The Long Peace — Violence, War, and the Human Record

The Claim

Steven Pinker’s The Better Angels of Our Nature (2011) makes a bold empirical claim: violence, measured across multiple dimensions and timescales, has declined over the course of human history. Not monotonically and not without reversals, but in the long run, humans today are less likely to die violently than their ancestors were. The 20th century, despite its world wars and genocides, was less deadly per capita than many earlier periods. The post-WWII period — what Pinker and John Gaddis have called the “Long Peace” — has been, in terms of interstate warfare, the most peaceful extended period in recorded history.

The claim is provocative because it cuts against the intuition that the 20th century’s organized mass violence — the Holocaust, the Gulag, the Great Leap Forward, Hiroshima, Rwanda — represents a nadir rather than a relatively peaceful period. The apparent paradox is resolved by the distinction between absolute numbers and rates. The 20th century had more war deaths in absolute numbers than any previous century primarily because it had vastly more people. As a proportion of population, the deaths from the 20th century’s violence are comparable to many earlier periods and less than some (the Mongol conquests, the 30 Years War as a proportion of European population, hunter-gatherer violence rates as documented by archaeology).

The Evidence

Pinker marshals evidence across multiple timescales.

Prehistoric violence. Archaeological evidence from pre-state societies — skeletal trauma analysis, contested burial sites, fortifications — suggests that violence rates among hunter-gatherers and early agricultural societies were higher, as a proportion of deaths, than in modern states. Lawrence Keeley’s War Before Civilization (1996) is the primary source: roughly 15-25% of skeletal remains from many pre-state archaeological sites show evidence of violent death. This compares to 1-2% in 20th century societies, even including the two World Wars.

The homicide decline. Using the records of English coroners from the 13th century forward, combined with later statistical records across Europe, researchers have documented a dramatic decline in homicide rates from roughly 35-70 per 100,000 per year in medieval Europe to roughly 1-2 per 100,000 in contemporary Western Europe. This is a 20-50x decline over five centuries. The decline is real, cross-national, and occurs without an obvious single cause — it is attributed by Norbert Elias’s “civilizing process” theory to the internalization of norms of self-restraint associated with court culture and state formation.

The Long Peace. Since 1945, no two great powers have fought each other directly. The number of interstate wars has declined. The duration of wars has shortened. The number of deaths from organized interstate violence per year is, at the time of Better Angels’ writing, near historic lows. Nuclear deterrence, the spread of democracy and economic interdependence, and the emergence of international institutions are the proposed mechanisms.

The Criticisms

Better Angels has attracted serious criticism from multiple directions.

The framing of the 20th century baseline. Nassim Taleb and Pasquale Cirillo argue that Pinker’s statistics are sensitive to the baseline chosen and the fat-tail properties of war. If you’re drawing from a distribution with fat tails (where rare catastrophic events are much more likely than a Gaussian model predicts), low recent violence doesn’t strongly imply a structurally lower violence process. The Long Peace could be a lucky run in a fat-tailed distribution rather than evidence of an underlying structural change.

The genocide accounting. Some critics argue that Pinker’s accounting underweights colonial violence, structural violence (deaths from poverty, preventable disease, and inequality that are caused by human arrangements even if not by direct physical force), and the long-term consequences of historical violence. If you count colonialism’s deaths from famine, disease, and displacement, the 20th century’s numbers look worse.

The reversal since 2011. The period after Better Angels was published has been rougher. The Syrian civil war, the rise of ISIS, the Russian invasion of Ukraine (2022), and increasing great-power tensions have complicated the “Long Peace” narrative. Whether these represent reversals of the trend or variations within a continuing decline depends on timescale and measurement.

The mechanisms are understated. Critics note that Pinker’s proposed mechanisms for the decline — the civilizing process, democracy, trade, international institutions — are less causally established than the statistical trends he documents. The correlations are real; the causal inference from correlation to mechanism is contested.

What Is Genuinely Established

The debates about Better Angels shouldn’t obscure what is robustly established:

Pre-state violence rates were high. Archaeological evidence of prehistoric violence is extensive and shows a level of violent death that would be considered catastrophic by modern standards. This is not a controversial finding.

Medieval European homicide rates were dramatically higher than modern ones. The coroner records don’t lie. The magnitude of the decline is striking regardless of its cause.

The post-WWII period has been unusual for great powers. No direct interstate war between nuclear powers has occurred. This is a historical anomaly that requires explanation even if its causes are disputed.

State formation generally reduces certain forms of violence while potentially enabling others. Hobbes was right that the pre-state condition is more violent in the sense of interpersonal and small-group violence. He was wrong, or incomplete, in not accounting for the state’s own capacity for organized mass violence against both internal and external populations.

The Normative Question

The debate about Better Angels is partly about statistics and partly about what questions we’re asking. If the question is “have rates of physical violence declined over long historical periods?”, the answer appears to be yes, with important caveats. If the question is “is the world becoming more peaceful?”, the answer depends on what you count as violence and what timescale you’re looking at.

Pinker’s work is most valuable as a corrective to a specific kind of pessimism: the intuition that violence is an inherent feature of human nature that cannot be reduced and that the modern period is uniquely brutal. The data doesn’t support this intuition. Human societies have reduced interpersonal violence dramatically over the past few centuries through identifiable mechanisms, which means further reduction is possible.

It is least valuable as grounds for complacency. The fat-tailed character of war statistics means that low recent violence doesn’t strongly predict low future violence. The Long Peace has been sustained partly by nuclear deterrence, which works until it doesn’t. The institutional and normative progress Pinker identifies has been real but is not guaranteed to be permanent.

The historical record suggests that violence is reducible — that it is not inevitable at any particular rate — but also that reductions achieved by specific institutional and cultural arrangements can be reversed when those arrangements are disrupted. The right response to Pinker’s data is not triumphalism but careful attention to what has produced the progress and what could reverse it.