Of Structures and Actors

International Legal Responsibility for “Necropolitical Structural Violence”?

Many of the gravest harms of our time are not produced by a single, spectacular act. People starve because relief does not reach them, die of treatable illness when aid programmes are cut, sicken because pollution and other environmental degradation go unregulated for decades, perish at the borders of militarised migration regimes. Johan Galtung famously called this structural violence: avoidable harm built into social structures. For the law of responsibility, which hinges on agency, the concept has a fundamental weakness: Galtung called violence structural where “there is no such actor” that commits violence (p. 170). That structural harm lies beyond the law seems baked into the sociological definition. There is, however, another interesting strand of social theory that shines a light on agency within such contexts: Achille Mbembe’s necropolitics conceptualises sovereign power as the capacity to expose people to (structural) conditions of premature death. Combining the two, as political geographers have done for the Khmer Rouge famine and Europe’s border regime, I use “necropolitical structural violence” (NSV) to describe severe and diffuse harm produced through a social structure that an actor can significantly influence and knows to be deadly. In environmental and climate cases, courts have already begun to attribute such harm to states. I argue that their reasoning could be carried into other contexts of structural harm.

Interdisciplinary Dialogues: Extra-Legal Concepts and Doctrinal Innovation

NSV is not a legal category, though. Why should doctrine engage with social theory at all? The reason, I think, is doctrinal innovation. Kant’s dictum that “thoughts without content are empty, intuitions without concepts are blind” holds for legal doctrine too. Abstract categories that refer only to one another without a bridge to real-world phenomena say little about how harm is actually produced. Conversely, observations of that harm remain blind and ineffective without legal concepts that can attribute it – and instil change. What NSV describes is agency in structures – agency to which legal responsibility can, and sometimes should, be tied. The concepts of structural violence and necropolitics admittedly carry normative aspects and are therefore contested. But so are the concepts of structural discrimination and vulnerability, which have nonetheless found their way into doctrine. I build on work that has long insisted that structural harm has authors and can be a legal wrong: among others, Susan Marks’s plea to speak of planned misery rather than root causes, Ioannis Kalpouzos and Itamar Mann on banal crimes against humanity, Randle DeFalco on invisible atrocities.

The claim is that such harm escapes the law less because of what it is than because of how the law is usually read. The default interpretations of the law filter out diffuse mass harm: knowledge as foresight of harm to identified persons, influence as control over individual acts or actors, harm as injury caused by a specific act. Epistemologically, the law individualises and zooms in on singular acts or actors to make the social world tangible. These legal “cuts”, however, obscure the relational nature of all human conduct. Recently, courts have started to refine these interpretations in isolated fields, for example when it comes to environmental risk, climate change or structural discrimination. As an analytical lens, NSV groups these doctrinal innovations and shows to which other contexts they could travel. Additionally, NSV can help identify limits of the law as it stands, since it makes visible whether or not the governing rules of a context leave room for such a reading.

Two features of state responsibility matter for NSV, with its multitude of actors and diffuse causal chains: first, conduct of state organs is attributed to the state under Art. 4 ARSIWA, conduct of entities exercising governmental authority under Art. 5. Everyone else, including state-owned companies without such authority, falls under the control test of Art. 8. Second, a wrongful act must be kept apart from proof of injury. Whether harm is an element of the breach depends on the primary rule. Obligations to prevent an event are breached only once the event occurs (Art. 14(3) ARSIWA). The ICJ’s Bosnian Genocide case illustrates the split (though expressly without setting a general rule): the breach did not depend on proof that compliance would have averted the genocide, but compensation failed for want of a “sufficiently direct and certain causal nexus” (para. 462). Causation thus primarily matters for compensation, not breach, unless the primary rule requires it.

Knowledge: From Identified Individual Victims to Known Structural Risks?

In the default interpretation of positive obligations, knowledge is tied to identified victims. A state must act once it knew or ought to have known of “a real and immediate risk to the life of an identified individual or individuals” (Osman v. United Kingdom, para. 116). Somewhat more collectively, in Campo Algodonero, the Inter-American Court required a risk to a specific individual or group. In that case, knowledge of a general pattern of violence explicitly did not suffice (para. 280). Structural harm rarely takes this individualistic form. Those who decide within contexts of structural violence usually only know that harm will be caused, not on whom it will fall specifically.

Courts have begun to recognise that this kind of knowledge can, however, suffice. In Öneryıldız v. Turkey – on dangerous activities rather than Osman’s operational duties – the European Court held that the authorities knew of a risk to “a number of persons” living near a rubbish tip, documented in an expert report (para. 101). In Cannavacciuolo and Others v. Italy, the Court based its finding of an infringement of the right to life to a significant degree on epidemiological reports concerning rising cancer rates. Under the Genocide Convention, the duty to prevent arises when a state learns, or should have learned, of a “serious risk”. The risk here is linked to an event, not to individualised victims (Bosnian Genocide, para. 431). In its March 2024 order on Gaza, the ICJ based its finding that “famine is setting in” – itself a structural condition – on UN and IPC data (paras. 19–21).

