Level III: International Structure as Material (The Problem of Anarchy)
This week’s move is to leave the actor behind and theorize the system. Waltz defines structure by three things, anarchy as the ordering principle, states as functionally “like units,” and the distribution of capabilities, and derives self-help and recurrent balancing from them. Axelrod and Keohane answer that institutions can lengthen the shadow of the future and make cooperation possible even under anarchy; Milner and Lake question whether “anarchy” is the right foundation at all; and Kertzer, Brutger and Quek take the security dilemma down to its microfoundations in mass opinion.
Reading list
- Waltz. 1979. Theory of International Politics, Ch. 5 & 6.
- Axelrod and Keohane. 1985. “Achieving Cooperation Under Anarchy: Strategies and Institutions.” World Politics.
- Milner. 1991. “The Assumption of Anarchy in International Relations Theory: A Critique.” Review of International Studies.
- Lake. 2009. Hierarchy in International Relations, Introduction & Ch. 2.
- Kertzer, Brutger, and Quek. 2024. “Perspective Taking and the Security Dilemma: Cross-National Experimental Evidence from China and the United States.” World Politics.
- Application: Glaser. 2024. “Fear Factor: How to Know You’re in a Security Dilemma.” Foreign Affairs.
Waltz (1979): Theory of International Politics
Ch. 5 & 6 · anarchy, like units, the distribution of capabilities, and balance-of-power theory
Many scholars and commentators label any alliance-building or military expansion meant to counter another state as “balance of power”, say, that China’s South China Sea policy and the United States’ Asia-Pacific alliances both “align with balance-of-power logic.” But that treats balance of power as a policy choice, a framework for describing phenomena, rather than the strict structural necessity Waltz envisions. Waltz’s key claim in Ch. 5 is tighter: in an anarchic international system, states are “like units” differentiated only by their material capabilities. The rest of his theory is built by walking carefully through each of those assumptions. Waltz treats states as the primary units because they are the major actors whose interactions form the structure of international politics; structures are defined not by all the actors within them but by the major ones, just as economists define markets in terms of firms. Non-state actors exist, but states “set the scene” in which everyone else acts. I’d push this further: many international organizations’ authority depends on the powerful states behind them (IMF conditionality backed by U.S. and EU preferences), and even large multinationals bend to strong states’ laws and sanctions, so often it is not “states yielding to non-state actors” but states responding to the powerful states backing those actors. What we gain is a clear, parsimonious definition of the system that lets us compare capabilities and explain system-level outcomes; what we lose is the detail of non-state actors, transnational movements, and the internal dynamics of states. States are “like units” in that they are all sovereign entities performing similar functions, defense, legislation, economic regulation, and differ not in function but in their capability to perform them. I read this as a claim about function rather than history, society, or culture: as long as the state, in our shared concept of it, exists to legally monopolize violence and provide order and protection from fear and mortal threat for a human community, the “function” of the state is the same. That is a logical judgment, not a factual one, historically the state may not have originated for this purpose, but in the now most widely accepted concept of the state it exists for it. As for anarchy versus hierarchy: a hierarchic system has centralized authority with differentiated functions and relations of super- and subordination; an anarchic system has no central authority, units are formally equal, and they coordinate through self-help. I agree with Waltz here, internationally there is no accepted superior authority that can legitimately and reliably punish rule-breakers.
In an anarchy of functionally undifferentiated units, what distinguishes states is how much capability they have, so Waltz uses the distribution of capabilities to define structure and to predict outcomes: when that distribution changes, expectations about behavior and outcomes change with it. Is capability the most important characteristic? Waltz thinks so for a structural theory; others would emphasize ideology, culture, or governance, but he deliberately abstracts these away to see the structure. I tend to agree with him, in practice “might matters.” Even a proposal for a new order requires the capability to supply the public goods that others will accept. In an anarchic self-help system we should expect states to prioritize survival, help themselves, balance against power, and emulate successful practices, while cooperation stays limited by security concerns, each unit spends part of its effort not on its own good but on protecting itself against others. Balance-of-power theory then predicts that balances will recurrently form whether or not states consciously aim at them, and whether some aim for balance or some aim for universal domination: states seeking at least to survive use internal means (arms, economy, strategy) and external means (alliances), and in a self-help system these generate balances through balancing or realignment. Importantly, it is a systemic theory, not a policy prescription. How would we test it? Look for balancing behavior even when states have reasons not to balance, the Franco-Russian alliance of 1894 is the kind of case where, given lopsided capabilities under anarchy, outcomes tilt back toward restoring balance. Waltz was writing in the bipolar Cold War, wanting to explain stability and conflict in a nuclear world of two superpowers. Classical realists like Morgenthau ground their theory in human nature, power hunger, and state motives; Waltz abstracts motives away and asks how the anarchic structure constrains behavior, modeling structure by (1) ordering principle, (2) no functional differentiation among states, and (3) the distribution of capabilities, explicitly keeping unit-level attributes out of the structural definition. He also distinguishes a theory of international politics from a theory of foreign policy: Allison’s Models II and III (organizational process, bureaucratic politics) work at the unit level, inside the state, whereas his own theory derives systemic outcomes, self-help, balancing, without explaining who inside a government chose a particular option. Offering the bureaucratic-politics approach as an alternative to the state-as-actor approach, Waltz says, is like saying a theory of the firm is an alternative to a theory of the market; I think the analogy is right. We still study him because his theory offers a powerful, parsimonious explanation of broad, recurring patterns, balancing, emulation, and remains the benchmark against which theoretical debate is run, even if it is less useful for explaining specific events or internal state dynamics.
