Challenging Common Assumptions About Civilian Protection in Large-Scale Combat Operations

Preparations for large-scale combat operations (LSCOs) by the U.S. military and other NATO militaries have led to a debate—no doubt familiar to Articles of War readers—about whether, or to what extent, the conduct of such operations requires modifying the prevailing approach to civilian protection obligations under the law of armed conflict (LOAC). But what if the debate is based on a series of false premises?

The starting gun was fired in 2021 when senior military lawyers, reflecting on decades of U.S. counter-insurgency (COIN) operations, identified a “COIN mindset” characterized by a “general aversion to collateral damage risk.” They argued that a constriction in “legal maneuver space” and the belief that “precision warfighting is legally required under the rules of war” constitute “one of the greatest dangers to our future success.”

By 2025, scholars identified an entire movement of “LSCO lawyering” that seeks “to reframe existing, well-established rules, not by changing them outright but by stripping them to their most minimal defensible core.” This movement was said to have reacted to “a dangerous misalignment between current legal expectations and the demands of high-intensity war against a peer adversary.”

In response, civilian protection advocates have warned that the U.S. military and its allies should take care “not to cast aside measures to protect civilians as a vestige of the counterinsurgency era,” but “should take steps now to adapt planning, training, tactics, and tools in order to better protect civilians in scenarios in which they may find themselves fighting in densely populated areas.” Building on the claim by former Secretary of Defense Austin that the U.S. Civilian Harm Mitigation and Response Action Plan was “scalable and relevant to both counterterrorism operations and large-scale conflicts against peer adversaries,” an important 2024 Center for Naval Analyses report commissioned by the Department of Defense concluded that “civilian harm mitigation does not equate to restraint.” Rather, it “promotes the effective and precise use of force.” Furthermore, the report found that while “[g]iven the scale of civilian harm, commands will likely need to assess civilian harm at the macro level rather than incident level,” in LSCO as in COIN “a comprehensive, effective approach to Civilian Harm Mitigation and Response will result in strategic gains for the US.”

In this post, I will not take a side in this debate or comment on its intensification under the current U.S. administration. Rather, I will seek to point out that arguments on both sides tend to rely on a series of common assumptions about civilian harm in LSCOs that are erroneous or at best misleading. Questions concerning the relevance of precision, restraint, and the limits of collateral damage are undeniably important, but they do not approach the central differences between COIN and LSCO or identify the salient legal questions for civilian protection in conducting operations against a peer adversary.

Assumption 1: Larger-Scale Operations Will Lead to More Civilian Harm

While it might seem a fair bet that more war creates more destruction, it does not necessarily follow that a LSCO against a peer or near-peer adversary will generate increased civilian harm compared to counterinsurgency operations.

The largest and longest contemporary LSCO is the conflict between Russia and Ukraine. The civilian death toll, currently estimated by the UN monitoring mission at some 15,000, is shocking but still not as high as that of recent counterinsurgency wars, including in Iraq and Afghanistan. Unlike those COIN wars, civilian deaths in the Russia-Ukraine conflict also constitute a small proportion of overall casualties in a conflict marked by many hundreds of thousands of combatant deaths. The reasons for this are easy to grasp. While there have been significant urban confrontations, including in Mariupol in 2022, major hostilities have seen regular armies in the field (albeit with auxiliaries on the Russian side) face off across a 1,000 km line of contact, most of which runs through sparsely-populated rural terrain.

Civilian harm is less a function of the scale of military operations than of their proximity to the civilian population. When hostilities are conducted against insurgents embedded in the civilian population, particularly in urban areas, civilian harm tends to be high, even when necessary precautions to mitigate such harm are taken. Conversely, when hostilities take place in areas of low population density, particularly between regular armed forces, military objectives can be identified with greater certainty, and civilian harm is easier to avoid.

The need to spare the civilian population is mandated by the LOAC obligation to take constant care and applies not just to the conduct of attacks but to all military operations (AP I, art. 57(1); DoD Law of War Manual, para. 5.2.3.5). It is clearly relevant to the choice of where to engage the enemy, and arguably the determining factor in the maneuver warfare that is often referenced in LSCO debates. (Professor Geoff Corn and Tyler Smotherman have proposed that the constant care obligation should be promoted to a core IHL principle in an era of maneuver war.) In the 2022 counter-offensive to retake the city of Kherson, Ukrainian armed forces were operating under a clear political imperative to avoid mass civilian casualties. A significant military victory was nonetheless secured by following a strategy based on severing Russian supply lines and enabling Ukrainian forces to engage Russia away from the civilian population.

How exactly the legal obligation to take constant care to spare the civilian population should act to limit operations in areas of high-population density is, however, a controversial question (as illustrated by the negotiations over the UN Political Declaration on the Use of Explosive Weapons in Populated Areas) and outside the scope of this post. My point here is that there is no necessary reason why civilian harm should increase with the scale of operations, as the Ukraine-Russia war demonstrates. This is not to deny that it might do so (including under the influence of factors other than scale, such as speed, or degraded intelligence and communications). But in conventional maneuver warfare against a peer adversary, whether in the eastern European theatre or on the high seas, it should not be assumed that harm to civilians will automatically be high, unless of course belligerents make civilian harm an objective.

