The Not-So-Secret Ingredients of Military Coups

July 27, 2017

For special access to experts and other members of the national security community, check out the new War on the Rocks membership.

Naunihal Singh, Seizing Power: The Strategic Logic of Military Coups (Baltimore: Johns Hopkins University Press, 2014).

In “Parkinson Law,” Northcote Parkinson’s witty essay on the rules of bureaucracy, there is an interesting section on how to sway parliamentary votes. Just surround the hesitant politician with your own partisans, Parkinson suggests, and convince him the die is already cast; that everyone is going to vote your way. Regardless of his views, he’ll be quick to jump on the winning wagon.

This is, at a glance, the most important insight of Naunihal Singh’s fascinating work on coups d’état. It is an original, well-written book on the dynamics of military revolts, a relatively neglected question in the recent decades. And as the last word on such an important subject, Seizing Power is well worth a review even today, three years after its initial publication, especially since coups, both successful and not, have continued to unfold in Burkina Faso, Burundi, and Turkey.

The goal of Seizing Power is to decipher the mechanism behind the success and failure of military attempts to overthrow an incumbent government. The literature on coups, writes Singh, tends to focus on their origins and results, but not on their dynamics. Scholars typically do not bother to study failed coups. This is a pity, because the success or failure of coups is a game changer in many historical situations, and the process that leads to either result merits serious attention. Singh rejects the two prevailing approaches, that coups are similar to either battles or violent “elections,” choosing instead to describe them as games of uncertainty where people tend to choose the side they believe most other actors support.

The battle approach, most associated with the seminal study of Edward Luttwak (Coup D’état – A Practical Handbook), looks mainly at tactics and the comparative strength of rebels and loyalists. Luttwak emphasizes secrecy, speed, intelligence, accuracy, and tactical acumen, assuming that the more competent side is likely to win the fight. Not exactly so, argues Singh, because unlike in a real war, belligerents in a coup d’état belong to the same national entity, and both would like to rule the country, not ruin it with a costly civil war. Therefore, they are often reluctant to use excessive force. When the tide turns against them, they usually surrender or flee rather than fight to the death.

The second approach, viewing the coup as a violent form of election, focuses on the ability of both belligerents to convince as many political actors as possible that they are right and their rivals are wrong. A coup d’état, according to this view, is first and foremost a political struggle, wherein both belligerents try to draw the greatest support among the undecided. When I researched The Plots Against Hitler, my book on the German military resistance during the Third Reich, I heard a neat summary of this approach from a prominent historian of Nazi Germany. “July 20, 1944,” he said, referring to the most famous attempt to kill Hitler and overthrow his regime by means of a coup d’état, “was the last ballot in Nazi Germany, held only among the generals. The leaders of the army, each and every one of them, had to choose between Hitler and the conspirators, and almost all of them voted Adolf.” Singh is deeply critical toward this approach as well. His own extensive research in Ghana included numerous interviews with officers involved in several failed and successful coups, on either the side of the rebels or the government. And most of them told him that their own political beliefs had little to do with their eventual decision of whom to support. Instead, their self-interest and their responsibility to their soldiers forced them to choose the side which was more likely to win.

With the help of an extensive, original database comprised of all known coup d’état attempts from 1950 to 2000, Singh examines the hypotheses of both “coups as battle” and “coups as elections,” and finds both of them wanting. Instead, he offers a third, alternative explanation as to why coups succeed or fail, the so-called “coordination theory.” 

The Coordination Theory

This theory was not invented by Singh, only adopted for his analysis. Originally part of game theory, it stipulates that people are likely to choose a certain course of action if they believe it’s an investment in a winning cause. “Believe” is a key word here. Nobody can foretell defeat or victory in real-life situations, giving rise to troubling uncertainty. Actors may choose the side they personally don’t like, even the side which is actually on the verge of defeat, as long as they believe, rightly or falsely, that most of their peers are likely to do the same.

That makes manipulation of information extremely important during a coup d’état. Singh argues, for example, that during the failed 1991 coup in the Soviet Union, the rebelling junta had much more power than Yeltsin’s loyalists, but that didn’t matter. Through an extensive campaign of misinformation, Yeltsin and his paladins convinced the public, the army, and the political system that the rebels were decrepit, elderly generals without support, and that officers and soldiers were leaving them in great numbers. This undermined the self-confidence of the junta, plaguing it with hesitation. Almost certain their troops would not obey them, the members of the junta canceled their attack on the parliament at the last moment. The coup crumbled shortly afterward.