Axelrod & Keohane (1985): Achieving Cooperation Under Anarchy
World Politics · payoff structures, the shadow of the future, and the institutionalist answer to Waltz
Harmony requires a complete identity of interests, whereas cooperation can only take place in situations that mix conflicting and complementary interests, actors adjusting their behavior to the actual or anticipated preferences of others. Cooperation is therefore not necessarily morally good. This matters because in world politics states rarely have fully aligned interests; most of the time they perceive conflicts, and pure harmony is very rare. The useful question is how cooperation is still possible even when interests clash, which is exactly what lets us study how to design conditions that encourage cooperation without harmony. The Prisoner’s Dilemma, Stag Hunt, and Chicken model different strategic interactions. In the Prisoner’s Dilemma players prefer mutual cooperation to mutual defection, yet each has a dominant strategy to defect (DC > CC > DD > CD), its fundamental problem is that defection is individually rational but leaves both worse off, mirroring arms races and trade wars. The Stag Hunt (CC > DC) shows cooperation works if both sides trust each other, but absent trust they settle for a less beneficial but safer outcome. Chicken (DC > CC, with DD worst) shows brinkmanship, where both risk disaster if neither backs down. Together they show cooperation is easier in some settings (Stag Hunt) than others (Deadlock), that beliefs and perceptions shape how actors read the game, and that repeated interaction with the expectation of punishment for defection helps overcome collective-action problems.
If states expect to interact repeatedly, cooperation becomes easier: the more future payoffs are valued relative to current ones, the less the incentive to defect today, because the other side is likely to retaliate tomorrow. This is the “shadow of the future”, caring about future consequences disciplines present behavior. Tit-for-tat operationalizes it: start by cooperating, then copy the other’s last move, cooperating if they cooperated and defecting if they defected. The necessary conditions for it to induce cooperation are that cooperation gives better results than mutual defection, that defection is tempting but is punished, and that interactions are repeated with a long shadow of the future. Axelrod and Keohane are more optimistic than Waltz: institutions can alter payoff structures, lengthen the shadow of the future, and break N-person games into games with fewer actors, so cooperation is possible even without a hegemon, through reciprocity, regimes, and adaptive strategies. Their critique of structural realism is that it underestimates institutions and overlooks how strategies and perceptions shape outcomes under anarchy. I’d add that this optimism is also tied to broader historical and social conditions: after World War II, growth, mobility, and innovation made cooperation and post-materialist values feel possible, but as societies grew more rigid, unequal, and economically slower, fear of downward mobility spread, competition and distrust rose, and publics turned more hostile to foreigners, making cooperation harder to sustain. I still agree that institutions matter, but they are not neutral: cooperation can be organized to exclude or target third parties, as when U.S. efforts to build the “first island chain” to contain mainland China show institutionalized cooperation intensifying conflict by being directed against a specific state. A structural realist would reply that institutions only reflect underlying power: cooperation happens when powerful states find it in their interest, and institutions are merely tools or arenas, not independent forces, consistent with Waltz’s claim that international institutions are weak.
Milner (1991): The Assumption of Anarchy: A Critique
Review of International Studies · two meanings of anarchy, interdependence, and the continuum
Milner separates anarchy as a lack of order, Hobbesian chaos, a war of all against all, often implied but contradicted by empirical work showing patterned order and regimes, from anarchy as a lack of government, meaning no common authority, which is the usage central to Waltz and to “Cooperation Under Anarchy.” The distinction matters because if anarchy meant “no order,” the neorealist findings of regular patterns like balancing and regimes would be paradoxical; but if it means “no common government,” order is fully compatible with anarchy, which cleans up a common confusion. It is important for Waltz precisely because his whole move is to derive systemic effects from anarchy plus the distribution of capabilities while still observing order. My own view is that the split between authority and coercion should not be treated as cleanly separable: in practice they co-produce each other, so the system lacks widely accepted legitimate coercion because it lacks widely accepted authority, and the two “definitions” often collapse empirically, though treating anarchy as a spectrum is still a useful methodological choice alongside the binary. What resonates: first, the sharp domestic-versus-international split is overstated, authority inside states varies along a continuum rather than a neat hierarchy, and the international side is not uniformly decentralized either, depending on issue-area and time. Second, Waltz’s three ordering principles pull against each other: there is real tension between “states as like units” and “capabilities shape functions,” which implies genuine differentiation and even hierarchy. Third, force is not what truly separates domestic from international politics, legitimacy and authority are, and they vary across levels and history, as with the Concert of Europe. Fourth, interdependence is a structural feature IR underplays: the system looks more like oligopolistic markets marked by strategic interdependence than a frictionless perfect market. What seems a bit unfair: on the fungibility of power, Milner says Waltz “curiously” assumes high fungibility to identify great powers, recentering power more than his market analogy suggests, a strong point, but partly interpretive, since Waltz sometimes treats fungibility as limited by issue-area. And her legitimacy-and-authority turn may conflate “lack of legitimate coercion” with “lack of authority”: she rightly warns that government and authority should not be conflated, but in practice the two co-produce each other. My nagging doubt is whether Milner is really critiquing what Waltz actually said, or a stricter caricature of it. Milner shows that domestic authority is often diffuse, U.S. checks and balances, federalism, and varies across countries along a centralization–decentralization continuum, so a rigid binary is misleading; she also questions whether the international realm is always highly decentralized, since issue-areas and periods differ, and she warns that over-emphasizing anarchy radically separates domestic and international politics and oversimplifies IR. I find treating hierarchy and anarchy as a continuum very helpful, since both inside states and among states the degree of centralization moves over time. But that does not make the binary useless: given strong nationalism and modern state capacity, many states can still mobilize nationwide and act with an internal solidarity that fosters cooperation in ways the international system cannot, so the binary remains analytically handy. I cannot accept Milner’s claim that the key to defining anarchy is a lack of authority rather than a lack of legitimate enforcement, these describe the same situation: there is no universally legitimate coercion in international society precisely because supranational bodies like the UN lack universally recognized, or even universally believed-in, authority. And “spectrum versus binary” are not really in conflict; they are a methodological adjustment for different analytical contexts, and rejecting the binary outright would make us overlook the power of nationalism and state capacity that lets even loosely centralized states mobilize and cohere.
