Classical strategic theory rests on a convenient architecture. Its foundational texts assume that war is a contest between organised political communities, each with a government that decides, an army that fights, and a population that pays. Clausewitz's dictum that war is the continuation of policy by other means presupposes a policy-maker who can formulate ends, calculate costs, and ultimately accept a verdict. For roughly two centuries, that assumption held well enough to make theory useful.

The adversaries that have dominated Western security agendas since the Cold War rarely fit this architecture. Insurgent movements, transnational terrorist networks, and armed criminal enterprises do not sign armistices, defend capitals, or field forces whose destruction compels capitulation. They disperse authority, blend into civilian populations, and often define success in terms their opponents struggle even to recognise. In 1991 Martin van Creveld declared 'trinitarian war' obsolete. Mary Kaldor and Rupert Smith offered parallel diagnoses of new wars and war amongst the people.

These critiques overstate the case, but they identify a real analytical problem. The difficulty is not that classical theory is wrong. It is that its implicit assumptions about who fights, why, and how wars end are carried into planning without examination. This article examines three such assumptions: actor rationality, the definition of victory, and the transferability of core concepts. It then asks which elements of the classical canon survive contact with non-state adversaries.

The classical model treats the adversary as a unitary rational actor: a single decision centre that weighs costs against benefits and adjusts its behaviour accordingly. This assumption underwrites nearly every instrument of coercion. Thomas Schelling's theory of compellence, deterrence by punishment, and the logic of escalation all depend on three conditions. The opponent must possess something of value that can be threatened, a leadership able to perceive the threat, and the authority to change course in response.

Non-state actors strain each of these conditions. Many lack a fixed asset base that can be held hostage. A network dispersed across a dozen countries offers few targets whose loss would alter its calculus. Authority is also frequently fragmented. Jacob Shapiro's work on the terrorist's dilemma shows that militant organisations face acute principal-agent problems. Leaders who want restraint cannot reliably control cells that benefit from escalation, and the secrecy required for survival undermines the command and control on which coercion depends.

It would be a mistake, however, to conclude that such actors are irrational. The more precise claim is that their rationality is organised around different objectives. Martha Crenshaw distinguished between an instrumental logic, in which violence serves political ends, and an organisational logic, in which violence serves the survival and cohesion of the group itself. Mia Bloom's analysis of outbidding shows how competing factions escalate violence to win recruits and constituents. That behaviour looks self-defeating against the state but makes sense within the competitive marketplace of militancy.

Criminal organisations introduce a further variation. Mexican cartels and Brazilian prison gangs pursue profit rather than political transformation. They therefore tend to seek accommodation with state institutions rather than their overthrow. Their violence against the state usually aims to preserve operational space, not to seize power. Applying a counterinsurgency template to such actors misreads their objectives. Applying a purely law-enforcement template may underestimate the scale of their coercive capacity and their ability to hollow out governance from within.

The analytical consequence is that coercion against non-state actors must begin with a disaggregated model of the adversary. Who actually decides? What do they value? Which constituencies, such as funders, recruits, host communities, and rival factions, constrain their choices? Strategy that skips these questions and assumes a state-like opponent will aim its threats at a decision centre that may not exist, or that responds to incentives the strategist has never mapped. The resulting failures are then misattributed to the enemy's fanaticism rather than to the strategist's model.

Takeaway

The question is rarely whether a non-state adversary is rational, but what its rationality is organised around and who actually holds the authority to act on it.

Clausewitz described the aim of war as compelling the enemy to do our will, achieved by rendering him defenceless. In interstate war, this aim has reasonably clear operational correlates: destroying the enemy's main forces, occupying territory, breaking the government's will. Victory, though rarely total, could be registered in a formal act such as an armistice, a treaty, or a surrender signed on a battleship's deck. A visible causal chain connected the military aim to the political object.

Non-state conflicts sever this chain. An insurgency without a capital has no centre whose fall forces capitulation. A terrorist network has no army whose annihilation renders it defenceless. Henry Kissinger captured the resulting asymmetry in 1969: the guerrilla wins if he does not lose; the conventional army loses if it does not win. For the weaker side, survival itself becomes a form of victory and protraction becomes its strategy, exactly as Mao's theory of protracted war prescribed.

