Book Review: The Diffusion of Military Power by Michael C. Horowitz



In his book ‘The Diffusion of Military Power” Michael C. Horowitz presents an adoption-capacity theory that seeks to explain when and how states successfully adopt military technology and the adopted technology’s effect on international politics. The dependent variable is diffusion or whether a state adopts a military innovation or not. The two independent variables are the financial and organizational capital required for a state to successfully implement a major military innovation (MMI).

Horowitz defines diffusion as “the process by which an innovation is communicated through certain channels over time among the members of a social system.”[1] Innovations often diffuse in the shape of an S curve. First, only risk-acceptant actors adopt; then the majority picks up the new innovation; lastly, late-adopters implement the innovation.[2] It is worth noting that it is not a theory of preferences or interests. It does not explain why some states are more willing to adopt innovations than others. Rather, it attempts to explain what determines the way these innovations spread.

MMIs are defined as “major changes in the conduct of warfare, relevant to leading military organizations, designed to increase the efficiency with which capabilities are converted to power.”[3] An MMI has a “demonstration point” which is the first time “the potential of its full capabilities is reasonably known in the international system through an action by a first mover.”[4] For the nuclear bomb, the demonstration point happened when the U.S. Air Force dropped Little Boy over Hiroshima. Following the demonstration point, other states will make an internal and an external response. An internal response either entails an attempt to adopt the innovation (i.e. the technology and the organizational practice) or an attempt to counter the innovation. An external response entails either balancing, bandwagoning, or becoming neutral.

Financial capital is defined as “the particular resource mobilization requirements involved in attempting to adopt a major military innovation.”[5] The cost per unit of the technology is the main driver of financial intensity. Furthermore, if the innovation is not exclusively military but can be used in civilian settings as well it is said to require less financial capital than purely military innovations.[6] The higher the financial intensity, the slower the spread of the innovation and the lower the probability is that a state will attempt to adopt the innovation.[7]

Horowitz does not explicitly define military organizational capital, but he posits that organizational capital for militaries “represents a virtual stockpile of change assets needed to respond to changes in the character of warfare. […] It is the nontechnological aspect of force generation demonstrated through doctrine, education, and training. ”[8] The higher the organizational capital required, the slower the spread of the innovation and the lower the probability a state will attempt to adopt the innovation.[9] There are three determinants of organizational capital: Critical task focus, experimentation resources, and organizational age. Critical task focus is the specificity of the organization’s goals.[10] If an organization has a broad focus, it is more likely to adopt an innovation. One example of difference in critical task focus is between the U.S. Army and U.S. Marines where the latter defines its objectives in a wide sense.[11] Horowitz uses internal documentation to code for the critical task focus of militaries. The experimentation variable is the organization’s commitment to experimenting with disruptive innovations. The more resources allocated towards experimentation, the higher the likelihood of successful adoption. Horowitz uses budget data to code for the experimentation variable. The organizational age variable is not explicitly defined, but the older the organization is, the more rigid and less likely to adopt it becomes.[12] Horowitz uses the time gap since a regime change or a major war that the state lost to code for organizational age.[13] The organizational capital required by an innovation is coded using the organizational capital of the first-mover.[14] The higher the financial intensity, the higher slower the spread of the innovation and the lower the probability a state will attempt to adopt the innovation.[15]

Horowitz then brings his theory from the system-level to the state-level in two parts. First, in order to predict diffusion at the state-level, one must also take into account the state’s utility function, the informational about the innovation which the state has acquired, the strategic environment of the states (its geography and threats), and domestic politics. One might view these as control variables. Second, whether a state will successfully adopt or merely attempt to adopt depends on whether the state’s financial and organizational capital exceed the required financial and organizational capital of the innovation.[16]

The second half of the adoption-capacity theory explains the impact of military diffusion on the international system through shifts in relative power. Diffusion of military innovations affects the international security environment in two ways: “through the implications of the system-level distribution of responses, and from first-mover advantages.”[17] MMIs can require either high or low financial and organizational capital, yielding four scenarios. In short, financially intense innovations reinforce the current power balance whereas less financially intense innovations threaten existing powers. Organizationally demanding innovations are likely to provide first-mover advantages of longer duration compared to less organizationally demanding innovations.[18]

Diffusion of MMIs has a specific effect on war through three intermediate variables: geography, speed, and intensity. MMIs can allow states to engage in disputes in a wider geography range. MMIs can also increase the speed by which a state can respond to threats. Lastly, MMIs that are diffused slowly can cause power asymmetries which can shorten then conflict—and rapidly diffused innovations are likely to be more intense as both sides can utilize the new technology to fight each other.