By interdependence Milner means strategic interdependence: an actor’s best move depends on others’ moves and expectations, relations are mixed-motive (both conflict and cooperation), and power is built into asymmetric ties through vulnerability leverage. It is structural because the system resembles an oligopoly, a few consequential actors, dense mutual responsiveness, recurring balancing and coordination, with tacit communication and institutions needed to sustain cooperation or manage cheating. The implication is that outcomes are often indeterminate and unstable, shaped by expectations, information, and institutions, which resists overly deterministic structural claims; her bottom line is that interdependence is at least as fundamental as anarchy, and that focusing only on anarchy is over-reductionist. My view is that its weight varies by place and time: before the Russia–Ukraine war, major European states treated gas trade with Russia as normal commerce that could stabilize relations, but once the war began the geopolitical reading of energy dependence returned and the EU moved to cut it; inside the EU a shared space and some common identity deepened ties, yet uneven costs and benefits still produced contestation and even exit pressures. So interdependence sometimes constrains and sometimes amplifies conflict, which leaves me with an honest open question: states are interdependent, but what exactly does that help us explain? Milner warns that overemphasizing anarchy pushes scholars to over-separate domestic from international politics, to embrace an over-simplified view of world politics, to hide the ambiguity in the concept of anarchy itself (its two meanings), and to divert attention from interdependence and from the institutions and communication that shape outcomes. I think we should be alert but not alarmist: treating anarchy as a spectrum, degrees of authority and legitimacy, and of effective coercion, makes IR cumulatively progressive without discarding the binary when it is analytically useful. I also disagree with fully separating authority from legitimate coercion, since in practice they co-generate each other. So the fix is not to abandon “anarchy” but to use it alongside interdependence, legitimacy, and institutions, adapting the lens to the case.
Lake (2009): Hierarchy in International Relations
Introduction & Ch. 2 · authority, the divisibility of sovereignty, and informal empire
Lake defines international hierarchy as a relationship in which one state, the ruler, possesses authority over another, the ruled, authority being the right to command and expect compliance, grounded in a social contract where the dominant state provides political order in exchange for legitimacy and compliance from the subordinate. Hierarchy varies in degree by how many policy areas the dominant state can legitimately regulate. Anarchy, by contrast, is a condition where no such authority exists between states: states are functionally undifferentiated and operate in a self-help system with no superior power to enforce rules. The key difference is the authority relationship, under anarchy states are independent and self-helping, while under hierarchy one commands and the other complies, and relationships are structured and differentiated. I agree with Lake that hierarchy is not well understood: mainstream IR has been dominated by the assumption of anarchy, which obscures the pervasive presence of hierarchy, and scholars have largely ignored or mischaracterized hierarchical relationships because they rely on a formal-legal conception of sovereignty and authority that excludes the possibility of authority between states. But importantly, Lake does not treat hierarchy as a systemic structural feature in the Waltzian sense. It is not the ordering principle of the entire system. Instead it is a dyadic, relational concept that varies between pairs of states; it exists within an anarchic system but mitigates the effects of anarchy for the states caught in hierarchical relationships. He implies it is as important as anarchy and interdependence because it explains behaviors anarchy alone cannot, why some states do not balance against powerful states, why defense spending varies, and why symbolic obedience such as joining coalitions occurs.