Andrew Mack's 1975 analysis of why big nations lose small wars added a further dimension. For the intervening power, the conflict is limited; for the insurgent, it is existential. This asymmetry of stakes produces an asymmetry of will. Material superiority cannot compensate for a political gap in how much each side is prepared to endure. As a result, the stronger party's domestic politics often decide the outcome long before battlefield attrition does.

The record of how such conflicts actually end reinforces the point. Audrey Kurth Cronin's study of terrorist campaigns found that groups rarely end through military defeat alone. More often they decline through loss of popular support, internal fragmentation, negotiation, or a shift into other forms of activity. Leadership decapitation applies the classical logic of striking the enemy's command, but its results vary widely. Jenna Jordan's research suggests that older, larger, and more bureaucratised groups prove remarkably resilient to it.

The practical implication is that victory against non-state actors must be defined as a political condition rather than a military event. Realistic objectives tend to be relative: reducing an adversary to a manageable level of threat, separating it from its constituencies, or bringing it into a political process, as Colombia eventually did with the FARC. This is uncomfortable for democracies that expect wars to end cleanly. Yet a theory that promises decisive outcomes against diffuse adversaries generates expectations that make perceived failure almost certain.

Takeaway

Against an adversary for whom survival is success, the decisive battle is not on the battlefield but in the definition of what counts as winning.

The claim that classical theory is simply obsolete does not withstand scrutiny. Van Creveld's attack on trinitarian war rested partly on a misreading. Clausewitz's trinity consists fundamentally of passion, chance, and reason, which he only loosely associated with people, army, and government. As Christopher Bassford and others have argued, this deeper trinity describes any organised violence, including violence waged by non-state actors. Clausewitz himself lectured on small wars and wrote on the 'people in arms.'

Several concepts therefore retain considerable force. The primacy of politics remains essential, arguably more so, since non-state conflicts are fought over legitimacy and allegiance. Friction applies with particular severity to campaigns conducted among civilian populations, where information is ambiguous and every action has second-order effects. Even the centre of gravity survives, though it moves. In counterinsurgency, theorists from David Galula to David Kilcullen located it in the population's support rather than in the enemy's forces.

What requires modification is the relationship between means and ends. Galula's view that counterinsurgency is roughly eighty percent political and twenty percent military inverts the classical weighting. Rupert Smith argued that force must now be used to establish conditions rather than to decide outcomes, which reframes the purpose of military action entirely. Coercion must be redesigned to target networks of enablers, such as financiers, sanctuaries, and sympathetic populations, rather than a single decision centre. In this setting, deterrence by denial often proves more useful than deterrence by punishment.

The adversary's trajectory also matters. Non-state actors are not fixed types; they move along a spectrum. When a group seizes and governs territory, it acquires state-like vulnerabilities. The Islamic State's caliphate between 2014 and 2017 and the Tamil Tigers' administration of the north before 2009 are examples, and in both cases conventional operations regained decisive potential. When a group reverts to clandestine networks, those same methods lose their purchase. Strategy must recognise which form the adversary currently takes and anticipate its transitions.

The most productive adaptation, then, is not abandonment but disaggregation. Classical theory should be treated as a set of conditional propositions, valid only under certain conditions of adversary structure, objectives, and relationship to territory and population. Its errors arise when conditional claims are treated as universal. Strategists who ask which assumptions hold in a given case will find the canon remarkably serviceable. Those who apply it wholesale will keep fighting the war they expected rather than the one they face.

Takeaway

Classical strategic concepts are best read as conditional propositions: the strategist's real task is identifying which conditions currently hold for the adversary in front of them.

Non-state adversaries do not invalidate strategic theory; they expose its hidden premises. The unitary rational actor, the expectation of decisive victory, and the location of the enemy's centre of gravity in its armed forces were always contingent features of interstate war. Strategists mistook them for universal laws.

The enduring value of the classical canon lies in its method rather than its specific conclusions. Clausewitz offered not a manual but a way of thinking about war as a political phenomenon shaped by passion, chance, and reason. That framework can still accommodate insurgents, terrorists, and criminal enterprises, provided strategists interrogate its assumptions rather than simply inherit them.

The refinement required is a typology of adversaries based on their organisation, objectives, and relationship to territory, linked to explicit conditions under which particular concepts apply. Such a theory would be less elegant than its predecessors. It would also be considerably more honest about the limits of force.