What We Are Really Arguing About: A Reply to Professor Daniele on Incidental Civilian Harm

[Dr Aurel Sari is a Professor of Public International Law at the University of Exeter]

Over the past two years, the Journal of Conflict and Security Law hosted an exchange on a question which until that point had received little attention in the literature: what does it mean for civilian harm to be ‘incidental’ within the meaning of the proportionality rule? Professor Luigi Daniele opened the exchange with an article arguing that civilian harm qualifies as incidental only where it is the lesser, side consequence of an attack, rather than its prevalent outcome. I replied that this reading does not reflect the law, contradicts State practice and is operationally unsound. Professor Daniele has now responded with a two-part rejoinder on Opinio Juris (Part I; Part II).

I feel that a line-by-line response would not serve the readers of this blog. Those who wish to follow the detail can read Professor Daniele’s article and my reply. I would like to do three things instead. First, identify what I take to be the core of our disagreement. Second, assess where Professor Daniele’s rejoinder leaves the debate. Third, explain why this debate is worth having at all. Professor Daniele suggests that my objections to his article betray an indifference to civilian suffering. This is a charge that I do not accept.

What the disagreement is actually about

Strip away the rhetoric and our disagreement reduces to a single doctrinal question: what is the nature of an indiscriminate attack? Article 51(4) of Additional Protocol I identifies three types of indiscriminate attacks: those not directed at a specific military objective; those employing a method or means of combat that cannot be so directed; and attacks using methods or means whose effects cannot be limited as required by the Protocol. It is common ground in the literature and State practice that disproportionate attacks fall within the third type, Article 51(4)(c). My argument is that, in so far as it captures disproportionate attacks, Article 51(4)(c) prohibits attacks that are insufficiently discriminate, in the sense that they are properly directed at a military objective, but their effect is to cause excessive levels of incidental harm.

Professor Daniele rejects this. For him, the closing words of Article 51(4):

‘of a nature to strike military objectives and civilians or civilian objects without distinction’

denote ‘the absence of distinction, not a merely insufficient degree of it’. On that reading, Article 51(4) demands distinction in an absolute sense.

It is worth following where that premise leads, because its consequences are more radical than they first appear. Attacks must be directed against military objectives. If an attack directed against a military objective becomes indiscriminate the moment it inflicts harm on civilians or civilian objects otherwise than as a secondary effect, then an attack that harms a co-located civilian before it reaches the military objective, or simultaneously with it and through the same damage mechanism, would strike the two ‘without distinction’ and therefore be indiscriminate. Consider artillery ranging onto a target, where some rounds fall among civilians before the target is found; or a tank round whose lethal radius takes in both an enemy position and a civilian standing beside it. On Professor Daniele’s premise, both are indiscriminate. The only incidental harm left standing would be harm of a second-order kind, such as a secondary explosion that, having destroyed an armoured vehicle, then injures a bystander. Everything else would amount to a violation of distinction per se.

My article explains, at length, why this does not reflect the law. It is not borne out by the historical evolution of the rule from Hague Convention IX and the 1923 Hague Air Warfare Rules; nor by the travaux of Additional Protocol I; nor by national military doctrine and the practice reflected in collateral damage estimation methodology; nor by other instruments such as Amended Protocol II to the Conventional Weapons Convention; nor by the jurisprudence of the ad hoc tribunals and the International Criminal Court (ICC), which routinely treat harm to civilians co-located with a target, including immediate, direct, first-order harm, as incidental in nature and subject to proportionality. The law has never required the absolute separation that Professor Daniele’s reading presupposes.

It follows that harm to civilians and civilian objects co-located with a military objective can, in principle, be incidental. But this raises an evidential problem: how do we distinguish harm that is genuinely incidental from harm inflicted on co-located civilians deliberately or recklessly? Where civilians and a lawful target are affected at the same instant by the same blast, the effects alone will not tell us whether the attack was directed only at the lawful target or also at the civilians around it.

The Katanga judgment offers the answer. The Trial Chamber distinguished, as a matter of law, violence intentionally directed against civilians from harm incidental to an attack on a lawful target, and held that the requisite intent for an attack intentionally directed against civilians could be inferred from objective indicia, including the means and methods used, the number and status of the victims and the nature of the acts. The decisive question is whether the attack was, as a matter of fact, directed against the civilians. Because that question cannot be answered from the physical effects alone in cases where civilians are co-located with the target, it turns on the attacker’s state of mind: a belligerent who must anticipate and therefore calculate civilian harm will, in such cases, knowingly and willingly expose civilians to it. What separates incidental harm from a direct attack is that the harm is an undesired by-product, not a desired goal.

