HomeDocuments on SitePersonalDemocratic PeaceDemocide20th C. DemocideMegamurderersLesser MurderersWhy DemocideDimensionsConflictMethodsTheoryPolicyLinks PHOTOS OF DEMOCIDEGalleries

Review of Power Kills

Ted Galen Carpenter's Review

Ted Galen Carpenter's rebuttal to Rummel's reply

H-C. Peterson's response to Carpenter' review and rebuttal.

Other Related Democratic Peace Documents On This Site


What is the "democratic peace"?

"Waging denuclearization and social justice through democracy"

"The rule of law: towards eliminating war"

"Freedom of the press--A Way to Global Peace"

"Convocation Speech"

Freeman Interview

City Times Interview


Bibliography on Democracy and War

Q & A On Democracies Not Making War on Each Other

But What About...?

"The democratic peace: a new idea?"


"Libertarianism and International Violence"

"Libertarianism, Violence Within States, and the Polarity Principle"

"Libertarian Propositions on Violence Within and Between Nations: A Test Against Published Research Results"

"Democracies ARE less warlike than other regimes"


Vol. 2: The Conflict Helix (see Chapter 35)

Vol. 4: War, Power, Peace (see e.g., Propositions 16.11 and 16.27

Statistics of Democide

The Miracle That Is Freedom


By R.J. Rummel

This is in response to Ted Galen Carpenter's negative review (The Independent Review, v.II, n.3, Winter 1998) of my book Power Kills: Democracy as a Method of Nonviolence (New Brunswick, N.J.: Transaction Publishers, 1997). In this book I bring to bear all the published systematic quantitative and historical evidence I could find and have generated on five propositions about the inverse relationship between democracy and collective violence. These propositions are that democracies

These propositions are overwhelmingly supported, and lead to a concluding and summary proposition:

But empirical results, no matter how sophisticated, are insufficient unless supported by a consistent theory. About half the book, therefore, presents alternative theories for explaining the nonviolent nature of democracy, and concludes that the best way of understanding this is in terms of the spontaneous society that freedom (liberal democracy) creates. This idea is similar to F.A. Hayek's spontaneous society, as described in his three volumes of Law, Legislation and Liberty, and is an idea well known to libertarians.

It is a free market writ large: of the economy, society, and politics within an overarching legal framework of civil liberties and political rights. A spontaneous society creates cross-pressures and an exchange culture in which negotiation, compromise, and tolerance reduce the tendency towards violence found in more hierarchically organized societies. Moreover, the natural bonds and linkages that develop between such societies (e.g., trade, social and cultural exchanges, treaties), and the perception that the other society is like one's own, favoring negotiation and compromise, reduces the possibility of violence between them.

Now for Carpenter's review.

(1) Rummel's "core thesis" is that democracies do not make war on each other (p. 437)

. No, the core thesis is that power kills, the title of the book. This thesis sums up all five of the book's propositions, while Carpenter mistakenly focused his review on only one of them.

I'm perplexed that he completely ignores what are the most important propositions and evidence of the book, that democracies are most internally peaceful and don't murder their own citizens. If the importance of each proposition is understood in terms of the number of people killed in the indicated violence (and the resulting or correlative misery), then far more people are killed in domestic collective violence than in international wars. For example, millions more people were killed in the Teiping Rebellion in China alone than died in battle in World War I and II. And in our century governments have murdered about four times those that have died in combat in all the domestic and foreign wars. Stalin alone is responsible for the murder of millions more than the combat deaths of both World Wars together. Therefore, even if Power Kills only concerned domestic violence and democide (genocide and mass murder), the results would be an incredible plus for freedom. For they say that promoting democratic freedom will eliminate or reduce to a minimum, by far, the largest number of deaths from collective violence.

(2) Overall, Rummel "fails to provide the necessary evidence" (p. 437)

I don't understand what Carpenter means by evidence. To me evidence for a general hypothesis, such as A does not do B, comprises all, or an appropriate sample of, the cases in which A does and does not do B over the relevant time period, the significance (possible randomness) of the cases, the historical context and understanding of the cases, and the findings on the same or similar hypotheses by other researchers and scholars. This is the evidence I brought to bear on the five hypotheses (propositions). My data or those of others I drew on cover all wars, going back to the ancient Greeks, in which democracies may have been involved; and for democide in this century, all democide and regimes. Moreover, my colleagues and I have subjected these data to both traditional and quantitative analyses. The overall result is that different investigators with different data collected under different definitions of democide, violence, war, and democracy, and applying different methodologies, verified the five propositions.

Perhaps Carpenter means the kind of evidence that would satisfy an historian. But I did elsewhere and referenced a number of historical and qualitative studies, as listed for criticism (6), below. I also referenced the historical analyses of others, as that of the historian Spencer Weart. In his book Never At War (New Haven: Yale University Press, 1998) he scoured written ancient and modern historical records to find a case in which democracies clearly made war on each other, including those proposed by Carpenter below, and found none.