Before moving on to case studies, Horowitz answers potential objections or alternative explanations to the diffusion of MMIs: strategic competition, international and domestic norms, and culture. In the proceeding chapters, he presents four cases that maximize variation in financial and organizational capital, covering aircraft carriers (high-high), nuclear weapons (high-low), battleships (medium-medium), and suicide terrorism (low-high).

Supply-side explanations

Horowitz presents a supply-side theory for the diffusion of military power. Regardless of their content, supply-side theories often face external criticism from proponents of demand-side, cultural, and normative arguments. Horowitz attempts to answer these three criticisms but does not provide any compelling arguments as to why his theory—or supply-side theories in general—are better than explaining diffusion of military innovations than these alternative theories. For example, when anticipating criticism from demand-side proponents (which he refers to as ”strategic competition,” and it should not be confused with Debs and Monteiro’s strategic-logic arguments which combine supply- and demand-side explanations. To avoid confusion, I apply Debs and Monteiro’s terminology), he presents three counterarguments.

First, demand-side explanations fail to account for the role of financial and organizational capacity.[19] He conflates good explanations with his own explanation. This is equivalent to arguing that demand-side arguments are limited by the fact that they do not include his supply-side arguments.

Second, demand-side explanations might be able to predict which states attempt to adopt an innovation, but they fail to predict which states succeed in adopting the innovation.[20] A demand-side proponent would disagree, however, and argue that states that really want a military innovation will get it. North Korea succeeded in acquiring nuclear weapons despite having one of the poorest economies in the world. If states attempt to adopt but does not fully implement a military innovation, it must be because the underlying security environment has changed. South Korea had a nuclear program which it abandoned once the United States extended its nuclear umbrella.[21] As such, Horowitz’s second-counterargument does not settle any intellectual disputes but carves out a deeper trench for supply-side theory to fight from.

Third, demand-side explanations are often tautological because threats are difficult to observe directly, and adoption is often a very good proxy for threat level.[22] This counterargument assumes that the supply- and demand-side debate is zero-sum and that by weakening alternative arguments he strengthens his own. Furthermore, this counterargument seems slightly odd given that his own theory struggles with tautological explanations (which I will show later).

Horowitz presents similar counterarguments to the cultural and normative arguments. Admittedly, his argument is not that supply-side explanations are sufficient to account for most variation in diffusion of military innovations but only that they are necessary.[23] However, while Horowitz’s theory may or may not be a good supply-side theory, it is un clear how necessary or important supply-side theories are in explaining the diffusion of military innovations compared to alternative theories.


Moving on from supply-side theories in general to Horowitz’s adoption-capacity theory, we investigate the definitions of the variables and find two problems. The first problem concerns the diffusion variable. The second problem concerns the organizational-capacity variable.

There exist two issues related to the diffusion, i.e. the dependent, variable. The first issue is that Horowitz creates two false dichotomies by distinguishing between major and non-major military innovations and between adoption and non-adoption. One example appears in his discussion of the aircraft carrier. Throughout the discussion, the level at which adoption of the aircraft carrier can be rendered successful is extraordinarily high. Successful adoption, according to Horowitz, seems to require not only a fully operational aircraft carrier but also an aircraft carrier group comprised of a supercarrier[24], three frigates, and a submarine, and anything that falls below this threshold is to be considered unsuccessful adaptation.[25][26] However, one might argue that the slightly smaller but still the French CATOBAR-type[27] Charles de Gaulle Class carriers also exemplify a successful adaptation.  Similarly, the British Queen Elizabeth Class is of a STOBAR-type[28], the British HMS Ocean helicopter carrier, and the American San Antonio Class amphibious transport docks are all examples of well-implemented supercarrier- or carrier-like innovations. Furthermore, aircraft carriers have evolved over time which makes it difficult to identify a single demonstration point or even to consider aircraft carriers as one innovation. Each country adapts demonstrated versions of military innovations to its own needs. As such, it is not obvious that the United States was the only successful adopter of the aircraft.  The issue could be solved either by making successful adoption a continuous variable, in which France would be almost successful, or by looking at military innovations more broadly instead of focusing on Major Military Innovations only. As such, he current theory underexplains the diffusion of military innovations.

The second issue related to the diffusion variable is that it does not distinguish among the quantity of an adopted innovation. One could argue that the theory only explains whether states adopt a certain MMI—not the quantity. First, the theory fails to explain why France chose to develop 10 aircraft carriers that were slightly smaller than the American Nimitz-class instead of building 5 supercarriers. The dependent variable for France would be coded differently while its financial and organizational capacity would be coded the same. As such, this issue allows for unexplainable oscillations in the dependent variable. Second, by ignoring quantity, the theory fails to account for much of the variation in balance of power. The Swiss Guard and the Chinese Army both possess rifles, yet, this fact says little about their relative power.

