11.11 — The Other Catastrophes: Famines, Purges, Genocides

The twentieth century's wars killed enormous numbers of people. Governments killed comparable numbers of their own citizens, or of people under their control, outside of any war.

This chapter covers those events, and it is written around a single question: is there a pattern, and can it be recognised before rather than afterwards?

There is, and the pattern is specific enough to be useful.

The famines that were policy

Chapter 6.17 established the principle with Amartya Sen: famine is rarely about the absolute availability of food. It is about entitlement — whether particular people have the means to obtain it — and it is therefore a political outcome.

The Soviet famine of 1932–33 (Chapter 11.5). Around five to eight million deaths. Grain procurement maintained during the famine, movement restricted, villages blacklisted, exports continued.

China's Great Leap Forward famine, 1959–61, is the largest famine in recorded history. Estimates range from around 15 million to 45 million deaths, with figures around 30 million widely cited. Chapter 12.3 covers the mechanism in full: communal farming that destroyed incentives, backyard steel furnaces that pulled labour from the fields, and — decisively — local officials reporting inflated harvests, so the state took a share of a crop that did not exist.

And the information failure is the specific point. Nobody could report the truth without being accused of defeatism. The system produced the numbers it wanted to hear and then acted on them.

North Korea's famine of the 1990s, several hundred thousand to over a million deaths, from the collapse of Soviet subsidies, floods, and a system unable to adapt.

And the colonial famines of Chapter 6.17.

The common feature is not weather. It is a government that has both the power to extract and no mechanism that requires it to hear bad news.

The purges

The Soviet Great Purge (Chapter 11.5): around 680,000 to 750,000 executions in 1937–38 alone, with quotas set centrally and increases requested locally.

China's Cultural Revolution, 1966–76 (Chapter 12.3): estimates of deaths range widely, commonly from several hundred thousand to two million, with tens of millions persecuted, universities closed for years, and a generation's education lost.

Cambodia under the Khmer Rouge, 1975–79, is the most extreme case relative to population. Around 1.5 to 2 million deaths out of a population of about 8 million — roughly a quarter of the country.

Its logic was total. Cities were emptied within days of the takeover; Phnom Penh's entire population was marched into the countryside. Money, markets, private property, schools and religion were abolished. The calendar was reset to Year Zero. Anyone associated with the old order — officials, teachers, doctors, monks, and, notoriously, people who wore glasses or spoke a foreign language — was a target.

And the Tuol Sleng prison kept photographic records of everyone it processed. Around 15,000 people passed through it; a handful survived. The photographs exist because the regime documented its own work, which is the same fact as Chapter 11.9's minutes.

The genocides

Four cases, and the pattern is what to take from them.

Armenia, 1915

The Ottoman government deported and killed its Armenian population during the First World War. Estimates run from around 600,000 to 1.5 million deaths, through massacre and through deportation marches into the Syrian desert without food or water.

Turkey does not accept the term genocide, arguing that the deaths were a consequence of wartime relocation and general chaos. A majority of genocide scholars and a growing number of states classify it as genocide.

Its importance beyond itself: Raphael Lemkin, who coined the word genocide, cited the Armenian case as a formative example. And a remark attributed to Hitler in 1939 — asking who, after all, speaks today of the annihilation of the Armenians — is disputed in its exact wording and captures a real point about impunity.

Rwanda, 1994

Around 800,000 to a million people killed in approximately a hundred days, principally Tutsi and moderate Hutu.

Its origins are colonial (Chapter 11.1): Belgian administration converted a partly fluid social distinction into a fixed legal category recorded on identity cards, and favoured one group in administration and education.

The preparation was visible in advance. Machetes were imported in quantities far exceeding agricultural need. Militia were trained. Radio Télévision Libre des Mille Collines broadcast dehumanising propaganda for months, calling Tutsi inyenzi — cockroaches — and, once the killing began, naming individuals and their locations on air.

