Violence has declined by orders of magnitude, relatively speaking

Pinker's central empirical claim is that rates of violent death — from tribal warfare, homicide, judicial torture, and war — have fallen dramatically across human history when measured as a proportion of population, even though absolute death tolls from twentieth-century events like the World Wars are horrifyingly large in raw numbers. He compiles archaeological and anthropological data suggesting that some non-state, tribal societies experienced war-related death rates far higher proportionally than even the bloodiest twentieth-century conflicts inflicted on their participant populations.

He's careful to emphasize this is a claim about relative rates, not a claim that modern atrocities were small or that suffering doesn't matter in absolute terms — the Holocaust and World War Two remain catastrophic tragedies regardless of population-adjusted comparisons. His point is narrower and statistical: the chance any given individual faced of dying violently has fallen substantially over the long historical run, even accounting for the horrific twentieth century.

This distinction between absolute numbers and per-capita rates is the foundation the rest of his argument builds on, and it's also the point critics most frequently contest, given how much the conclusion depends on specific historical death-toll estimates that are often genuinely uncertain.

Takeaway: comparing violence across eras requires comparing rates, not raw body counts — and by that measure, we're living in an unusually peaceful era.