There exist two issues related to the organizational-capacity variable, as well. The organizational variable is weakly defined and difficult to observe. Horowitz never defines the organizational variable, but he compares it to the organizational capacity of a company which is measured by subtracting the value of the physical assets of the firm from the total stock value.[29] In the business example, organizational capacity is a residual that captures anything that the financial capacity variable does not capture. In order to avoid this, Horowitz divides the variable into three subcategories or proxies. However, these subcategories are either very vaguely or narrowly defined—for instance, the organizational capacity variable does not capture the education level of the workforce which is arguably an important determinant of a state’s ability to successfully adopt an innovation—and fail to explain much of the variation in militaries’ organizational capacities.

It is unclear how organizational age directly leads to lower organizational capacity. One can see how critical task focus and experimentation can directly allow militaries to adopt MMIs more easily, however, organizational age does not seem to directly determine anything. Horowitz argues that older organizations tend to be more rigid and less capable of adopting innovations.[30] Yet, the reverse could also be true in that old organizations have been able to survive because they have been successful at adapting to innovations. Even if we stick to Horowitz’s assumption that organizational age has an inverse relationship with organizational capital, one could argue that organizational age does not have to describe anything itself because it serves as a proxy variable for organizational capacity. Yet, we do not know which exact mechanisms organizational age should account for because organizational capacity has not been properly defined.

Overall, organizational capacity is both undefined and, consequently, difficult to measure. This yields the potential for drawing tautological conclusions: States possess the organizational capacity to adopt an innovation because we observe that it succeeds in adopting it.

Causal relationships

If we move beyond the problems concerning the definitions of the variables and consider the logical arguments in the theory, we find four issues that limit the theory’s ability to account for variation in adoption.

The first is is that the theory fails to account for variation in military expenditures. States typically spend between 1% and 10% of GDP on their militaries. If a state decuples its military expenditure, it can arguably acquire more financially intense military innovations. When do states decide not to acquire innovations and reduce military expenditure? What is the maximum amount of financial capital that a state can allocate to acquiring and adopting military innovations? The theory does not answer these questions. And the questions are important because they could potentially provide the source of tenfold variation in the diffusion of military innovations. Demand-side proponents would argue that they could describe these variations in military expenditure through variations in threats which cause some states to allocate more financial capital to the acquisition of innovations than economically like states facing fewer and weaker imminent threats. Proponents of normative arguments would argue, for example, that Japan has not acquired nuclear weapons because of an international norm that prevents otherwise capable states of acquiring nuclear weapons. Proponents of cultural arguments would argue that Germany spends less on military than what supply-side theories predict because World War II sparked a culture of pacifism among the German people. Proponents of Horowitz’s theory have no arguments to present on this matter.

A second limitation is that this theory does not account for variation across different types of military innovations.  A Gotland-class submarine costs approximately $100 million per unit, has a crew of 25, and although it requires some training in order to operate successfully, the submarine can operate fairly independently of other naval and military units.[31][32] Both the financial and organizational capital required is medium. In comparison, a Lockheed-Martin F18 fighter jet costs $85 million per unit, needs a crew of approximately 20 per jet on average including mechanics, logisticians, and ground personnel, and even though it requires a fair amount of coordinated training with other units to be fully operational, some of this information is already available through experience from the civilian aviation industry.[33] The financial and organizational capital required is also medium. Both innovations are among the most sophisticated submarines and fighter jets, respectively. According to the theory, these two innovations should be diffused at approximately the same rate, yet, while most states operate a fighter jets, far fewer states operate submarines. The theory does not account for this type of variation.

A third limitation is the states’ lack of foresight. In the theory, it is the financial and organizational intensity of the military innovation that determines whether states attempt to adopt or not. When a state has adopted an innovation, it is then the financial organizational capacity of the state that determines whether or not the state will adopt successfully.[34] The theory then assumes that states have no ability to take their own financial and organizational limitations into consideration when deciding to adopt a military innovation or to foresee financial and organizational obstacles in the adoption process. Thus, every state should attempt to adopt new military innovations, but this is clearly not the case. Horowitz claims that the theory explains why some states attempt but fail to adopt.[35] However, in doing so, it becomes impossible to explain why some states choose to attempt to adopt and others choose not to attempt to adopt.

The Strategic Logic of Nuclear Proliferation

Horowitz’s adoption theory bears similarities and differences with Debs and Monteiro’s ‘Strategic Logic of Nuclear Proliferation.’ They both seek to describe the diffusion of military innovation, and they both suffer from issues with definitions, measurement, and shallow causal relations. In Debs and Monteiro’s case, the alliances seem to explain most cases of nuclear proliferation and nonproliferation, however, the question of why alliances want to allow or prevent nuclear proliferation remains unanswered.[36] In Horowitz’s case, organizational capacity seems to explain much of the variation in diffusion; however, it is not entirely clear what organizational capacity entails.

