This article is part of our joint symposium with EJIL: Talk! on Chatham House’s “Proportionality in the Conduct of Hostilities” Report.

Chatham House’s research paper by Emanuela-Chiara Gillard, Proportionality in the Conduct of Hostilities, provides an outstanding analysis of interesting, important, and challenging legal questions. Nevertheless, I was asked to find a nit to pick, and so I have. At one point, the Report states that

In certain situations it is evident that the expected incidental harm will be excessive in relation to the military advantage, and in others it is evident that it will not be. To use some examples referred to in military manuals, bombing an isolated fuel tanker in the middle of a densely populated city would be excessive, while an airstrike against an ammunition depot beside a farmer ploughing a field would not be. ‘Excessive’ is a wide but not indeterminate standard. (para 81)

This passage seems entirely correct, yet invites the question: What should a commander do when she confronts borderline cases, in which the expected civilian losses are neither ‘evidently’ excessive nor ‘evidently’ non-excessive in relation to the military advantage anticipated? Is she free to attack, since she cannot assess with confidence that the expected losses would be excessive? Or must she refrain from attack, since she cannot assess with confidence that the expected losses would not be excessive?

Rather than directly address this question, the Report states that

Provided they do what is required to collect information on which to base their assessment, and conduct the assessment in good faith and in a manner that is reasonable, belligerents have ‘a fairly broad margin of judgment’, in the words of the ICRC Commentary to the Additional Protocols of 1977, to determine whether the expected incidental harm would be excessive. (para 81)

Now, to tell a commander that she enjoys ‘a fairly broad margin of judgment’ is not yet to tell her how best to exercise that margin of judgment. Similarly, telling a commander to conduct herself reasonably and in good faith does not tell her anything that she does not already know (or so one should hope).

For its part, the ICRC Commentary states that,

[o]f course, the disproportion between losses and damages caused and the military advantages anticipated raises a delicate problem; in some situations there will be no room for doubt, while in other situations there may be reason for hesitation. In such situations the interests of the civilian population should prevail, as stated above. (para 1979)

On this view, commanders should simply refrain from attack in borderline cases. The ICRC Commentary appears to ground this view on the general precautions rule (codified in Additional Protocol I, art. 57), stating that

Other more complex situations may pose difficult problems for those responsible [that is, for commanders]. The golden rule to be followed in such cases is that contained in the first paragraph [of API art. 57], i.e., the duty to spare civilians and civilian objects in the conduct of military operations. (para 2215)

In other words, the duty to take constant care to spare civilians excessive harm (among other things) supplements the rule of proportionality by requiring restraint in borderline cases. Commanders may not flip a coin or resolve serious doubts in favor of their own military advantage.

The Report underscores the duty of commanders to do everything feasible to verify that proposed attacks will not violate the proportionality rule (see herehere, and here). This duty seems to imply that a commander who tries but fails to verify conformity with the proportionality rule must refrain from attack. Among other things, it would seem to weaken a commander’s incentives to do ‘everything feasible’ if her failure to verify were instead to be rewarded with freedom to attack.

This view also leads to results that are logical rather than unreasonable or absurd. Assume the following scenario:

Attackers verify that a building is a military objective. Attacking the building will almost certainly kill a number of people nearby.

Now consider two variations:

I) Attackers suspect that the people nearby are combatants, but remain in serious doubt. If the people nearby are civilians, then their expected deaths would be excessive in relation the military advantage anticipated.

II) Attackers verify that the persons nearby are civilian. However, their expected deaths would be neither clearly excessive nor clearly non-excessive in relation to the military advantage anticipated.

In variation I, attackers must presume that the people nearby are civilian (under API 50(1)) and therefore refrain from attack. To ignore their serious doubts and attack would seriously risk violating the proportionality rule. What about variation II? On the view we are exploring, attackers must again refrain from attack, for the same reason: to avoid serious risk of violating the proportionality rule.

There is no obvious reason to refrain from attack in the face of grave factual doubts but proceed with attacks in the face of equally grave evaluative doubts. The risk of error—excessive harm to civilians—is the same either way. It would be particularly illogical to preclude an attack which may not kill any civilians (variation I) but permit an attack which will definitely kill a number of civilians (variation II).