By means of this theoretical and quantitative analysis, as well as in-depth qualitative studies of seven test cases in Ghana and one in the Soviet Union, Singh reaches several important conclusions about coups:

  1. In military coups, ideology doesn’t matter. Actors don’t join the side they necessarily agree with, but the side they think most other actors are likely to support. Therefore, highly unpopular rebels may succeed, while popular rebels may fail.
  2. It is mandatory for the rebels to take over the main TV and radio stations, in order to create a fait accomplice. That, by itself, is a very old insight. It is mentioned in almost all famous essays on coups. Singh’s originality is in arguing that after taking over the stations, the content of the speech delivered over the airwaves is of paramount importance. The conspirators do not need to waste time rumbling about their righteousness. Instead, they have to hammer home the idea that they have already won, that most important actors have already pledged their support for the coup. Even if untrue, such a broadcast may create a self-fulfilling prophecy (p.29).
  3. Coups led by senior generals (contrary to those led by middling or junior officers) are more likely to succeed, because the leaders of the army have a greater ability to use soft power, connections and influence in order to control the flow of information, cementing the belief the coup is already a done deal. Coups can also be led by middling and low officers, but such coups are harder and more likely to deteriorate into bloodshed and civil war. The chance of the coup succeeding, argues Singh, grows proportionally with the rank of its ringleaders.
  4. Civilians don’t really matter during a coup, unless they can exercise strong influence on officers.
  5. Street demonstrations and mass civil action are not likely to stop a successful coup.

The Problems with Singh’s Argument

And yet, there are also weak points in Seizing Power. Taking them into account, it seems that notwithstanding the usefulness and general validity of Singh’s coordination theory, some of his conclusions merit at least some revision.

The problem begins with Singh’s methodology. I do not intend to criticize his quantitative analysis, but the chronological boundaries of his sample are arbitrary at best. Why 1950 to 2000? Why not to begin in 1940, 1930 or 1920? In fact, Singh’s neglect of pre-World War II coups blinds him to some major test cases — such as interwar Eastern and central European coups, Chinese coups during the Warlord and the Nanjing Periods (1915 to 1937), as well as the February 1936 Incident in Japan and two important coups in Germany — the Kapp Putsch of 1920 and the July 1944 Plot. This is not a quibble. 1950, the earliest boundary of Singh’s sample, marks the beginning of decolonization, the Cold War and the consolidation of the postwar order in Europe. Therefore, there is a disproportionate number of African and Latin American coups in the sample, and not enough European, Chinese, and Japanese ones.

Singh compounds the problem by his biased choice of cases for qualitative study. Of the eight cases assessed, seven are from one African country, Ghana, and the remaining case is from the Soviet Union. This choice colors the entire study in African (and specifically Ghanaian) hues, and makes one wonder whether Singh’s conclusions reflect African conditions rather than the general dynamics of a global phenomenon.

As important is Singh’s dismissal of ideology and civilian elites as major factors in military coups d’état. A thorough examination of the two coups I have studied, the July 1944 plot in Nazi Germany and the incident of February 26, 1936 in Japan, suggest ideology and civilian elites cannot be so easily ignored.

Nazi Germany and the Role of Ideology

In certain ways, the abortive coup of July 20, 1944 in Nazi Germany gives some support to Singh’s coordination theory. In the early afternoon of that day, when it was still uncertain whether Hitler was dead or alive, many officers were influenced by the conspirators’ assurances that the entire army had joined them. However, when the conspirators failed to take over the radio stations, it was clear to every officer that the coup was not a fait accomplice, and that the bulk of the army opposed it. In the late afternoon, after Propaganda Minister Goebbels announced by radio that Hitler was alive, even erstwhile supporters of the conspiracy began to desert.

However, the decision-making of all actors on that day was influenced by certain basic assumptions, all of which were related to Nazi Germany’s ideological framework. For example, both the conspirators and their enemies concurred that Hitler’s death was a necessary condition for the coup, regardless of the deteriorating situation on the fronts. Not matter how shameful Hitler’s military failures had been, it was extremely difficult to challenge his leadership. Wehrmacht soldiers, both sides agreed, were immersed in Nazi ideology and admiration for the Fuhrer. They would not follow the senior officers as long as Hitler was alive. Therefore, when it became clear the assassination attempt had failed, the conspiracy crumbled quickly, even though the conspirators continued to assure their comrades they had universal support in the army. With Hitler alive, it just didn’t sound credible. Maybe the conspirators could have isolated Hitler from the world long enough to overthrow the regime had they succeeded in taking over the radio stations, but that would have been extremely difficult. In other words, the preponderance of Nazi ideology in the Wehrmacht was a preexisting condition that all actors had to consider. Accordingly, it had a serious influence on the decision-making of everyone in the system. For coup participants in Nazi Germany, ideology was important.

Imperial Japan and the Role of the Emperor

In the 1930s, coups were very common in Imperial Japan. There were at least four serious attempts between 1930 and 1936. The last of these — the February 1936 Incident — was a large-scale insurgency of more than a thousand soldiers led by a group of junior officers.

