[Editor’s Note: On Tuesday, the Department of Defense released an updated version of its Law of War Manual. Just Security is hosting an online “mini forum” to discuss the new changes to the Manual.]

On June 12, 2015, the United States Department of Defense issued its long-awaited Law of War Manual. The Manual seemed to take the alarming position that harm to human shields would not prohibit an attack under the proportionality rule, no matter how great the expected harm or how small the anticipated military advantage. This position seemed to apply to civilians who voluntarily choose to serve as human shields; to civilians who are involuntarily forced to serve as human shields; and to civilians who are involuntarily used as human shields without their knowledge or consent.

Ten days later, I wrote my first post for Just Security, criticizing the Manual’s position with respect to involuntary human shields. Many other scholars and practitioners voiced similar concerns, both publicly and privately. At the time, the focus on involuntary human shields seemed appropriate: the law was clear and the stakes were high. In mixed battle spaces like Mosul, Aleppo, and Sanaa, the Manual’s position on involuntary human shields would deny vital legal protection to many (perhaps most) civilians.

On Tuesday, the DoD issued an updated Manual. The Manual’s position on human shields has changed dramatically, though it requires careful parsing of multiple provisions scattered across multiple subsections. The new position seems to be as follows:

(i) Harm to human shields may render an attack unlawful under the proportionality rule. (16.4)
(ii) In some cases, voluntary human shields may directly participate in hostilities and thereby lose their protection under the proportionality rule. (12.3.4)
(iii) The responsibility of the defending force is a factor that may be considered in determining whether harm to voluntary human shields is excessive. (12.1.4)

Proposition (i) is clearly correct. In particular, involuntary human shields retain their ordinary protection under the proportionality rule. However, propositions (ii) and (iii) raise a number of questions.

First, the Manual does not explicitly state that involuntary human shields may not directly participate in hostilities and thereby lose their protection under the proportionality rule. Nor does the Manual explicitly state that the responsibility of the defending force is not a factor that may be considered in determining whether harm to involuntary human shields is excessive. In my view, such vitally important matters should not be left to negative implication.

Second, it is not clear whether the Manual means to say that

(a) civilians who voluntarily serve as human shields may also engage in other acts that constitute direct participation in hostilities;
or that
(b) voluntary human shielding may itself constitute direct participation in hostilities.

The former (a) seems obviously correct; the latter (b) is not obvious and, perhaps, not accurate.

For example, the ICRC takes the view that voluntary human shielding constitutes direct participation in hostilities only if it creates a physical obstacle to attacking a lawful target. Some experts take the broader view that voluntary human shielding also constitutes direct participation in hostilities if it creates a legal obstacle to attacking a lawful target, that is, if the presence of voluntary human shields would otherwise make attacking a lawful target disproportionate. In my forthcoming book, I take the narrow view that voluntary human shielding constitutes direct participation in hostilities only if it creates a physical obstacle to attacking lawful targets as an integral part of a military operation likely to directly cause serious harm.

Unfortunately, the relevant provision simply directs the reader to the Manual’s general discussion of direct participation in hostilities, which does not mention human shielding at all. This may be some evidence in favor of interpretation (a) over interpretation (b). Alternatively, since the Manual’s general discussion of direct participation in hostilities seems to contemplate only physical participation in hostilities, there is perhaps some reason to believe that the Manual favors the ICRC’s view, or one closer to mine, over the broader view of some experts.

Finally, we come to the claim that

5.12.1.4 . . . When the attacking force causes harms that are the responsibility of the defending force due to its use of voluntary human shields or due to the employment of civilian personnel in or on military objectives, the responsibility of the defending force is a factor that may be considered in determining whether such harm is excessive.

Put another way: harms caused by the attacking force to voluntary human shields are the responsibility of the defending force; therefore, it may not be excessive to cause some amount of harm to voluntary human shields even if causing the same amount of harm to ordinary civilians would be excessive in relation to the same military advantage.

In my view, this argument proceeds from an ambiguous premise to a highly dubious conclusion. As I argued in my original post:

Of course combatants who use human shields are responsible for the foreseeable harm they occasion. It hardly follows that combatants who kill human shields are not responsible for the foreseeable harm they inflict, particularly when the harm they expect is not justified by the military advantage they anticipate.

More precisely, it hardly follows that the attacking force is less responsible for the harm that it inflicts such that it is permitted to inflict more harm in pursuit of less military advantage. The responsibility of the defending force, as such, does not affect the permissibility of the attacking force’s decision to strike.