Hierarchy produces several observable patterns: subordinate states reduce their own defense spending (free-riding under protection); they bandwagon instead of balancing against the dominant state; functional specialization emerges, as when states rely on the U.S. nuclear umbrella; subordinates offer symbolic obeisance by following the dominant state into wars or alliances; and dominant states discipline or intervene against subordinates who violate the social contract. This helps explain why the U.S. intervenes more in the Caribbean than elsewhere, why European states did not balance against the U.S. after the Cold War, and why some states dollarize their economies or join U.S.-led coalitions. The Dominican Republic case shows a state voluntarily subordinating itself to a dominant power in exchange for political order and protection, forming an informal empire, illustrating that hierarchy rests on a social contract, not just coercion, and that sovereignty is divisible. Beyond Lake’s own examples (the U.S. in the Caribbean, the Soviet Union in Eastern Europe), I would add contemporary informal empires: Russian influence over Belarus and Central Asian states through economic and security dependence, and Chinese influence over Cambodia or Laos through debt diplomacy and infrastructure investment. The classical view holds sovereignty to be indivisible: ultimate authority must reside in a single entity, exclusive and absolute. Lake disputes this because historically sovereignty has often been divided, protectorates, spheres of influence, unequal treaties, and because states in practice transfer authority over domains such as monetary policy or security to other states or international bodies, so indivisibility is a political and normative aspiration, not an empirical reality. I agree with him; the classical view is overly idealized and does not reflect the layered nature of authority in the real world. Figure 2.1 captures this by plotting hierarchy on two dimensions, economic and security, as continua rather than a binary, which allows a more precise classification of relationships, sphere of influence, protectorate, dependency, empire. The typology usefully moves beyond simplistic labels like “empire” and acknowledges that states can hold varying degrees of authority in different domains, reflecting real-world complexity better than unidimensional models.
Kertzer, Brutger & Quek (2024): Perspective Taking and the Security Dilemma
World Politics · a cross-national survey experiment on the microfoundations of the security dilemma
The paper argues that both U.S. and Chinese publics display the asymmetric attributions that fuel a security dilemma, and that perspective-taking, often recommended as a fix, can help or backfire depending on which beliefs it activates: reciprocity among Americans, but opportunism or deterrence-style reactions among Chinese. Classic IR holds that under anarchy uncertainty makes states misread each other’s defensive moves as offensive, producing a spiral, the security dilemma of Herz, Butterfield, and Jervis. Kertzer et al. restate that logic and then test its microfoundations experimentally rather than through historical cases or formal models, fielding parallel dyadic survey experiments in the U.S. and China. A survey experiment lets them causally manipulate what “the other side” does (escalates versus de-escalates) and whether respondents engage in perspective-taking, then observe changes in policy preferences, isolating whether the spiral or deterrence logic shows up in mass opinion. Public opinion matters because mass attitudes can accelerate or inhibit assertive foreign policies in U.S.–China relations; the authors do not deny elite manipulation but argue public sentiment is a constraint and a resource that elites anticipate and shape, in both democracies and autocracies. I tend to agree. We saw the security dilemma in Jervis (1976): a state acts for defensive reasons but fails to appreciate the other’s perceptions, leading to a spiral neither wants. Kertzer et al. measure attribution asymmetry directly: respondents rate whether “we increase” versus “they increase” activity in the South China Sea is defensive or offensive on a 7-point scale, and a larger “they greater than we” gap is security-dilemma thinking. The asymmetry is present in both countries and 3.1 times larger in China (2.89 versus 0.94; p < .001), a clean, face-valid micro measure of the classic attribution mechanism. They then show spiral responses in behavioral preferences: when the other side escalates, respondents shift toward escalation by +0.50 in the U.S. and +0.59 in China (p < .001), again matching the spiral model rather than classic deterrence.
The literature offers two solutions: perspective-taking or empathy to reduce misperception, and costly reassurance or signals to credibly convey benign intent. Personally I find reassurance more robust when beliefs are already hostile, and the authors’ own results caution that perspective-taking is not universally palliative. Perspective-taking can backfire because it activates a person’s existing beliefs about self and other: if people feel identity threat from strong nationalism, or see negative goal interdependence, then taking the other’s perspective can amplify distrust or encourage them to exploit perceived “weakness.” I find the mechanism plausible, and the conditional effects in their data support it. The design (Figure 1) has an attribution module, within-subject, rating motives from 1 (defensive) to 7 (offensive); and a perspective-taking module, between-subject and parallel-dyadic, which randomizes whether the other side escalates or de-escalates, asks treated respondents to write what their country would infer, and then measures policy choice from 1 (decrease a lot) to 7 (increase a lot), an encouragement design, since some controls may already take perspective and some treated may not comply. The sample is a national sample of 1,820 American adults and 1,485 Chinese adults, with the individual respondent as the unit of analysis. The key independent variables are the other side’s behavior (escalate versus de-escalate, randomized) and the perspective-taking treatment (randomized), with moderators including attribution asymmetry and image of the other or negative goal interdependence; national attachment and identity threat come out weak and not significant. The dependent variables are the attribution motive rating and the preference for one’s own military activity. Reading the coefficients by direction and significance: attribution asymmetry exists in both countries and is about 3.1 times larger in China; the spiral response to escalation is positive in both (U.S. +0.50, China +0.59); and perspective-taking has conditional, interactive effects, where negative goal interdependence is high it reduces reciprocity in the U.S. and reduces backing down in China, and where attribution asymmetry is high it loses its palliative effect and can even promote escalation when the other de-escalates. What surprised me is the directional split: Americans reciprocate when China de-escalates, but Chinese respondents are more likely to exploit a perceived softening and back down only when faced with U.S. strength, a deterrence-style response. My critiques concern external validity (a 2016 baseline; South China Sea salience differs across publics), compliance with the writing task, and whether effects would generalize to elite decision-makers, though the authors treat compliance as a conservative encouragement design. The policy implication, which the authors themselves stress as a cautionary note, is that perspective-taking is not a silver bullet: in high-threat belief environments it may activate hostile narratives, so policymakers should combine empathy efforts with credible signals and manage identity and goal-threat narratives.