Just like in the Nazi case, these coups give some support to Singh’s model. The two attempts in 1930 were plotted by the Cherry Blossom Society (Sakura-kai), a cabal of mid-ranking officers. In both cases, they backed off when it became clear that senior generals had either withdrawn support or were about to offer effective resistance. That sits well with Singh’s argument on the difficulties of coups led by middling officers. Even such a bloodthirsty group as the Sakura-kai, which was planning to massacre the entire political leadership of Japan with naval bombers and poisonous gas, gave up when it became clear most military actors had turned against the plot.

This tendency was even more pronounced during the incident of February 1936. The ringleaders, mostly lieutenants and captains, opened the coup with assassinations of political and military leaders and occupied large parts of central Tokyo. However, after the first wave of violence subsided, they became timid, almost paralyzed, whenever they encountered senior officers. When it became clear the coup had entered a phase of stagnation, the senior allies of the young officers (who were always ambivalent) withdrew their support. During the third and fourth day of the coup, the army as a whole turned against the plotters, leading to the desertion of the soldiers under rebel command. In accordance with Singh’s model, both sides were wary of a civil war and the plotters did not fight to the finish, even though it was probable that they would face the firing squad.

However, yet again, the coup took place in a certain historical framework, strongly intertwined with ingrained ideological presumptions. In Imperial Japan, the emperor reigned, but his office was divorced from the day to day affairs of government. The sovereign usually refrained from intervening, and yet, when he said something, his word was final. Even military rebels recognized his authority as divine, and only hoped to replace his “traitorous advisors” with “patriotic subjects” (namely, themselves and their supporters). On the first day of the February Incident, it seemed the senior leadership of the army was divided between proponents and opponents of the coup. In the first few hours, the proponents were better placed and had more power than the opponents. There were some signs undecided officers had started to believe the rebels would win, and considered joining their camp.

The single event that changed that dynamic was the explicit intervention of the emperor against the rebels. When the emperor heard about the coup, he explicitly ordered it to be put down. He told his generals that he would fight the coup in person at the head of the Imperial Guard if they didn’t act. In a country where the emperor was considered divine, his explicit orders could not be resisted, not even by enthusiastic supporters of the coup. From the moment the emperor decided to use his personal influence against the coup, the fate of the plotters was sealed. It took three days to defeat them, only because the senior generals wanted to avoid a violent showdown in central Tokyo. But when the imperial wish became clear, nobody that mattered could support the rebels any longer. The emperor, and the civilian courtiers around him, had decisive influence because of the prevailing ideology. Hence, this test case sheds serious doubt on two of Singh’s key findings: that ideology doesn’t matter for the success and failure of coups, and that civilians don’t have serious influence on their dynamics.

Turkey and the Failed TV Broadcast

It is impossible to not include last summer’s failed coup d’état in Turkey in our discussion, which took place two years after Singh’s book was published. The rebels in Turkey did just what Singh would have recommended: They occupied a major TV station and used its facilities to broadcast to the nation. They, too, tried to create an image that their coup was a fait accomplice. However, it didn’t help them. As they failed to kill President Recep Erdogan, he was able to communicate with the nation through an aide’s iPhone. Civilian resistance played a large role in the defeat of the rebels, as angry crowds surrounded their tanks and all major opposition parties denounced them. In a modern country, with numerous TV, radio, social media, and internet channels, it is extremely difficult for rebels to create a fait accomplice through a single TV broadcast. Singh’s emphasis on the importance of the broadcast seems therefore a little dated. His insistence that street demonstrations don’t matter should be reexamined as well.

The Future of Coups

The analysis above may give us some clues about future coups d’état, and the ways in which incumbent governments can cope with them. Singh rightly argues that coordination theory can be a guideline for both belligerents in a coup: Each of them has to prove as early as possible that the struggle is already decided in his favor. Yet, contrary to Singh’s argument, the way to do so cannot be a monopolization of all or even most channels of information. As the Turkish case has shown us, this is next to impossible in a modern country, even if we ignore the role of social media and the influence of foreign TV stations. Instead, both sides will have to play the coordination game, taking into account that their rivals could not be silenced.

What should they do? The key is in Singh’s own analysis of the failed coup against Gorbachev in 1991. In that coup, no side could silence the other. However, Yeltsin’s side was more adept at media manipulation, rather than outright coercion. In contemporary coups, the winners will probably have to win more allies in the media already in their early stages of planning. In addition, they’ll be able to skillfully frame their own news, often by using proxies at home and abroad, while creating the impression that the news of the other side is faked. In that sense, coups are much more similar to real election campaigns than Singh would like us to believe.


Danny Orbach, The Hebrew University of Jerusalem, author of The Plots against Hitler (New York: Houghton Mifflin Harcourt, 2016) and Curse on this Country: The Rebellious Army of Imperial Japan (Ithaca: Cornell University Press, 2017).

Image: David Broad, CC

We have retired our comments section, but if you want to talk to other members of the natsec community about War on the Rocks articles, the War Hall is the place for you. Check out our membership at!