NSV makes visible what these cases share, which is knowledge of a structure’s lethality. They also show the form such knowledge takes: alerts, monitoring data, mortality projections – statistical rather than individual. This essentially “scientific” knowledge is also what the ICJ’s climate advisory opinion (also for a structural context) highlights as an “important factor”, since it helps “to assess the probability and seriousness of possible harm, informing the required standard of due diligence” (para. 283 et seq.). While this was established for environmental harm, the open doctrinal texture of, for example, the right to life would (and in Öneryıldız and Cannavacciuolo already did) also allow for this form of knowledge to be taken as a basis when determining a breach: the Human Rights Committee, in its General Comment No. 36, holds that a deprivation of life involves “intentional or otherwise foreseeable and preventable [emphasis added] life-terminating harm or injury, caused by an act or omission” (para. 6). This structural type of knowledge, however, also has its limits, especially in international criminal law, which still requires awareness that death “will occur in the ordinary course of events” (Art. 30(2)(b) Rome Statute).

Influence: From Controlling Actors to Shaping Structures?

On the level of actors, the standard reading ties responsibility – outside of the state’s own conduct – to control over specific actors. Private conduct becomes state conduct only if the state effectively controlled the specific operation (Art. 8 ARSIWA; Nicaragua, para. 115). In a similar vein, human rights obligations abroad follow control over territory or persons. On that reading, structural harm seems intangible, as it often emerges from multiple actors of which the state seldom controls all, if any.

With the expansion of due diligence obligations in international law, jurisprudence has started to shift the focus from the harmful act per se to the structural environments from which those acts emanate as the anchor point for responsibility. In La Oroya v. Peru (2023), the Inter-American Court emphasised states’ duty “to organize the governmental apparatus and, in general, all the structures [emphasis added] through which public power is exercised” so as to ensure “free and full enjoyment of human rights” (para. 108). In the Court’s interpretation, this entails regulating companies so that they are obliged to respect human rights while operating (para. 114). While the Court acknowledged “that compliance with these obligations was a highly complex logistical, financial and technical process, which could not be carried out immediately, but required a progressive approach” (para. 163), it found that over several years, Peru had, among other things, twice extended the operator’s deadlines for environmental compliance, thereby failing to properly use its own power to regulate, supervise and monitor. This approach is mirrored in other environmental cases (cf. the ICJ’s climate advisory opinion, para. 428, or Sacchi et al. v. Argentina, para. 10.5), which arguably make up a significant share of today’s mass harm.

Due diligence shifts the perspective from whose conduct counts to what a state did with its influence over a structure. A failure to exercise due diligence is the state’s own omission, directly attributable under Art. 4 ARSIWA. The focus thus shifts from attribution to the primary obligation, whose intensity grows with the degree of influence. This pathway circumvents the strict requirements of Art. 8. Beyond environmental cases, the reasoning could, for example, extend to the abrupt withdrawal of life-saving aid. There, the donor state does not control the situation in the recipient state, but fully controls the source of harm, which is its own funding decision. Following Sacchi, such control could ground jurisdiction. The limits lie in the primary rules. There is no general duty to fund aid, and treaty bodies have (in cases not related to aid) so far extended jurisdiction where foreseeable harm came with control or dependency, which ongoing treatment programmes could arguably establish. This pathway also carries a clear risk: if any influence over any structure sufficed, powerful states would answer for everything, even though remedies might be beyond their reach. Ascribing responsibility hyperbolically could then undermine the law’s regulatory power and therefore its legitimacy. The “necropolitical” in NSV, however, carries with it an element of significant influence as well as severity that would have to be met to trigger responsibility.

Harm: From Individual Compensation to Structural Remedy?

If we turn to harm, the default understanding treats it as injury or damage to an identified person, caused by a specific act. Individual complaints require victim status. Compensation for an injury or damage then requires a “sufficiently direct and certain causal nexus”. Structural harm is statistical: we know roughly how many are injured, but not who. Also, there is often a confluence of multiple causes. Causation therefore seems incompatible with NSV.

This is true if compensation for individual injuries is sought. As stated above, however, causation is not a general condition of a wrongful act. State responsibility also offers remedies that do not depend on individual causation: cessation and guarantees of non-repetition under Art. 30 ARSIWA attach to the continuing breach itself. Arguably, they might fit diffuse harm better than compensating a limited number of individuals. Indeed, cessation, i.e. stopping the wrongful act if it is continuing in nature, could be read as dismantling a harmful structure. Moreover, causation in structural contexts remains an open question: the ICJ’s climate advisory opinion observes that causation for damages from climate change “involves two distinct elements”, one of which is “whether a given climatic event or trend can be attributed to anthropogenic climate change”, which “may be addressed by recourse to science” (para. 437). Climate change is not the only instance of anthropogenic mass harm. If primary obligations – as in the climate case – were also interpreted to prohibit the furtherance of other such structural conditions, expanding scientific knowledge about those contexts could also be used to establish a causal nexus direct and certain enough for remedies under Art. 31 ARSIWA.

Conclusion

Structural harm in its many forms poses a fundamental challenge to the world’s societies, yet it often looks unregulable by definition. Concepts like NSV, built from sociological theory, offer us a vocabulary to think anew about established doctrine and might enable novel ways of conceptualising legal solutions to social problems, thereby illuminating the innovation – be it ongoing in case law or de lege ferenda in academic discussion – we urgently need to address the 21st century’s many problems.