Iraqi army in 2003 (Chapter 12.9).

And the post-2001 legislation passed within weeks of an attack that lasted 102 minutes.

What it predicts. The consequential decisions are made in the days after a shock, and they are made from whatever was on the shelf.

8. Information that cannot travel upward destroys the system that blocks it

The mechanism. Where reporting bad news is dangerous, the reports become favourable, and the leadership acts on false information.

China's Great Leap famine, where inflated harvest reports determined procurement (Chapter 12.3).

Soviet planning (Chapter 9.6).

The Ottoman and Mughal courts' late awareness of relative decline (Chapter 8.10).

And the corporate and military versions, which recur constantly.

And the counter-case is the argument's strongest evidence. India has had no famine since independence, and Sen's explanation is elections plus a free press (Chapter 6.17). The mechanism is not democracy as a value; it is information as a function.

9. Formal abolition does not abolish the structure

The mechanism. A legal status is created by law and maintained by economics and custom, so removing the law leaves the rest.

American emancipation in 1865, followed by a century of sharecropping, convict leasing, disenfranchisement and segregation (Chapter 10.8).

Untouchability abolished by Article 17 in 1950 and present in employment, marriage and violence statistics today (Chapter 6.28).

And formal decolonization with economic dependence intact (Chapter 10.9).

What it predicts. A reform that changes the law without changing the incentives will be absorbed.

10. Universal language cannot be confined to those who used it

The mechanism. A principle stated in general terms is available to everyone the words cover, and the people who stated it cannot subsequently narrow it without conceding that it was never a principle.

"All men are created equal", written by a slave owner, used by abolitionists, suffragists and the civil rights movement (Chapter 10.2).

The French Declaration, taken literally in Haiti (Chapter 10.5).

British constitutional principles quoted back at Britain by Indian nationalists, which was the most effective weapon available (Chapter 6.20).

And the Universal Declaration used against the states that drafted it.

What it predicts. This is the most optimistic pattern in the volume, and it is the mechanism by which the moral circle has actually widened.

11. Prices and rules do not defeat each other

The mechanism. Where a law contradicts an incentive strong enough, the law is evaded rather than obeyed.

The Statute of Labourers after the Black Death, which fixed wages at pre-plague levels and failed because employers needed the harvest in (Chapter 7.7).

Diocletian's price edict (Chapter 9.8).

Prohibition, and the drug trade.

And the corollary, which is the useful half. A rule aligned with an incentive works almost effortlessly. The Montreal Protocol succeeded partly because compliance was profitable for the largest producer (Chapter 14.5).

12. Contingency is real

The one that limits all the others.

A driver's wrong turn in Sarajevo (Chapter 11.2). A Great Khan's death that turned the Mongol armies around at the edge of Europe (Chapter 7.6). An asteroid arriving when a sulphur-rich shelf was underneath rather than deep ocean (Chapter 1.10). Arkhipov saying no (Chapter 12.2).

Which means the patterns above describe tendencies in a system with enormous variance, and any account that reads history as inevitable is wrong.

What to do with these

They are not predictions. They are questions to ask.

Is this institution defended by what it once achieved? Is this advantage being retained past its usefulness? Where is the slack in this system? Who has a stake in this arrangement surviving? Can bad news reach the top here? Does this reform change the incentive or only the rule?

Those questions are usable in an organisation, a career and a country, which is the practical claim this volume makes for the study of history.

What the next page covers

Chapter 16.2 takes the sharpest version of the question: why do civilizations collapse? It examines the actual cases in this volume — the Bronze Age, Rome, the Maya, the Indus, the Soviet Union — identifies what they share, tests the popular single-cause explanations, and asks the uncomfortable question of what the model implies about the present.