This is where Professor Daniele misreads me. I do not suggest that the line between incidental and direct harm turns on the attacker’s private, self-certified intentions, so that a commander could excuse committing a massacre by professing that he did not really desire it. My point, made expressly in the article and grounded in Katanga, is the opposite: whether the harm was truly treated as undesired is established by objective factors. The most telling of these is whether the attacker took precautions to avoid or minimize the civilian harm, as required by Article 57(2)(a)(ii) of Additional Protocol I. A party who took no feasible step to spare co-located civilians can hardly say that their deaths were unwanted.

Where the rejoinder leaves the debate

Much of Professor Daniele’s first post is devoted to the proposition that civilian harm which the law requires to be carefully calculated cannot sensibly be described as ‘fortuitous’ or ‘accidental’ and that neither of these words is a good synonym for ‘incidental’. On reflection, I think he is right. While ‘fortuitous’ may mean inadvertent or unintended, and it is this sense that I had in mind, it can also mean random, haphazard and unforeseen. Since the proportionality rule describes incidental civilian harm as something that is ‘expected’, Professor Daniele and myself are in complete agreement that it must be capable of being foreseen. The bottom line is that incidental harm is collateral to the effect that the attacker seeks to achieve, not that it comes as a surprise. However, none of this affects the substance of our disagreement.

The textual hook of Professor Daniele’s original thesis was the inference drawn from the area bombardment rule in Article 51(5)(a) of Additional Protocol I that an attack expected to affect a similar or greater number of civilians than military objectives must be indiscriminate. Part of my article was devoted to showing that Article 51(5)(a) does not support this inference: the numerical comparison in the rule runs between localities and areas, not between military objectives and civilians, and the rule cannot be stretched into a general principle. To his credit, Professor Daniele now accepts that the phrase ‘similar concentration’ in that provision ‘refers to cities, towns, and villages, not to numbers of lawful targets and protected persons’. That is a significant concession which removes the textual footing for the strongest form of his thesis.

In his article, Professor Daniele offered the hypothetical example of the:

‘bombing a high-rise tower building full of civilians because an enemy unit was located in its basement’.

I argued that this example would not involve an indiscriminate attack per se within the meaning of Article 51(4)(a) and (b) of Additional Protocol I, provided that the attack was directed at the enemy unit and that the methods and means of combat employed were capable of being directed in that way with some degree of accuracy. This is so because, as noted earlier, harm caused to civilians or civilian objects in the course of an attack against a military objective is not precluded from being incidental in nature by the fact that it affects those civilians or civilian objects first or affects them simultaneously to affecting the military objective being targeted. The bombing therefore may engage Article 51(4)(c) of Additional Protocol I if the anticipated incidental civilian harm was excessive. In his rejoinder, Professor Daniele rejects this analysis and suggests that the ‘limitations’ mentioned in Article 51(4)(c) include, ‘first and foremost’, distinction itself. That begs the question: if the limitations in Article 51(4)(c) merely incorporated those already set out in Article 51(4)(a) and (b), would that not make Article 51(4)(c) redundant? Of course, the reason why Article 51(4)(c) is not redundant is because, unlike Article 51(4)(a) and (b), it is not concerned with how methods or means of combat are directed, but with the question of whether their effects can be limited as required by Additional Protocol I. These are two distinct forms of distinction, the first is concerned with what is being targeted by an attack and the second is concerned with the effects of an attack. Professor Daniele continues to collapse that difference.

Professor Daniele also invokes the Bemba and Lubanga jurisprudence to argue that dolus directus in the second degree, that is virtual certainty, suffices for the war crime of directing attacks against civilians, so that my emphasis on ‘desire’ impermissibly raises the threshold to one of purpose. This conflates two distinct questions. The authorities he cites concern the fault required once we know what the attack was directed at. They do not tell us what it was directed at in the first place. My distinction between desired and undesired effects operates at that prior, target-selection stage which, as the Ntaganda Trial Chamber put it, is a matter of ‘selecting the intended target and deciding on the attack’. Professor Daniele refers to a passage in Katanga which declares that an intent to target civilians may be inferred where the damage to civilians is ‘so great’ that the perpetrator appears to have meant to cause that damage. That is precisely my point: the magnitude of civilian harm may constitute evidence, amongst other indicators, that the attacker did in fact desire to harm civilians and directed the attack against them. However, the magnitude of the civilian harm is not a freestanding arithmetic threshold that, by itself, converts collateral harm into a direct attack.

For these reasons, the rejoinder does not, in my view, make out its case. It abandons the textual argument that was critical to the original article, restates the circularity it needs to break and presses an international criminal law argument that provides no support.

