When a head of state authorizes military force, they rarely possess the kind of certainty that classical just war theory seems to presuppose. Adversary intentions are opaque, weapons inventories are estimates, and the second-order consequences of intervention ramify in ways that defy reliable prediction. Yet the tradition stretching from Augustine through Walzer often speaks as if the moral agent has access to facts that real decision-makers simply do not.

This gap between idealized theory and the actual epistemic situation of those who wield force has become more, not less, urgent in an era of cyber operations, drone strikes, and contested intelligence about weapons of mass destruction. The 2003 Iraq War, debates over preemptive strikes against nuclear programs, and humanitarian interventions all turn on judgments made under conditions of irreducible uncertainty.

What follows is an attempt to extend just war theory into this terrain—to ask not what would justify force if we knew everything relevant, but what justifies force when we cannot. The question is not merely practical. It bears on how we understand sovereignty, the moral standing of probabilistic harm, and the burden of justification states owe one another and to those who will bear the costs of war.

Traditional just cause analysis treats the triggering wrong as a fact to be discovered: aggression has occurred, atrocities are underway, a treaty has been violated. The moral question concerns the proportionate response. But in practice, the existence and gravity of the wrong is itself contested, and the evidentiary standards governing that contest have rarely been articulated with the rigor accorded to proportionality or last resort.

Consider the structural analogy to criminal law. Domestic legal systems do not demand metaphysical certainty before authorizing coercion; they demand specified standards—probable cause, preponderance of evidence, beyond reasonable doubt—calibrated to the stakes. Just war theory has nothing comparable. Yet the stakes of military action are categorically higher than most domestic legal interventions, which suggests the evidentiary threshold should be correspondingly demanding.

A defensible framework would distinguish at least three epistemic conditions: the credibility of the threat or wrong, the reliability of the inferential chain from evidence to conclusion, and the openness of alternative interpretations. Force premised on intelligence that fails any of these tests carries an additional burden of justification—not because uncertainty negates just cause, but because acting on uncertain grounds shifts moral responsibility for error onto the actor.

This has institutional implications. The legitimacy of recourse to force depends partly on the procedures by which evidence is gathered, scrutinized, and exposed to challenge. Unilateral assertions by interested parties warrant deeper skepticism than findings vetted through independent bodies. The procedural cosmopolitanism implicit here is not merely a check on power; it is constitutive of epistemic justification itself.

What remains contested is whether good-faith error under genuine uncertainty exculpates. The honest answer is that it mitigates without erasing responsibility. To act on incomplete information is to accept a form of moral risk, and the agent who imposes that risk on others bears it even when their epistemic conduct was reasonable.

Takeaway

The right to use force is not just about whether a wrong exists, but about whether you have done enough to know it does. Epistemic conduct is itself a moral question.

Preemption against an imminent attack has long been accepted within the just war tradition. Prevention—striking against a threat that has not yet crystallized—is far more controversial, and the controversy intensifies under uncertainty. The preventive logic asks decision-makers to weigh present, definite harm against future, probabilistic harm. The asymmetry is morally significant.

The strongest case for prevention treats catastrophic and irreversible threats differently from ordinary security concerns. A nuclear-armed adversary believed to be hostile presents a future risk so severe that even a low probability might warrant action. Yet this reasoning generalizes dangerously. Almost any rivalrous state can be characterized as potentially threatening, and the historical record of preventive wars is a record of confident predictions that proved wrong.

The deeper philosophical problem is that preventive war substitutes the actor's judgment about another state's future conduct for that state's actual conduct. It treats sovereignty as conditional on predicted intentions rather than on present behavior. This corrodes the framework of mutual recognition that distinguishes a society of states from a war of all against all.

A more defensible position narrows preventive justification to circumstances where three conditions converge: the anticipated harm is catastrophic and largely irreversible, the evidence of intent and capability is robust across multiple independent sources, and non-military alternatives have been genuinely exhausted. Outside these conditions, the burden of uncertainty should fall on the would-be intervener.

It is worth noting that the preventive impulse often masks other motives. States rarely undertake costly military action against hypothetical futures unless present interests are also implicated. Honest analysis requires distinguishing the genuine preventive logic from the rhetorical use of future threat to legitimate present ambition.

Takeaway

When we use force against a threat that has not yet materialized, we are betting that our prediction is more reliable than another sovereign's right to be judged by what they actually do.

Proportionality requires that the harms inflicted by military action not exceed the goods achieved. The standard formulation treats both sides of the ledger as estimable. In real conflicts, both are deeply uncertain: civilian casualty projections rest on contested models, strategic outcomes depend on adversary responses, and long-term consequences include political and humanitarian aftershocks that no planner can foresee.

Expected-value reasoning offers an obvious framework: multiply each outcome by its probability and compare aggregated harms to aggregated goods. But this approach struggles with the moral asymmetry between definite and probable harms, between harms one inflicts and harms one fails to prevent, and between identifiable victims and statistical lives. The mathematics is tidy; the morality is not.

A more adequate account would recognize that proportionality under uncertainty is not a single calculation but a structured judgment with several components. It must address how to weight low-probability catastrophic outcomes, how to handle systematic biases in casualty estimation (which historically underestimate civilian harm), and how to account for the irreducible difference between intended and foreseen consequences.

There is also a temporal dimension. Conflicts unfold, and proportionality assessments made at the outset rarely survive contact with reality. A morally serious framework requires ongoing reassessment, with mechanisms to halt or alter operations when emerging evidence shifts the calculus. Proportionality is not satisfied once at the moment of decision; it is a continuing constraint.

Finally, who bears the residual uncertainty matters. When estimates are unreliable, someone absorbs the cost of error—usually civilians in the territory where force is applied, who had no voice in the decision. This distributive question is rarely surfaced in proportionality debates, but it is central to whether the calculation is morally legitimate at all.

Takeaway

Proportionality is not a one-time equation but an ongoing obligation, and the question of who bears the cost of our miscalculations is itself a question of justice.

Just war theory was built for a world in which moral facts were assumed to be more accessible than they are. Extending it into conditions of genuine uncertainty does not soften its constraints; it tightens them. The actor who proposes to use force under uncertainty bears not only the traditional burdens of just cause and proportionality but an additional burden of epistemic justification.

This reframing has implications for institutions as well as individuals. The legitimacy of military action depends increasingly on procedures—evidentiary standards, independent review, multilateral deliberation—that distribute the cognitive labor of moral judgment beyond the interested party. Sovereignty in the use of force cannot be unilateral epistemic sovereignty.

The honest conclusion is uncomfortable. Many wars that seemed justified at the time fail the standards uncertainty imposes. A theory adequate to global political reality will license force less often than its proponents wish, and demand more humility than its critics expect. That is the price of taking both the gravity of war and the limits of knowledge seriously.