The Sorting Machine: Innocence as Political Demand in the Laws of War

The law does not merely protect the innocent. It produces them. To receive protection under the laws of war, a person must first satisfy a set of conditions that have nothing to do with their moral worth and everything to do with their political behavior. They must remain still. They must remain disengaged. They must remain legible as harmless to whoever holds the authority to sort. This demand has been present in the laws of war since at least 1900. What has changed across more than a century of humanitarian progress is not the demand itself but the sophistication of the machinery used to enforce it and the range of actors authorized to operate that machinery. The history of international humanitarian law, read through the category of innocence rather than through the institutions that claim to protect it, is a history of refining the conditions under which protection attaches — and of explaining away, at each successive stage, the suffering of those who failed to satisfy them.
Before the Hague Conventions codified anything, the classical law of war was at least honest about what it was doing. The civilian population was bound to the fate of its sovereign. Citizens were passive enemies. They warranted some protection, but only conditionally, only insofar as they genuinely did nothing. The franc-tireur — the farmer who took up a rifle without a uniform or a chain of command — forfeited everything. This was not squeamishness about civilian suffering. It was a considered theory: that the relevant distinction was between those who submitted to the state’s monopoly on violence and those who challenged it. Protection followed submission. The irregular fighter who blurred the line between soldier and civilian was the primary threat the law feared, not because he was more dangerous than a regular soldier but because he made sorting impossible. Innocence, from the beginning, was a status conferred on those who did not complicate the picture.
The Hague Conventions of 1899 and 1907 formalized this understanding. The civilian keeps protection by staying in place. Sorting authority rests with the commander in the field, who looks at a body and decides whether what he sees is a lawful combatant or an unlawful participant. The law rewards stillness and penalizes the defender who acts outside approved structures. The franc-tireur debates of the 1870 Franco-Prussian War sat just beneath every negotiating session. Germany, which had fought through irregular resistance in France, was the most insistent that combatants must carry arms openly, follow a responsible command, wear a fixed distinctive sign. The requirement was presented as a humanitarian advance — soldiers also have rights and deserve to know who they are fighting. The effect was to define any deviation from state-sanctioned military organization as a forfeiture of protection. Innocence was conditional on visible compliance with the very institution that was doing the killing.
The First World War did not change the rule. It changed the image used to enforce it. Allied propaganda, responding to the German invasion of Belgium, produced the modern template of innocence: the violated woman, the orphaned child, the elderly civilian shot in the street. These images were not simply false, though some were. They were selective. They fixed in the international imagination a figure of innocence that was feminized, passive, and entirely without agency. The propagandists understood, correctly, that the most compelling innocent is the one who cannot possibly threaten anyone. The raped woman and the bayoneted baby are innocent precisely because they could not have done anything to deserve what happened to them. They had no politics, no organizational capacity, no weapons. They were acted upon absolutely.
This imagery fed back into law by stabilizing a template. The protected person is the one who does not act, does not organize, does not resist. The more passive the figure, the more legible the innocence. Sorting authority began to migrate outward from the battlefield commander toward the political elites and public narratives that fixed these images in the international imagination, deciding whose suffering would be recognized as wrongful and whose would be explained away. The person who fought back, even against an invading army, became harder to place in the category the law was building. Defense looked like participation. Participation looked like forfeiture.
The Nuremberg Tribunal hardened the template while appearing to transcend it. The paradigmatic innocent was now the Jew in Nazi Europe — someone subjected to state violence of unimaginable scale without having chosen the conflict, without having done anything to invite it, without any capacity to alter the machinery directed against them. The moral power of this figure is genuine and the historical justification for it is complete. But it locked in a structure whose implications extended beyond its immediate context. Innocence, at Nuremberg, meant radical passivity in the face of organized power. Responsibility was concentrated entirely at the top, on the leaders who planned and launched the violence. The civilian remained a tragic backdrop — someone acted upon rather than acting. Political agency on the part of the victim was not just irrelevant to the legal analysis; it was invisible. The law built its new architecture around the figure of the wholly defenseless, and that figure became the standard against which future claims to protection would be measured.
The 1949 Geneva Conventions expanded the circle of protected persons and introduced real, enforceable rules about the treatment of prisoners, the wounded, and civilians under occupation. These were genuine humanitarian advances. But they came with a condition that has done more work than any other clause in the history of the laws of war. Protection attaches to the civilian who refrains from taking a direct part in hostilities. The moment that threshold is crossed, protection is suspended — for such time as the direct participation continues. The clause polices the boundary of innocence through activity. Feed a fighter, carry a message, scout a road for an armed group, and your status becomes contestable. The law had always demanded passivity. Now it stated the demand explicitly, as a formal legal condition, and handed sorting authority to military lawyers and targeting processes who would operationalize it in real time.
The requirement sounds neutral. In practice it is not. The people most likely to feed fighters, carry messages, and provide shelter and intelligence are the people who live in conflict zones, whose communities are occupied, whose families are threatened, and who respond to those conditions by doing what people in their position have always done: organizing their own survival and resistance. These are ordinary responses to extraordinary danger. Legally, they are also the moments when protection can thin or vanish. The law does not protect the person who defends themselves in ways it has not pre-authorized. It protects the person who endures.
The 1977 Additional Protocols responded to the era of decolonization and appeared to mark a significant departure. Wars of national liberation were recognized as international armed conflicts. Guerrilla fighters who did not wear uniforms could qualify for combatant status under certain conditions. The law seemed, for a moment, to be acknowledging that resistance to colonial occupation was politically and legally legitimate. What it actually did was absorb that acknowledgment into its existing framework without changing the framework’s fundamental demand. The civilian still had to remain not directly participating. The more a population organized its own defense under occupation, the more it risked sliding out of protection at the moment of action. The law refined the language of distinction and proportionality, placed new burdens on military planners, and expanded the class of professional legal advisers authorized to define the limits of permissible participation. The underlying demand — remain passive, remain disengaged, or accept the consequences — survived every revision intact.