What remains therefore is a disagreement about when proportionality applies. Professor Daniele suggests that proportionality does not apply at all to the example of the high-rise building. In my view, proportionality does apply, provided that the attack complies with Article 51(4)(a) and (b) of Additional Protocol I. However, my analysis does not stop here, a point that Professor Daniele fails to acknowledge. First, the attacking party is bound by Article 57(2)(a)(ii) of Additional Protocol I to take all feasible precautions in the choice of means and methods of attack with a view to avoiding, and in any event to minimizing, incidental civilian harm. It is never a case of simply ensuring that incidental civilian harm is not excessive: everything feasible must be done to avoid or at least minimize even non-excessive civilian harm. Second, a party that took no such steps cannot claim that the incidental civilian harm it inflicted was unintended or ‘undesired’. A failure to take feasible precautions does not automatically convert an attack into a deliberate attack on civilians, since a party may fall short of its precautionary duties without directing its attack against the civilian population. However, it does raise the question whether the belligerent concerned has limited the effects of the attack in a manner required by the Protocol and therefore brings the attack within the scope of Article 51(4)(c). Applied to Professor Daniele’s example, an attack that levels a tower ‘full of’ civilians to reach a handful of fighters in its basement will, once the precautionary duty and the excessiveness standard are applied, almost certainly be unlawful, both because it accepts excessive levels of civilian harm and because the failure to further minimize that harm, assuming this was feasible in the circumstances, renders it an indiscriminate attack within the meaning of Article 51(4)(c).

Professor Daniele charges me with ‘not once’ referring in my article to the reality of the Israeli attacks on the residential towers in Gaza. He is right: in that paper, I engaged with his example on a normative level, without relating it to the ongoing hostilities in Gaza and the systematic violations of the law of armed conflict committed there. I have written about the attacks on the Gaza tower blocks in a post elsewhere. As Professor Daniele will know, given that he cites the post in his article, I argued that the complete destruction of Gaza’s high-rise buildings raises questions about the proportionality of these attacks and whether Israel has complied with its precautionary duty to take all feasible steps to avoid or to minimize incidental civilian harm.

Why the debate matters

This brings us to a broader point. Professor Daniele ends his rejoinder by expressing doubt that:

‘any sector of our discipline should, in this moment, be preoccupied with the presumed dangers of excessively protecting civilians’,

suggesting that to worry about rendering military operations ‘all but impossible’ is to be ‘divorced from current war realities’. It is difficult not to read this as suggesting that my article is, in effect, lending cover to those who read the law of armed conflict too permissively or to those who would prefer to dispense with it altogether. I reject that suggestion and think that it mistakes what is at stake here.

We agree on the essentials: the law of armed conflict must be respected. There is a grave and widening gap between the obligations it imposes and the conduct of contemporary hostilities. Our disagreement is about what to do in response. Professor Daniele’s answer is to re-read the proportionality rule in favour of an ‘unapologetic protection of civilians’, even at the risk of upsetting the balance between military necessity and humanity that Additional Protocol I seeks to maintain.

I think that this approach is misguided. It conflates the law as it is with the law as we might wish it to be. The line between lex lata and lex ferenda may be fluid, but it is not infinitely so. A re-reading of the rules that is not supported by the text, the travaux, State practice and jurisprudence cannot be sustained merely by the urgency of the cause. Even if it could, Professor Daniele’s re-reading is operationally unsustainable: pressed to its conclusion, the thesis would characterize an extremely wide range of attacks as indiscriminate per se. As I have argued in my article, a rule so far removed from the conditions of battle has little prospect of commanding compliance, as the International Committee of the Red Cross warned in 1975. A rule that is ignored protects no one.

The answer is not a more permissive law. It is to recognize that the proportionality rule serves a limited function: it imposes a maximum ceiling on incidental harm. We need to look elsewhere to reinforce the protective purpose of the law of armed conflict. In my article, I argued that a failure to comply with Article 57(2)(a)(ii) by not employing means or methods that would avoid or minimize incidental harm where this was feasible should itself render an attack indiscriminate under Article 51(4)(c). That reading expands civilian protection at exactly the point where it can be tested against the conduct of belligerents, but without asking armed forces to accept a rule that is operationally unworkable. Our disagreement therefore does not pit a progressive, civilian-friendly reading of the law against a morally bankrupt, military-friendly interpretation, as Professor Daniele seems to imply, but involves a disagreement about how best to ensure civilian protection in a way that remains rooted in the law as it exists and in a manner that is capable of being implemented.

Photo attribution: by Levi Meir Clancy on Unsplash