A factual note on a widely shared claim about government killing in the twentieth century.
A graphic circulating on social media attributes to political scientist R.J. Rummel the claim that governments killed roughly 170 million people during much of the twentieth century and that concentrated political power is a greater danger to human life than war itself. The graphic captures an important part of Rummel’s argument, but it compresses several distinct claims and should be read with care.
Rummel, a political scientist at the University of Hawaiʻi, coined the term “democide” for killing by governments, including genocide, politicide, massacres, extrajudicial executions and other forms of intentional state killing. In his original twentieth-century work, he estimated that governments and related regimes killed about 170 million people from 1900 through 1987. In later revisions, he raised his global estimate substantially after changing his treatment of colonial deaths and deaths during China’s Great Leap Forward.
Rummel’s larger argument was not simply that governments kill. It was that the concentration of unchecked political power is strongly associated with mass killing. His shorthand was: “Power kills; absolute power kills absolutely.” He argued that systems with open political competition, independent institutions, freedom of expression and meaningful checks on rulers create stronger restraints against mass state violence.
That thesis remains important, but the numbers should not be presented as uncontested fact. Rummel’s estimates were built from thousands of historical estimates, often with very wide ranges. He himself acknowledged that the underlying material was approximate and sometimes politically biased. Later scholars have criticized parts of his methodology. Tomislav Dulić, for example, argued that Rummel’s estimates for Tito’s Yugoslavia relied too heavily on unreliable or partisan sources and questioned whether similar problems might affect other cases. Rummel disputed that critique and argued that one disputed case did not invalidate his much larger comparative dataset.
The most defensible conclusion is therefore not that one precise number has been permanently established. It is that the twentieth century produced an enormous record of intentional killing by governments, and that the relationship between political institutions, accountability and mass violence deserves serious attention. Rummel’s work is best used as a large-scale comparative framework whose individual estimates should be checked against later archival and demographic scholarship.
For readers concerned with mainland China, this distinction matters. Historical analysis should separate wartime deaths, famine mortality, executions, deaths in custody, political campaigns and other forms of state-caused mortality rather than collapse them into a single rhetorical number. The same standard should be applied across regimes and countries. Strong claims require transparent definitions, dated sources and the willingness to revise estimates when better evidence appears.
The enduring question raised by Rummel’s research is therefore institutional: what limits exist on the power of the state, who can challenge official decisions, what information can citizens obtain, and what mechanisms can hold rulers accountable? Those questions can be studied without treating any single numerical estimate as beyond dispute.
Sources and further reading
- R.J. Rummel, “Why Do Governments Kill?” — summary of his 1900–1987 estimate and definition of democide.
- R.J. Rummel, “Power Predicts Democide” — statement of the relationship he proposed between regime power and democide.
- R.J. Rummel, “Democide Since World War II” — discussion of the approximately 170 million estimate.
- Tomislav Dulić, “Tito’s Slaughterhouse: A Critical Analysis of Rummel’s Work on Democide,” Journal of Peace Research (2004) — methodological critique.
- R.J. Rummel, response to Dulić (2004).
Editorial note: This article distinguishes Rummel’s own estimates and conclusions from later scholarly criticism. Numerical claims are attributed rather than presented as settled historical totals.