No other general propositions of international relations and foreign policy have been so widely tested and thoroughly supported by empirical analysis. Indeed, as a result, some scholars are now asserting that the lack of war between democracies is an iron law of international relations, so well established that further research should focus on democratization. In fact one publisher's reader recommended the book not be published because there was nothing new in it.

(3) Rummel fails to consider alternative explanations or factors

Neither do I understand this criticism. Consider some of the alternative explanations and factors that I or those I cited tested for on one or another of the five propositions: geographic distance or size, small number of democracies, economic development, culture, power parity or lack thereof, ideology or religion in general and specific ideologies and religions in particular (e.g., communism, Islam, Christianity), war or revolution (possibly accounting for democide), population density, resources, education, technology (a factor Carpenter believes important), among others. I think that my colleagues and I have covered the most popular alternative explanations and factors, and even those that most analysts consider only remotely possible.

I have also covered alternative theories. These range from those trying to account for sociopolitical violence by social distances, in-group perception, cross-pressures, economic forces, and concomitant values, to political theories about the natural peacefulness of people, political bonds and interests, and the role of power.

And yet Carpenter writes that I am "oblivious to or casually dismissive of alternative explanations...." (p. 437)

(4) Rummel fails to consider cases of war between democracies

Those mentioned are the American Civil War, Boer War, and World War I. These and other possible cases have been carefully considered and dismissed by my colleagues, such as Bruce Russett in Grasping the Democratic Peace (Princeton: Princeton University Press, 1993), James Lee Ray in Democracy and International Conflict (Columbia: University of South Carolina Press, 1995), and the aforementioned Spencer Weart. Space does not allow for a consideration of all these possible exceptions, but let me focus on the war between Germany and the democracies that Carpenter writes "gives democratic-peace theorists intellectual indigestion" (p. 439). In my view this possible counter example is easily disposed of. Consider: the Chancellor of Germany served at the whim of the Kaiser, by whom he was appointed and dismissed. Moreover, the Kaiser had considerable power over foreign affairs, and the army was effectively independent from control by the democratically elected Reichstag. For all practical purposes, in foreign policy Germany was autocratic, without a democratic leash, and thus World War I hardly contradicts the proposition that democracies don't war on each other.

(5) Rummel does not consider cases where democracies almost went to war

Carpenter mentions several, and focuses on the 1898 Fashoda Incident, a war scare between France and Great Britain over control of the Nile River. But note this: negotiation between both sides was respectful and straightforward, neither side seeking to end up dominant; both sides thought they could count on the other side being reasonable; and indeed, in his book (Chapter 13) Weart quotes a French diplomat as saying that France assumed "England would never initiate hostilities." Rather than questioning the proposition that democracies don't make war on each other, this crisis supports it by illustrating why war crises do not escalate to war among democracies.

(6) Rummel robotically invokes statistical data (p. 438)

What? Should I not throw a wide net for statistical evidence? Or organize such evidence by proposition, dates, and methods? This is a strange criticism for a libertarian, since so many libertarian policy recommendations are based on economic statistical evidence. Anyway, the "robotically" presented systematic evidence prove that Power Kills.

The context for this criticism is Carpenter's discussion of such incidents as Fashoda and the claim that I should have engaged these cases. But a book can only cover in depth so much, and has to fill in by reference to other work. Carpenter ignores that on the most important democide proposition I also did extensive historical and qualitative analyses. In Death By Government I wrote case studies on each of fourteen cases in which a regime murdered at least 1,000,000 people, and also wrote separate histories of the Soviet democide (Lethal Politics, New Brunswick, N.J.: Transaction Publishers, 1990), Chinese democides (China's Bloody Century, New Brunswick, N.J.: Transaction Publishers, 1991), and that by Nazi Germany (Democide, New Brunswick, N.J.: Transaction Publishers, 1993). Moreover, I presented all the democide estimates, their sources and qualifications, and qualitative considerations underlying them, in my Statistics of Democide (Charlottesville: Center of National Security Law, Law School, University of Virginia, 1997). On the war propositions, I wrote five volumes of Understanding Conflict and War (Beverly Hills, CA: Sage Publications, 1975-1981). As should be clear, far from robotically invoking statistical data, I have immersed myself deeply in historical and qualitative analyses before coming to my conclusions.

(7) Rummel assumes that correlation means causation (p. 438-9)

This invocation of Statistics 101 misses the essence of the results. Fundamentally, the theory, described above, lead to hypotheses, that were then tested and retested and replicated by others. The theory dominates, the five hypotheses (propositions) flow from it, and these are successfully tested empirically, qualitatively, and historically.