As I explained in a later article, “[i]n the context of war, attributing responsibility is not a zero-sum game.” With some modifications, what I wrote there seems applicable here:

No doubt, [defending forces have] acted wrongfully, even criminally, by using []voluntary shields. But now the choice is yours: to kill the[m] or to spare them. If you choose to kill the[m] then you cannot deny responsibility for your choice. The [defending forces are] responsible for [their] choice and you are responsible for yours. The[y are] wrong to use []voluntary shields and you would be wrong to kill the[m to achieve a military advantage too small to justify their deaths].

In sum, the Manual’s new position seems to rest on an unsound foundation.

Might the Manual’s new position stand on its own merits? After all, the United Kingdom’s Law of Armed Conflict Manual states:

Any violation by the enemy of this rule [prohibiting the use of human shields] would not relieve an attacker of his responsibility to take precautions to protect the civilians affected, but the enemy’s unlawful activity may be taken into account in considering whether the incidental loss or damage was proportionate to the military advantage expected.

Such passages of the UK Manual are often cited for the proposition that collateral harm to human shields may be discounted—though not entirely disregarded—in determining the legal proportionality of an attack.

However, in my view, the use of the passive voice, as well as the past tense, leaves the meaning of these passages somewhat unclear. These passages do not say that attackers may take a defender’s conduct into account in considering the legality of their own future attacks. Instead, these passages seem to address a third party considering the legality of an attacker’s past attacks. The distinction between ex ante conduct rules and ex post decision rules is often quite important. For example, it is one thing to tell a war crimes tribunal to acquit defendants who do not inflict clearly excessive harm on civilians; it is quite another thing to tell attacking forces that they are permitted to inflict collateral harm on civilians that is excessive but not clearly excessive.

Interestingly, the peculiar phrasing of the UK Manual tracks the following passage from the scholarly work of APV Rogers, the General Editor of the UK Manual:

Those carrying out attacks in such circumstances [including the use of human shields] are not relieved of their obligation to attack military objectives only and reduce incidental damage as much as possible, but in considering the rule of proportionality, any tribunal dealing with the matter would be obliged to weigh in the balance in favour of the attackers any such illegal activity by the defenders.

The similarity between the passages suggests that the parallel language of the UK Manual is addressed to tribunals considering past attacks and not to attackers considering future attacks. Viewed in this light, the passages invite additional questions. Do these passages mean that tribunals should consider an attack proportionate if the civilians harmed are human shields but disproportionate if they are not? Or do these passages mean only that tribunals should bear in mind that substantial harm to civilians may turn out, on closer inspection, to have been proportionate if the military targets they were used to shield were very important? It is hard to say.

The UK Manual is not a strong foundation for the view that attackers may discount harm to human shields in determining the proportionality of an attack. Certainly, the UK Manual does not offer any legal reasoning to support such a view. Nor is such legal reasoning easy to find. Many of us have looked.

In general terms, I agree with the Manual’s statement that “[i]n determining whether the expected incidental harm would be excessive, the totality of the circumstances must be considered. This holistic judgment should consider any relevant moral, legal, and military factors.” I also certainly agree that

[t[he evaluation of expected incidental harm in relation to expected military advantage intrinsically involves both professional military judgments as well as moral and ethical judgments evaluating the risks to human life (e.g., civilians at risk from the attack, friendly forces or civilians at risk if the attack is not taken). (5.12.3)

Indeed, my own view places such risks to human life at the very heart of proportionality analysis.

Ultimately, however, the proportionality rule compares only two variables: harm to civilians and military advantage. The fact that certain civilians are unlawfully used as human shields by defending forces does not affect the legal or moral weight of either variable. Nor does the standard of comparison—excessiveness or disproportionality—provide any obvious entry point for seemingly exogenous considerations such as the illegal conduct of defending forces. Finally, the proportionality rule considers only military advantages that are concrete and direct. Accordingly, a priori speculation that discounting harm to human shields may deter their future use has no place in proportionality analysis.

It is true that many voluntary human shields can easily avoid collateral harm. We can imagine circumstances in which this may seem like a morally relevant consideration. However, as Janina Dill reminds us, it is often impossible for attackers to know whether civilians are in or near lawful targets (i) voluntarily and (ii) with the specific intent to shield those targets from attack.

Some experts take the view that attackers should presume that civilians are not voluntary human shields absent decisive evidence to the contrary, but that given such decisive evidence attackers may discount collateral harm to such civilians. Since such decisive evidence is hardly ever available, the practical effect of this view would be that attackers must consider all civilians fully protected by the proportionality rule.

Importantly, this view is simply not available to the Manual, which continues to deny that persons should be considered protected civilians in cases of doubt, and still fails to even mention the customary rule that attackers must do everything feasible to verify that persons are not protected civilians before attacking or collaterally harming them.

Accordingly, we should reject the updated Manual’s apparent position that attackers may discount harm to voluntary human shields who do not directly participate in hostilities. Given other mistakes in the Manual, this position is too dangerous to leave in place.