Understanding Waltz in depth
These are questions I raised for myself while working through Waltz, not from the course.
This is the part of the week I worked hardest on. For me a theory’s value does not rest only on whether it can explain everyday reality, Waltz himself defined “theory” so that it should not be judged by direct comparison with the empirical world. When I read a theory I attend to three things: the problem it responds to (for Waltz, the behavioral revolution and a defense of a structural understanding of IR); the intellectual tradition it continues (I read Waltz inside a long philosophy of categories, form, and structure); and the internal clarity and consistency of its framework. If a theory defines its assumptions clearly and builds a logical system step by step, it is very hard to “disprove” from the inside. One can only challenge it by rejecting its basic assumptions. If we only knock down each other’s assumptions, dialogue across theories becomes almost impossible. That is why I cannot simply “agree” or “disagree” with Waltz: such judgments rest not just on reason but on belief, identity, and values, which are deeply tied to who I am. Some of this I can only think through clearly in Chinese, because it connects to Kant and the philosophy of science; here I set it down in full.
Waltz’s conception of theory
Waltz’s view of what counts as theory is strict and distinctive, rooted in the philosophy of science and especially positivism, and it can be summarized in five points. First, theory versus laws: laws are observed regularities reached by induction (“balance of power has repeatedly appeared in history”), whereas a theory is not a list of laws but an abstract framework that explains why such laws exist, supplying the causal mechanism behind the pattern, theory is “the cause of the laws,” not a catalog of them. Second, theory is an intellectual construction that shapes reality rather than a mirror that copies it: a good theory is created through simplification, abstraction, and omission, deliberately ignoring most empirical detail, leaders’ personalities, a state’s culture, historical accidents, to isolate a few powerful variables and reveal their logical relations, so that a highly abstract, seemingly “unrealistic” model (like the perfectly competitive market in economics) can explain more than an exhaustive description. Third, theory is built by deduction, not induction: induction can discover regularities and hypotheses but cannot create a theory, which is constructed by starting from a few core assumptions and deriving necessary, testable implications, with evidence used to test the derived propositions rather than to build the theory itself. Fourth, the criterion is explanatory power, not the prediction of specific events: the goal is to explain why certain behaviors recur, judged by explanatory power, logical consistency, and parsimony, the fewer variables needed to account for the most outcomes, the better; a theory tells us what to expect, not when to expect it. Fifth, theories are displaced only by better theories, not refuted by a fact or two: because a theory is not a perfect copy of reality, a single counterexample cannot defeat it. It is replaced only when a more rigorous, more parsimonious, more far-reaching theory appears. The competition is a whole-paradigm competition.
Why balance of power is a systemic theory
On this view of theory, Waltz argues that balance of power is the theory of international politics, not simply a theory of foreign policy. It explains a system-level regularity, not unit-level behavior: he sharply criticizes reductionism, explaining international outcomes from state attributes or individual psychology, by noting that the same outcome (war) can arise from different domestic causes, and similar domestic structures can produce different outcomes, so the regularity and recurrence of international results must be sought at the system level. Balancing does exactly that: whatever a state is internally, democratic or autocratic, led by the peaceful or the warlike, the pressure to survive in an anarchic self-help system pushes its behavior toward an overall tendency to check the strong. The balance arises from the structure, not from unit intentions: where traditional balance-of-power thought saw balance as a shrewd strategy consciously pursued by statesmen, Waltz’s revolutionary claim is that balance is a structurally necessary product, even if no state aims at it, the system “selects” the outcome by rewarding states that unintentionally balance and punishing those that do not (by conquest or weakening). This is a parsimonious, elegant deductive system: from three structural elements, the ordering principle (anarchy, the absence of a supreme authority), the character of the units (functional similarity, all states seeking survival above all), and the distribution of capabilities, he derives logically that a concentration of capability, the rise of a potential hegemon, threatens the survival of all other units and triggers balancing that restores some form of balance, with no need for any assumption about motive, culture, or ideology. Modeling it out: hold anarchy constant; assume unit similarity; let the key variable be the distribution of power (polarity); then regardless of internal politics or leaders, states are constrained and pushed to act alike, pursuing security and survival, and balancing is complex and unstable under multipolarity but simple and stable under bipolarity. In short, Waltz offers a theory of the systemically imposed constraints states face when choosing, not a theory of how states make choices. Balance of power is the central expression of that structural constraint.