What changed in the 1990s was not the demand but the visibility of the victim it centered. The ad hoc tribunals for the former Yugoslavia and Rwanda, and later the International Criminal Court, built their jurisprudence around identity-based persecution. The civilian population became the primary object of protection against genocide and crimes against humanity. Innocence attached to membership in a targeted group. This made certain harms newly legible that previous frameworks had missed or minimized. The specific suffering of specific communities — their systematic destruction as groups — became the organizing concern of international criminal law. But the core intuition persisted. Innocence was strongest, most legally compelling, when the victim appeared entirely defenseless and non-participating at the moment of attack. Judges and investigators reconstructed patterns of victimization after the fact, and the cases that landed most cleanly were the ones where the victims had done nothing. The tribunal system could accommodate the organized massacre of a passive population far more easily than it could accommodate the ambiguous violence of a war where victims were also sometimes fighters.
It was after 2001 that the passivity requirement showed its sharpest edge. Counterterrorism created conditions in which armed groups lived among civilian populations, and states argued that this embedding made sorting nearly impossible without expanding the definition of who counted as a participant. The concept of continuous combat function entered the legal vocabulary. A fighter who put down a weapon and picked up a phone to organize the next operation did not, under this analysis, return to civilian status between operations. Participation was defined not by the act of pulling a trigger but by the ongoing role a person played within an armed group’s organizational structure.
The Israeli Supreme Court’s ruling in the targeted killings case of 2006 crystallized what this expansion meant in practice. The court rejected what it called the revolving door of protection — the idea that a fighter could shed and resume civilian status by setting down arms. It recognized a category of persons who perform a continuous function within an armed group, including organizers, recruiters, and logisticians who may never personally commit an act of violence. Under this framework, the political organizer in a conflict zone who coordinates community resistance, manages logistics, or recruits participants in an armed movement becomes a legitimate target. Not because of anything they have done with their hands. Because of their role, their associations, their history of engagement.
The court presented this as a natural extension of the duty to distinguish combatants from civilians. In a sense it was. The logic follows directly from the demand that has governed the field since 1900: innocence requires non-participation, and participation now includes influence, organization, and political agency exercised in the service of armed resistance. What the court made explicit was something the framework had always implied. The only fully innocent civilian is the one who does nothing. The one who organizes a community, coordinates a defense, sustains a resistance movement — this person has complicated the picture. And to complicate the picture is, in the law’s terms, to accept a degree of risk that the fully passive civilian does not face.
The cost of this logic is not hard to identify. When direct participation is defined broadly enough to include organizers and recruiters, the space of legitimate political resistance under occupation effectively disappears. Anyone who takes a role in sustaining the organized life of a community resisting occupation — and almost everyone in such a community takes some such role — becomes potentially targetable. The law does not protect the political actor. It protects the person who has withdrawn from politics entirely. Under occupation, withdrawal from politics is not a neutral act. It is a form of submission. The law, by protecting only those who submit, is not neutral between occupier and occupied. It is the occupier’s instrument, dressed in the language of universal protection.
The latest developments have not reversed this logic. They have automated it. Data-driven targeting turns the person into a pattern of signals. Innocence becomes a function of whether a phone travels with known fighters, whether a location matches a target profile, whether a pattern of life deviates from what an algorithm defines as civilian behavior. The demand for passivity survives in the data stream. The system rewards those who remain within expected behavioral thresholds and flags those who move in ways associated with participation. Humans sign the warrants, but the shape of the decision is set upstream by training data and system design. The sorting authority has migrated into workflows and code, but the underlying question is the same one the Hague commanders were asking in 1900: does this person look like a civilian or a participant?
Alongside state targeting systems, NGOs and humanitarian organizations have become powerful rival sorters. They document harm, classify the dead, and publish counts of civilian casualties. A person may be recorded as a legitimate target in a military file and as an innocent civilian victim in an Amnesty International report. The category has become a site of institutional competition. The state sorts for action and justification. The NGO sorts for recognition and accountability. Neither system asks whether the underlying demand — remain passive or accept the risk — is itself legitimate. Both compete to control the answer to the same question: who counts as innocent.
Followed across more than a century, the category of the innocent reveals a consistent political demand beneath its shifting forms. In 1900, innocence required visible submission to the state’s monopoly on violence. In 1914, it required the appearance of helplessness. In 1945, it required radical passivity before organized power. In 1949, it required formal abstinence from direct participation in hostilities. In 1977, it required the same abstinence even under occupation and colonial domination. After 2001, it required not just behavioral abstinence but biographical distance from any organizational role in armed resistance. Today it requires passivity legible to an algorithm.
Each refinement was presented as a humanitarian advance. Each made some victims more visible. Each simultaneously moved the line so that a different set of people, those who responded to danger by organizing rather than enduring, fell outside the category at the moment of their greatest need. The law does not protect people who challenge organized force from within the civilian population. It protects people who accept what is done to them. It tells the occupied, the besieged, the persecuted: you may survive, or you may resist, but if you resist in ways we have not pre-authorized, you do so without our protection.
This is the demand the field has never openly stated and cannot state, because stating it would expose the humanitarian framework’s deepest tension. A law that protects only the passive is not a law for people in danger. It is a law for people who have already given up. The sorting machine has grown more sophisticated, more precise, more distributed across institutions and systems and code. The act at its center has not changed. It is the continuous classification of human beings into those whose suffering the world is permitted to mourn and those whose suffering the law explains away by their failure to remain still.