Assumption 2: Most Civilian Deaths Will Be as a Result of Collateral Damage

Both military lawyers and civilian protection advocates in the United States have tended to focus their civilian harm mitigation efforts on the limitation of collateral damage. The assumption is that, although isolated instances of deliberate wrongdoing or criminality may occur, the United States and allies seek to act according to LOAC and that therefore most if not virtually all the direct civilian harm that may result from their actions is lawful, either incidental and proportionate to attacks on military objectives or the result of reasonable error or malfunction. Part of this focus is no doubt due to the Pentagon’s own messaging about precision warfare over two decades.

Looking more widely across history, however, most direct civilian harm in high-intensity warfare is not collateral damage, properly understood, but rather the result of attacks directed at the civilian population or indiscriminate attacks which fail to distinguish between military objectives and civilian objects. The Second World War, increasingly referenced in the LSCO debates, is a case in point. There was, of course, extensive collateral damage, but most civilians were killed, across both European and Pacific theatres, by the exterminatory policies of Nazi Germany and Imperial Japan or, to a lesser extent, by “strategic” bombing of urban areas by the Allied Powers undertaken for the purpose of undermining enemy morale. On the whole, civilians were killed because belligerents sought to kill them.

The lessons for civilian protection in LSCO are as clear as they are uncomfortable. While the LSCO debates share a preoccupation with the relative merits of precision, tolerance levels for incidental harm, and delegation limits for strike authority, the greater challenge is likely to come from the interpretation and implementation of the fundamental principle of distinction.

Unlike many of its allies, the United States has long held that “war-sustaining” objects qualify as military objectives, including in particular objects that generate finance for the enemy (DoD, Law of War Manual, para. 5.17.2.3). But while this may be used to justify destroying the Taliban’s opium business (an example cited in the Manual), what does it mean when applied to an enemy State, to whose finances the activities of millions of civilians may contribute (or indeed the majority through the tax system)? Surely civilians who assist war-sustaining activities cannot be held to be directly participating in hostilities, as suggested in a recent Justice Department memorandum.

The United States previously condemned Russian air strikes on Ukraine’s energy grid, and the UN monitoring mission described them as probably in violation of IHL (they have also been the subject of ICC arrest warrants against four Russian military officers for, inter alia, the war crime of attacking civilian objects). But in a LSCO, one may wonder whether the United States recognizes the inviolability of an adversary’s energy infrastructure, or rather treats (parts of) it as a lawful target due to nature, location, purpose, or use. In an era of multi-domain operations, questions arise about which elements of essential infrastructure can survive being (re)classed as military objectives.

The recent roll-out of AI systems such as NGA Maven and Maven Smart System across combatant commands has seen targeting timelines, from sensing to engagement, drop from hours to minutes, with targeting decisions made at the potential rate of 1,000 per hour. Will the automated processing of threat data include such pattern-of-life or proxy indicators that characterized “signature” strikes against military-age males under COIN? In the shift to inter-State war, will civilian leaders and domestic law enforcement be designated “terrorists” or “militants” and targeted for attack?

Again, many of these questions raise complex or controversial issues for LOAC, and I will not analyze them here. But in the context of an existential LSCO against a peer adversary, how they are decided will have immense consequences for civilian protection and how closely the world approaches the “total war” of the 1940s.

Assumption 3: Kinetic Action Is the Primary Cause of Civilian Harm

The move to modify LOAC guidance in contemplation of a LSCO against a peer adversary has mostly concerned the rules on the conduct of hostilities, and in particular targeting law. However, humanitarians should pay equal attention to the implementation of that other part of the jus in bello, the law protecting victims who are in the power of a party to conflict, including civilian populations under occupation.

It is now widely accepted that most conflict-related civilian deaths are not caused by people being directly killed but rather by the effects of displacement and disease and the degradation of healthcare, sanitation and access to food and water. We should not allow our heads to be so turned by the LSCO rhetoric of “lethality” as to ignore the non-kinetic aspect of war, which in the long run can prove more deadly. From the legal perspective, that includes the permission and facilitation of humanitarian access, an occupying power’s positive “duty to ensure the food and medical supplies of the population” (GC IV, art. 55), as well as the need to control the reverberating effects of attacks, at least where such effects are reasonably foreseeable.

In a rare but instructive recent article, Jenny McAvoy summarises some of the specific risks that a large-scale conflict would pose to civilians in the Asia-Pacific region and underscores how the prospective belligerent parties must prepare now to mitigate civilian harm. She cautions that “damage to power, water and sewage, roads, and transportation systems, resulting from … missile attacks, cyberattacks, and force-on-force land battles would cause displacement at an enormous scale and create life-threatening public health consequences for millions.” She further notes that an “estimated one-third of world’s shipping passes through the South China Sea and small island nations and archipelagos across the Pacific depend on commercial shipment of goods for food security and basic provisions, all requiring safe passage.”