The intellectual lineage I felt: from Kant to Waltz
I read Waltz inside a long philosophical lineage that wrestled with necessity versus freedom, subject versus object. In the Critique of Pure Reason, Kant argued that knowledge of nature is possible because the mind imposes categories, conceptual molds such as causality and substance, on sensory input, so necessity belongs to the phenomenal world organized through these forms; in the Critique of Practical Reason, freedom belongs to the noumenal realm where the will legislates the moral law as the categorical imperative; and in the Critique of Judgment he sought to bridge the two through aesthetics and purposiveness, yet the “stubborn content” of empirical givenness, contingency, accidents, particular facts, resisted full absorption into rational form, for unlike mathematics, which is pure human-made rational form, political and historical facts remain contingent. Hegel rejected this strict dualism: form and content are inseparable, the categories of thought are the logic of being itself, and rationality unfolds dialectically in history so that subject and object, necessity and freedom, are ultimately united. Fichte identified the hiatus irrationalis, the irrational gap between subject and object, and proposed starting not from given facts (Tatsachen) but from an original act (Tathandlung), a decisive move from contemplation to practice. Lukács later described this tradition’s greatness and tragedy: it confronted contingency but remained within philosophy, and the “system as system” must be transcended once contingency is faced, Marx grounded the reconciliation of necessity and freedom in social practice, treating structures not as fixed but as historically produced and transformable. Against this backdrop, Waltz’s core concepts, anarchy, unit similarity, the distribution of capabilities, function like Kantian categories: not observations but conceptual molds imposed on reality to make it intelligible. So from one angle Waltz reads as a neo-Kantian, like Weber, who holds that scientific knowledge relies on conceptual constructions and that the test of theory is logical coherence rather than direct access to noumenal reality; from another angle his style is Spinozist, since faced with the contingent “content” of international politics, leaders’ choices, historical accidents, cultural meaning, shifting identities, he excluded them as “reductionist” to preserve a closed, rational system, gaining clarity and necessity only by bracketing contingency. Waltz is thus an heir of the Kantian–Hegelian tradition who builds a precise logical framework to make the world graspable; but the very precision of that framework marks its limits.
The hidden fissures: is and ought
On the surface Waltz presents his theory as value-neutral, confined to systemic necessity, yet the fissure between “is” and “ought” reappears in three ways. First, surface neutrality versus implicit normativity: what is feasible under anarchy (states must balance) quietly becomes what is legitimate (states ought to balance), and structural necessity becomes a strategic imperative. Second, methodological closure as a normative choice: privileging structure over domestic politics, culture, or ideas is itself a normative decision about what counts as real explanation. Third, the policy uptake of theory: once theories circulate among policymakers, description becomes guidance, and the “logic of anarchy” functions as a manual for rational statecraft. So Waltz offers a refined categorical construction, but it cannot grasp the aspects of international politics that resist categorization: the contingency and rupture of history (the sudden end of the Cold War), the fluidity of identities (how an “enemy” becomes a “rival” or partner), and the dimension of meaning and value (why some issues are deemed existential, not simply interests). These dimensions are not residual; they may be central to political life, and they can only be shown and grasped in historical narrative and concrete practice, not fully captured by formal theory. Ultimately it is not theory that constructs the meaning of practice; rather, practice provides the basis and the limits of theory, through language games, political action, and historical events.
Reexamining “balance of power has repeatedly appeared”: Europe versus China
The claim that “balance of power has repeatedly appeared” comes mainly from European history, though historians debate how “automatic” balancing really was, emphasizing order-management and postwar reconstruction over an automatic equilibrium. From the Peace of Westphalia (1648) to World War II, Europe was a multipolar arena in which a rising contender, Louis XIV’s France, Napoleonic France, Wilhelmine and then Nazi Germany, was repeatedly checked by counter-coalitions. Waltz abstracted this experience into the model “anarchy + changes in the capabilities of major units → a balancing trend,” and then universalized it as holding across time and space wherever those conditions are met, explaining outcomes regardless of actors’ intentions. But is that universalization valid? Chinese history suggests a very different structural trend: unification. Although periods of fragmentation were long in total, the idea of unification became a powerful political myth and structural expectation. In times of division, the Warring States, the Three Kingdoms, the Northern and Southern Dynasties, temporary multipolar balances did emerge, yet for the major players balance was not the end goal but a stage on the path toward “sweeping away rivals and unifying the realm.” Their aim was to end anarchy itself, not to maintain it, so the underlying logic of the Chinese system was “break the balance for unification,” not “preserve the balance for survival”, fundamentally different game rules and ultimate objectives from Europe’s. Waltz’s focus on power distribution and balancing makes him blind to variables like cultural identity, ideology, and imperial memory; for him these belong to the unit level and are filtered out as noise. But in China those very filtered-out factors, a shared script, the idea of “Zhonghua,” a civilizational identity higher than loyalty to any one regime, created the cultural and ideological foundation for repeated unification. This suggests that a “structural trend” may sometimes be a theoretical artifact, constructed jointly by the researcher’s choice of variables, the boundaries they set, and the philosophy of science they hold: focus on power distribution and you see balance; focus on cultural identity and you see unification.