And the international response is the part that indicts everyone. A UN force was present. Its commander, Roméo Dallaire, sent a cable in January 1994 reporting an informant's account of the plan, including the arms caches and the intent, and requested authority to act. He was refused. When the killing began, the force was reduced rather than reinforced. The word "genocide" was deliberately avoided in official statements because using it would have triggered obligations under the Genocide Convention.

The identity cards meant people were sorted at roadblocks by a document a colonial administration had issued.

Bosnia, 1992–95

The break-up of Yugoslavia produced ethnic cleansing — a term that entered general use here — and, at Srebrenica in July 1995, the killing of around 8,000 Bosniak men and boys.

Srebrenica had been declared a United Nations safe area. It was defended by a lightly armed Dutch battalion which was overrun and which handed over people who had taken refuge with it.

The International Criminal Tribunal for the former Yugoslavia classified Srebrenica as genocide, and it convicted senior political and military figures including Radovan Karadžić and Ratko Mladić.

Denial of Srebrenica remains politically active in the region.

And others

Darfur from 2003, described as genocide by the United States and by some other bodies, with estimates of several hundred thousand deaths.

The Rohingya expulsions from Myanmar from 2017, with over 700,000 people driven into Bangladesh, and a genocide case brought at the International Court of Justice.

The pattern

This is the point of the chapter. Genocide scholars, principally Gregory Stanton, have described a sequence, and its value is that the early stages are visible and are not violent.

Classification. Us and them, formalised. Identity documents, census categories, legal definitions.

Symbolisation. Names, badges, physical markers.

Discrimination. Legal exclusion — from office, from the professions, from citizenship. Chapter 11.9's Nuremberg Laws.

Dehumanisation. The target group described as vermin, disease, insects. This is the single most reliable predictor — cockroaches in Rwanda, rats and lice in Nazi propaganda, and the same vocabulary in Cambodia and in Bosnia. It removes the ordinary moral inhibition against killing a person by removing the person.

Organisation. Militias, lists, training, weapons distribution.

Polarisation. Moderates within the perpetrator group are silenced or killed first. Intermarried families are targeted. The middle ground is destroyed deliberately, because it is the obstacle.

Preparation. Deportation lists, transport, camps.

Persecution, then extermination, then denial, which is the final stage and which begins immediately and continues for generations.

The practical value is in the early stages. By the time killing begins, only force stops it. Dehumanising language in mainstream media, the legal classification of populations, and the silencing of moderates are all visible years earlier and are all cheap to notice.

Why bystanders do not act

The record is consistent and the reasons are structural.

Sovereignty. Chapter 8.7's Westphalian principle says states do not interfere in each other's internal affairs, and mass killing of a state's own citizens is, in that framework, internal.

Cost. Intervention is expensive, dangerous and politically risky.

Definitional evasion. Rwanda demonstrated that avoiding the word avoids the obligation.

Information. It is always contested at the time. Reports are dismissed as exaggerated, as propaganda, as atrocity stories of the kind that turned out to be false in earlier wars. The scepticism is not always unreasonable and it is exploited.

And prior commitment. Governments allied to the perpetrator find reasons.

The Responsibility to Protect doctrine, adopted at the UN in 2005, was an attempt to establish that sovereignty entails a responsibility and that the international community may act when a state fails it. It was invoked in Libya in 2011, where the intervention exceeded its stated mandate and produced state collapse — and the consequence has been that R2P has been very difficult to invoke since. Chapter 13.7 covers it.

Where this shows up in your life

The Genocide Convention of 1948, the International Criminal Court from 2002, and the tribunals for Yugoslavia and Rwanda are the institutional response, and their record is mixed: real convictions of senior figures, and no jurisdiction over the most powerful states, several of which have not ratified.

And the practical application is the dehumanisation test. When a political movement describes a group of people as an infestation, a disease, a flood, or vermin — consistently, in mainstream outlets, with no cost to the speaker — that is the recognisable early stage. It is worth objecting to at that point, when objection is cheap.

Chapter 16.4 makes it a habit rather than a historical observation.

What the next page covers

Chapter 11.12 closes this Part with the accounting — what the twentieth century actually cost in lives, what it delivered in the same period, and the honest question of whether the world got better or worse, answered with numbers rather than with mood.