Continuing the previous discussion of supply- versus demand-side explanations, the comparison to the ‘Strategic Logic of Proliferation’ reveals the limitations of supply-side theory versus strategic-logic theory, which captures both supply and demand. Debs and Monteiro’s theory offers a complete, rationalist framework for understanding nuclear proliferation that allows for other theories to account for demand-side and supply-side variation and explore micro-level questions on each side. Horowitz’s theory is incomplete because both supply and demand affect states’ decision and ability to adopt military innovations. Instead, he builds upon demand-side explanations and, independently of those, presents his own supply-side theory.

Here, it is important to notice that the two theories attempt to describe different observations: Horowitz focuses on military innovations in a broad sense and tries to capture both the diffusion of the technology as well as the ability to successfully incorporate the technology into the military organization and apply it for maximum military power. Explaining both technology and the application of technology is important because technology itself often only partially explains variation in military power—the ability to utilize the technology is also highly important.

This is not the case for nuclear weapons, though. Having one nuclear weapon can significantly alter the balance of power, and it requires comparatively little skill to utilize a nuclear device. This is why, when focusing on nuclear weapons, Debs and Monteiro are able to include demand-side explanations without presenting a more complex theory than the adoption-capacity theory. Horowitz, on the other hand, applies a broader scope and by considering all types of military innovations, he has to describe militaries’ ability to utilize that technology which forces him to look at supply-side explanations only. Eventually, Debs and Monteiro and Horowitz all try to explain the diffusion of military innovations’ impact on the balance of power, but they apply different scopes which, respectively, allow and force them to alter the specificity of their independent variables.


This book review has argued that Horowitz’s adoption theory faces multiple challenges. Some of these challenges are common to most supply-side theories as they face competition from alternative explanations. Other challenges stem from vague definitions and immeasurable variables. And other challenges relate to the underlying causal logic of the adoption theory. This book review could also have looked at the empirical cases presented in the book, but in an attempt to hit the most birds with one stone—and to avoid boiling the ocean—I have chosen to focus exclusively on the theory. Even if Horowitz only considers supply-side explanations, the scope—major military innovations—is tremendously wide, and this makes it difficult to develop one coherent theory.



[1] Horowitz, Michael C.: “The Diffusion of Military Power”, Princeton University Press (2010), pp. 19

[2] Horowitz, pp. 19

[3] Horowitz, pp. 22

[4] Horowitz, pp. 24

[5] Horowitz, pp. 31

[6] Ibid.

[7] Horowitz, pp. 32

[8] Horowitz, pp. 31, 32

[9] Horowitz, pp. 39

[10] Horowitz, pp. 35

[11] Horowitz, pp. 36

[12] Horowitz, pp. 37

[13] Horowitz, pp. 38

[14] Horowitz, pp. 39

[15] Horowitz, pp. 39

[16] Horowitz, pp. 41

[17] Horowitz, pp. 45

[18] Horowitz, pp. 49

[19] Horowitz, pp. 56

[20] Ibid.

[21] Monteiro, Nuno P. & Debs, Alexandre: “The Strategic Logic of Nuclear Proliferation”, International Security, Vol. 39, No. 2 (2015), pp. 41

[22] Horowitz, pp. 56

[23] Horowitz, pp. 53

[24] The only supercarriers are the American Nimitz Class and the American Gerald R. Ford Class. U.S. Navy, “Technical Details.” (accessed November 22, 2017)

[25] The typical constellation of a U.S. aircraft carrier group

[26] Horowitz, pp. 89, 92, 94

[27] Catapult Assisted Take-Off, Barrier Assisted Recovery carriers are the largest and most complex carriers. Bender, Jeremy, “These graphics show the differences between the world’s 3 types of aircraft carrier.” (accessed November 22, 2017)

[28] Short Take-Off, Barrier Assisted Recovery carriers are technologically simpler and thus easier to operate that CATOBAR carriers, but the aircraft must be lighter to successfully take off from the deck. Bender, Jeremy.

[29] Horowitz, pp. 33

[30] Horowitz, pp. 37

[31] Gotland Saab. ”Kockum Gotland Class.” (accessed November 22, 2017)

[32] SvD. “Norge suget pa svenska ubåtar.” (accessed November 22, 2017)

[33] Lockie, Alex. ”Boeing has an updated F-18 in the works—and there is how it’s ’comparable’ to the F-35.” (accessed November 22, 2017)

[34] Horowitz, pp. 41

[35] Ibid.

[36] See my second reading response