A dual-mechanism framework
To make this comparison testable rather than merely rhetorical, I would propose two mechanisms. Mechanism A, the European type: anarchy plus high conquest costs, strong geographic barriers, and maritime constraints, plus fragmented identities that are hard to integrate, yields a world where external balancing dominates and balance is sustainable. Mechanism B, the East Asian type: a quasi-hierarchical social order plus low conquest costs, inland connectivity, and economies of scale, plus a legitimacy myth of unification, yields a world where internal consolidation dominates and unification is the main pattern. To let these propositions be confirmed or falsified, I would watch observable indicators: the ordering principle (is there a center with substantive, not simply ceremonial, final authority and enforcement?); alliance turnover (frequent short-to-medium-term shifts, as in balance, versus long cycles of disunity followed by reunification, as in unification); the distribution of war aims (limiting the opponent’s capacity, as in balance, versus absorbing regime and territory, as in unification); legitimacy discourse (sovereign equality, as in balance, versus orthodoxy and a Sinocentric hierarchy, as in unification); and military-geographic-technical variables (maritime barriers and a defensive advantage, as in balance, versus open plains, an offensive advantage, and logistical depth, as in unification).
The constructivist rebuttal
Constructivists would argue that norms and identities are not simply “unit-level properties” but part of the structure of the international system itself. Take NATO: its cohesion is explained not by the internal culture of each member state but by a shared identity of “who we are,” created through decades of interaction and institutionalized norms, and for constructivists that collective identity is itself a feature of system structure in a social sense. From this perspective Waltz makes structure too narrow by reducing it to the distribution of material capabilities, tending to treat the social dimension as a reductionist concern when it is in fact another kind of systemic factor: roles, shared norms, and intersubjective expectations organize behavior across states and are themselves structural. Expanding the concept of structure means recognizing that it is not just about material positions but also about social relations. In short, constructivists see structure as a mix of “hard” and “soft” elements, material power and social meaning together, and norms and identities do not just constrain actors; they also constitute their very preferences.
My remaining puzzles: a Waltz–constructivism Q&A
What are the two core critiques of Waltz?
Critique one is a determinism-versus-tendency tension: if structure (anarchy plus the distribution of capabilities) is claimed to determine state behavior, observed variation, bandwagoning, neutrality, free-riding, contradicts it; but if structure only tends to shape behavior, the theory must specify when and under what conditions convergence should occur, or it becomes hard to falsify. Critique two is thin socialization: Waltz’s “socialization” largely renames behavioral convergence under pressure without explaining where norms come from, how they are internalized, and how they persist, so it lacks mechanisms tied to the social dimension of shared meanings, roles, and identities.
Does talking about norms shaping preferences and identity fall into Waltz’s “reductionism”?
By Waltz’s definition, yes, that is unit-level “who I am.” But constructivists counter that “who we are”, shared identities, expectations, and norms, is part of international structure itself, not a unit property. The dispute is ontological: Waltz restricts structure to material positions and capability distributions, whereas constructivists include social relations as structural, so “who I am” versus “who we are” is a neat shorthand for the divide, for Waltz structure is positions and capabilities and identity is unit-level, while for constructivism structure not just constrains but also constitutes interests and preferences.
Is using “balance of power” to explain a specific state’s policy a level error?
Under Waltz’s strict view, yes. Balancing is a systemic, long-run tendency, not an explanation of a state’s short-run strategic choice, and many applications conflate the two by treating balancing as a discretionary strategy. The cleaner practice is to use Waltz for system-level patterns, why unipolarity is hard to sustain over the long run, and unit-level theories, bureaucratic politics, domestic politics, leaders’ cognition, for specific policies.
Why does Waltz say constructivism “is not a theory,” and does that contradict “assumptions need not be realistic”?
There is no contradiction. Waltz’s objection was not that constructivist assumptions are unrealistic but that early constructivism lacked clear, falsifiable propositions, making it more a perspective than a theory. For Waltz, theory requires parsimony, then explicit mechanisms, then testable hypotheses; he rejects attacking assumptions for being “unrealistic,” but insists that the propositions deduced from those assumptions be tested empirically. His complaint is precisely that the propositions and mechanisms were not clear or testable enough, not that the assumptions strayed from reality.
How can “unrealistic” assumptions still have their deductions tested by reality?
Assumptions are abstract starting points whose value lies in generating observable propositions; the deductions, not the assumptions, are what get tested. Think of it as rules, moves, and a game: the assumptions are the rules, the deductions are the moves, and reality is the game. We test whether the moves fit the game, not whether the rules “resemble” reality. Practically, this requires specifying scope conditions and auxiliary assumptions so the theory does not slide into explaining everything and therefore nothing.
Is constructivism only a “perspective,” or can it become a theory in Waltz’s sense?
Historically, early constructivism was more a research program or lens emphasizing that norms and identity matter. But it later produced middle-range theories: Wendt’s cultures of anarchy (enemy, self-help, friend) institutionalized through interaction; Finnemore and Sikkink’s norm life cycle with thresholds and cascades; Checkel and Risse’s socialization mechanisms, persuasion, role learning, peer pressure, with observable conditions; and Adler’s security communities, where shared identity plus trust make war unthinkable. These specify mechanisms, scope conditions, and observable implications, moving much closer to Waltz’s conception of theory.
Is Waltz’s theory really “value-neutral,” and why does structural realism look more objective?