Finally, on a different approach, which I explore in a forthcoming article, international law requires parties to armed conflict to respect and ensure respect for the proportionality rule (among others). Parties can best fulfill this requirement by instructing their armed forces to refrain from attack in borderline cases. Evidently, such an instruction will ensure greater respect for the proportionality rule than the contrary instruction­—‘in borderline cases, fire at will’—or no instruction at all. Moreover, borderline cases of excessiveness may invite motivated reasoning (or even out-group bias), as commanders may consciously or unconsciously assign too little weight to the lives, bodies, and belongings of civilians (particularly civilians supportive of the adversary) and too much weight to their own military advantage. Instructing commanders to refrain from attack in borderline cases may help counteract such psychological dynamics.

Toward the end of the Report, it states that “the obligation to cancel or suspend an attack only arises if it becomes ‘apparent’ that the [proportionality] rule would be violated. The word ‘apparent’ sets a high threshold.” This passage might suggest that, once an attack commences, attackers may carry it out even if substantial doubts arise regarding its conformity with the proportionality rule, so long as it does not become ‘apparent’ that the attack would violate the rule. But things are not quite so simple.

The Report continues by stating that “[i]n circumstances that fall below this threshold, those executing an attack who become aware of facts that they believe could affect the proportionality assessment must report this to those coordinating the operation, so that such information can be taken into account.” It later adds that “[w]hen those executing attacks become aware of information that may affect the proportionality assessment and report it back to those who are coordinating the operation, the result should be that the attack will be interrupted.”

Of course, ‘interrupted’ is just another word for ‘suspended’. Presumably, if the factual doubts raised in such cases cannot be resolved, then the interruption or suspension will result in cancellation. So, effectively, an attack must be suspended or even cancelled if serious factual doubts arise during its execution. Not such a high threshold after all.

Notice, however, that the Report calls for the ‘interruption’ of attacks in progress only if those executing the attack encounter new facts or information. If their doubts are evaluative, regarding the excessiveness of the expected harm, rather than factual, regarding the extent of expected harm or anticipated advantage, then it seems that they may ignore these evaluative doubts. This is the distinction that, I suggested above, should make no difference.

In an earlier passage, the Report states that the obligation to cancel or suspend attacks if it becomes apparent that they would violate the proportionality rule

neither entitles nor obliges those conducting an attack to carry out a new proportionality assessment on the basis of the same facts as those who planned or decided it – i.e. replacing the original assessment with their own assessment of whether the expected incidental harm would be excessive. It does require them to cancel or suspend the attack if, on the basis of new facts or for any other reason, it becomes apparent to them that it would violate the rule.

While the phrase “or for any other reason” creates some ambiguity, the Report seems to take the view that international law allows combatants to defer to the evaluative judgments of their superiors unless the factual basis for those evaluative judgments is called into question. This view seems hard to accept. Those carrying out attacks remain bound by the proportionality rule itself, as well as by the precautionary obligation to take constant care to spare civilians, and these requirements do not suggest that factual assessment must continue but evaluative judgment may be suspended once an attack is underway.

Ideally, a combatant’s superiors would be so legally, ethically, and professionally reliable that she would best conform to the proportionality rule by deferring to their evaluative judgment, never doubting theirs and never exercising her own. She could do everything feasible to verify that her attacks conform to the proportionality rule simply by following their orders, since any independent efforts would be wasted. Unfortunately, combatants may confront a very different reality, placed under superiors undeserving of such unquestioning obedience. The law must speak to these combatants, too.

For one thing, combatants may have reason to doubt that their superiors are interpreting the proportionality rule correctly. For example, their superiors may take the view that the proportionality rule excludes civilians forced to serve as involuntary shields, while the combatants know better. Second, combatants may have reason to doubt that their superiors are applying the proportionality rule in good faith. Superiors, too, may be prone to undervalue civilians and overvalue military advantage. Finally, the expected civilian losses may be so extensive, and the military advantage discernable to the combatant so minor, that the combatant may not carry out the attack—even if she cannot identify any specific mistake of fact by her superiors—unless and until some great undisclosed military advantage is revealed to her. When the stakes are so high, she may not blindly trust her superiors.

To sum up, I have argued that combatants may not carry out attacks in those borderline cases in which the expected civilian losses are neither clearly excessive nor clearly non-excessive in relation to the military advantage anticipated. This prohibition may be imposed by international law directly, within the precautions framework, or indirectly, as a means to ensure respect for the proportionality rule. This prohibition applies to all combatants, at all stages of attack, irrespective of whether the doubts in question are factual or evaluative or whether the combatant is planning, ordering, or carrying out the attack. Of course, it is possible that I am mistaken. But I doubt it.

IMAGE: Maintenance crews aboard the USS John C. Stennis Naval aircraft carrier ready a F/A-18 Hornet for takeoff. (Photo by Marco Garcia/Getty Images)