Following the downgrading of the Civilian Protection Center of Excellence, it is unclear whether the U.S. military is considering these issues.

Key Takeaways

Militaries always have the tendency to fight the last war. That is to say, tactics and strategy in any new conflict may be determined less by its immediate context and unique challenges than by the insights gained from fighting the last conflict (and the technical innovations and training they spawned). For LOAC practice, which generally evolves more slowly than other military disciplines, the current trajectory appears to approach LSCO as an exponential function of COIN: LSCO = COIN^X. The main outcome is to accept much greater margins of tolerance for civilian harm.

This post has argued that such an approach is erroneous. It is based on a series of false assumptions about the nature of civilian harm in LSCOs, including the notion that mass civilian harm is somehow inevitable when large military forces engage one another or that allies will have no choice but to fight in city centres. It upends centuries of understanding an essential difference between, on the one hand, wars between sovereigns, pitting military force against military force, and, on the other, “fratricidal” civil wars which pit neighbour against neighbour and constrain local populations to take sides. Theorists of insurgency warfare, from Mao to Frank Kitson to Stanley McChrystal, have long identified the defining feature of civil insurgencies, in distinction to conventional warfare, as their proximity to local populations. This proximity carries profound implications across several dimensions: the garnering of support; the gathering of intelligence; the shaping of public opinion; and the risk of civilian harm.

Envisaging LSCOs as civilian-proximate operations only undertaken at greater scale and speed than COIN operations, with concomitantly greater rates of civilian harm, would be to make a category error. The COIN era saw intense debate around specific practices, including target verification, precision weaponeering, and managing the variables in the proportionality calculus. But the switch to planning for civilian protection in LSCOs must mean more than simply downgrading such practices, or alternatively, in the views of civilian protection advocates, scaling them up.

The starting point should rather be the assumption that LSCOs between peer or near-peer adversaries should result in less civilian harm compared to COIN operations, particularly if they can be conducted away from population centres. Where support can be won or coerced from the local population, an insurgent force with limited capability can embed itself among civilians. For a peer adversary in an inter-State conflict, this scenario is much less likely. The arts of camouflage may have come a long way, but few commanders would choose to embed and coordinate a major battlegroup among a civilian population. To suggest that in a LSCO against a peer adversary, fighting inside cities could always be avoided would be naïve. But it would be equally unrealistic to base LSCO planning on civilian-proximate scenarios drawn from COIN, but overlain with intensifiers for scale and speed.

Realistic civilian protection planning for LSCOs needs to pay more attention to other areas of LOAC: how the obligation to take constant care to spare civilians should be implemented across the conduct of military operations, including in the choice of battlespace; key challenges to the principle of distinction, including the automated use of proxy indicators for threats and the targeting of “war-sustaining” objects and critical dual-use infrastructure; and effective planning for the protection of, and facilitation of humanitarian assistance to, vulnerable civilian populations, including those under occupation.

There are indications that NATO allies may pick up the lead on civilian protection if the United States withdraws. At the last meeting of the International Contact Group on Civilian Harm Mitigation and Response, outside London in January 2026, the United States for the first time failed to attend. Non-governmental organisations such as Ceasefire Centre for Civilian Rights, Airwars, and PAX have presented detailed recommendations to the United Kingdom and Dutch governments on strengthening civilian protection in LSCOs .

Conclusion

While the prospect of a LSCO against a peer adversary remains (thankfully) at some distance, there is also a more immediate concern. As recent operations in Yemen and Somalia attest, it is highly likely that the United States and allies will continue to conduct COIN operations overseas for the foreseeable future. And as the attacks on Venezuela and Iran demonstrate, asymmetric operations may be conducted against States too.

Yet as Professor Naz Modirzadeh warns, “LSCO lawyering is not dependent on the actual outbreak of a great power war.” The legal arguments “are already shaping how LOAC is taught, operationalized, and applied.” The introduction of a new LSCO mindset into military lawyering carries the danger that, as operations continue to be conducted against insurgencies, embedded in the civilian population, or other asymmetric enemies, they will be conducted with a greatly expanded tolerance for incidental harm. That would be disastrous for civilian protection and would negate 20 years of hard-won lessons about how to run COIN operations.

***

Mark Lattimer is executive director of Ceasefire Centre for Civilian Rights.

The views expressed are those of the author, and do not necessarily reflect the official position of the United States Military Academy, Department of the Army, or Department of Defense. 

Articles of War is a forum for professionals to share opinions and cultivate ideas. Articles of War does not screen articles to fit a particular editorial agenda, nor endorse or advocate material that is published. Authorship does not indicate affiliation with Articles of War, the Lieber Institute, or the United States Military Academy West Point.

Photo credit: Marine Corps Sgt. Mitchell Johnson