On the surface Waltz presents a value-neutral scientific model, but at a deeper level his assumptions, survival-seeking states, constant anarchy, functional homogeneity, naturalize a particular picture of power politics and embed a pessimistic stance, which constructivists and critical theorists call value-laden, merely masked by scientific rhetoric. Structural realism looks more “neutral” because it deliberately focuses on measurable material variables (capabilities, polarity, war frequency) and formal principles (anarchy, functional similarity), reducing the interpretive load while emphasizing parsimony and generality; but this does not expunge values, it only abstracts and naturalizes them at the level of the model.
Are constructivist concepts too subjective, and can they be operationalized?
Concepts like shared ideas and collective identity are intersubjective and require interpretive methods, but they are not arbitrary. They can be operationalized through discourse and text coding, archival records, UN or NATO voting, transnational network data, elite interviews and surveys, and process tracing; observable indicators include the density and reproduction of normative clauses in treaties, role consistency among members, the long-run predictability of interactions, and the sanction or shaming that follows violations. The key is to tie mechanisms to indicators and to counterfactuals, asking what patterns we should observe in the absence of the norm or identity.
Why is Waltz called a “balance-of-power” theory and why does it look weak under hegemony?
Waltz’s core implication is balancing, and his model works best under bi- or multipolarity, where competition and balancing are most legible. Under unipolar hegemony the order becomes more hierarchical, and secondary states may bandwagon rather than balance, weakening the predicted convergence; to preserve parsimony, Waltz largely sidestepped a systemic account of durable hegemony. This is exactly where Lake’s hierarchy and the institutionalists pick up the slack.
Can constructivism be used to analyze the rise and spread of realism itself?
Constructivism can be turned on realism itself, at the meta-theoretical, sociology-of-knowledge level: one can analyze how realism emerged, came to dominate, and naturalized its premises by tracing the social construction of its assumptions within epistemic communities. This evaluates the conditions of knowledge production rather than the empirical implications of realism, so it is a different level of analysis from using constructivism as a substantive empirical theory. The one-line takeaway for the whole debate: Waltz demands a parsimonious, mechanism-rich, falsifiable systemic theory, and while early constructivism was a lens, its later work, the norm life cycle, socialization, security communities, yields exactly the testable middle-range propositions he asked for.
Application · Theory in current events
Glaser. 2024. “Fear Factor: How to Know You’re in a Security Dilemma.” Foreign Affairs.
Glaser argues that the recent intensification of great-power competition can be read in either of two ways: rising challengers like China and Russia are “greedy expansionists”, states that want to revise the geopolitical status quo for nationalist, ideological, or leadership ambitions, or they are “security seekers” that act aggressively because they feel insecure and are trying to protect themselves. I do not think it is always useful to sort states into just one bucket, because many are mixed, both greedy and insecure at once, which makes the policy response harder. For a pure security seeker the appropriate response is to reduce the security dilemma, defensive strategies, unilateral restraint, and negotiated agreements to limit offensive capabilities, so as not to make the adversary feel more threatened.
On Russia and Ukraine, the two perspectives offer different origins: a greedy Russia driven by Putin’s beliefs, imperial nationalism, and expansionist goals, versus an insecure Russia threatened by NATO expansion. Glaser suggests Russia is likely a mixed state, which implies the West had no good options, abandoning NATO expansion might have delayed but not stopped the invasion, while accelerating it would have made Russia feel more insecure. On East Asia, he argues the security dilemma intensifies the competition over Taiwan: even defensive U.S. and Taiwanese actions are read by China as threats because they could encourage Taiwan toward independence, creating a cycle of tension and military buildup, so the U.S. is left with only bad options, strengthen Taiwan’s defenses and risk war, or abandon Taiwan and invite invasion. I agree with this analysis; Taiwan is rooted in both national identity and security perceptions, a classic case where even defensive measures can be misread as offensive. The logic travels: Glaser notes the security dilemma has been applied to alliance formation (Snyder), ethnic conflicts after imperial collapse (Posen), and status competition (Wohlforth), and today I would point to India–Pakistan tensions over Kashmir, Iran–Saudi proxy conflicts, and North Korea’s insecurity-driven nuclear program.
But Taiwan also shows the limits of a purely material reading, and here Glaser’s security dilemma meets a constructivist objection. From a realist angle Taiwan matters for geography and military strategy, it sits on the first island chain and affects sea power, yet in practice both the U.S. and China treat it as far more than territory: for Washington it has become a symbol of credibility, since abandoning it would make Asian allies doubt U.S. commitments, while for Beijing it is tied to national unity and legitimacy, framed as the unfinished civil war and central to “national rejuvenation.” Material security is not the whole explanation; interests are shaped by beliefs, identity, and legitimacy narratives, which is exactly what constructivists mean when they say interests are socially constructed rather than simply given by survival. This is also where Waltz’s structural realism reaches its limit: the survival of the United States does not depend on Taiwan, and China’s security is not directly threatened by Taiwan’s current status, yet both define it as a “core interest.” The reason is that the structure of international politics is not just material but also social, carrying identity and legitimacy, the central constructivist critique of Waltz, and a fitting bridge into next week.
← Back to the genealogy tree · ← Prev: Week 3, Foreign Policy Decision-Making · Next: Week 5, Constructivism →