(8) Rummel gives a "vague and slippery treatment" of democracy (p. 440)

This is yet another criticism I do not understand. I try to carefully define democracy, for without such a definition I could not collect data on democracy or understand the results. Moreover, in Chapter 8 ("What is to be Explained") of Power Kills I extensively detail the meaning and nature of democracy, carefully delineating it from authoritarian and totalitarian regimes, and placing the three types of regimes within the space of a political triangle that encompasses the variety of variation among these three types. Moreover, as noted in the book, this analysis is informed by several factor analyses of political variables I carried out in Statistics of Democide and elsewhere.

Carpenter is most concerned about my applying a contemporary definition of democracy to previous centuries. The problem here is the historical limitation on equal rights and the franchise, as before women achieved equal rights, or slavery was eliminated. For previous centuries the definition of democracy was loosened to include at least two-thirds of the males having equal rights (as long as the lower classes were not excluded), while maintaining the other characteristics, such as open competitive elections. For one, democracies so defined in previous centuries, such as the United States in 1800 and democratic classical Athens, saw themselves as democratic, called themselves democratic, and were perceived by other nations as democratic. Second, even with this looser definition, well established democracies so defined still did not make war on each other. Well established means that a regime had been democratic long enough for it to be stable and democratic practices to become established.

The fundamental question about any definition is: does it work? Does it define something in reality that predicts systematically to something else. If we have so defined an x such that it regularly predicts to y, and a theory explains this relationship, then that is a useful and important definition of x. Both the definition of contemporary democracies, and the limited one of those past, predict to a condition of continuous peace (nonwar) between these nations. If one does not agree that these are democracies, fine. Then call them xcracies. We then still can say that xcracies do not make war on each other, as given by theory, and by universalizing xcracies we have a solution to war.

(9) Rummel ignores conflicting evidence, such as that the United States carried out covert action against other nations, even democracies (p. 440)

Covert action is not war (military action) and is therefore irrelevant to the propositions. But Carpenter did catch me in a misstatement about a world of democracies eliminating the need for secret services. I had in mind covert violence against hostile nations, but the statement does not come through this way. There would be a place for spies in a democratic world, as there is in a free market where companies spy on each other. But in such a world hostile enemies would be absent, the expectation of war gone, and thus a secret war unnecessary.

But what about past American covert action? This was during the Cold War and was part of the largely successful policy to contain communism, particularly Soviet power. Mistakes were made, actions were taken that in hindsight embarrass democrats. Even then, there was no military action between democracies.

This having been said, there is also a deeper explanation. Democracies are not monolithic; they are divided into many agencies, some of which operate in secrecy and are really totalitarian subsystems connected only at the top to democratic processes. The military, especially in wartime, and the secret services, such as the CIA, are examples. These near isolated islands of power operate as democratic theory would assume. Outside of the democratic sunshine and processes, they do things that were they subject to democratic scrutiny would be forbidden. The answer to this problem is more sunshine and democratic control.

(10) Rummel ignores that the peace among democracies since World War II is due to a "powerful totalitarian threat" (p. 441)

The data are not limited to this period. As mentioned, other tests have been done for different years, including 1816-1960. Now it may be true that the Cold War accounted for the particular lack of war between democracies, but what about other periods? Also, ignore the statistics and consider Europe, the historical cauldron of war, and what has happened since the end of the Cold War. Unity has continued to grow, rather then hostility. And, incredibly, those old enemies, France and Germany, have even considered forming a common army. Moreover, once the former enemies became democratic, they have tried to join and are being integrated into a larger Europe.

(11) And finally Carpenter's conclusion: "Rummel makes an array of extraordinary claims, but he ultimately fails to prove his case" (p. 441)

But Carpenter ignores most of the book and concentrates on the proposition that democracies do not make war on each other. He titled his review "Democracy and War." It should have been "Power Kills." Moreover, even for the one proposition on which he did focus, his criticisms do not make sense, are incorrect, or are irrelevant as shown above.

I would have hoped that Carpenter, a Vice President of the Cato Institute, a libertarian think tank, would be overjoyed at what is proved about freedom. Not only does freedom promote the greater economic and social welfare and happiness, as libertarians believe, but it also promotes life and security. To the best of our knowledge, its universalization would end war and virtually eliminate other forms of collective violence, particularly the most pervasive and greatest cause of violent death--democide. Power has killed people by the hundreds of millions. Freedom would have saved nearly all of them and avoided the attendant suffering and misery. This is the miracle that is freedom.


* This is the prepublication draft of R.J. Rummel, "Democracy and War: Reply," (The Independent Review 3 [Summer 1998]: 103-108), which is a reply to Ted Galen Carpenter's "Review Essay: Democracy and War," (The Independent Review 2 (Winter 1998): 435-441) on Power Kills.

You are the visitor since 11/27/02

Go to top of document