Gregory S. McNeal of Pepperdine University School of Law writes in with the following contribution to our ongoing discussion of Newtown, drones, targeting, and accidental versus intentional killing of children---one that responds to Benjamin Farley's earlier point that collateral damage is not unexpected or entirely accidental:
I thought I’d weigh in on your discussion regarding accidental deaths and targeting. First, I want to express my disagreement with the paragraph you submitted from Benjamin Farley. Given that it's only a paragraph I may be reading too much into it, but Farley writes in fairly definitive terms that I think require an answer. He states without equivocation "[d]eaths due to collateral damage are not unexpected or absent intent." And he gets to that point by skipping straight to the law governing proportionality. However, if we are to understand the possible causes of accidental deaths, I think its best that we talk through more than just proportionality.
Let’s start by briefly noting that international law requires that attackers refrain from mounting indiscriminate attacks, this is a customary law norm and is also articulated in Article 51(4) of AP I. Moving on, the general principles of targeting that I think are most relevant to our discussion of accidental deaths are (1) the principle of distinction, (2) precautions, and (3) proportionality. First, (1) distinction requires us to look to Article 48 of AP I which prohibits direct attacks on civilians or civilian objects. The basic rule expressed is that attackers must "distinguish between the civilian population and combatants and between civilian objects and military objectives and accordingly direct their operations only against military objectives." (This prohibition is also discussed in 51(2) and 52(2) whereas 51(3) renders civilians targetable “for such time as they directly participate in hostilities.” (DPH)). Article 57 (discussed below) also states a precautionary rule of “constant care” and in 57(2)(a)(i) it requires attackers “do everything feasible to verify that objectives to be attacked are neither civilians nor civilian objects . . .” Feasible means this is a duty that must take into account all circumstances ruling at the time, certainty is not a requirement as that would be impossible to comply with (but see the API Commentary, paragraph 2195). In any case, at this threshold stage a commander will make a determination based on the available intelligence as to whether harm to civilians is expected. Thus, one possible source of accidental killings that we shouldn't neglect is mistakes made while attempting to comply with the principle of distinction---for example targeting a group of persons believed to be directly participating in hostilities who turn out to be civilians not DPH’ing. That collateral damage would fall outside of Farley’s unequivocal “deaths…[that] are not unexpected or absent intent” framework, and would certainly count as accidental.
Moving on, even if an attacker has identified that harm to civilians is likely to result from an attack, our analysis doesn’t end. The next step for that commander would be to (2) assess feasible precautions (in U.S. policy this is referred to as collateral damage minimization). Precautions are codified in Article 57 of AP I, which requires the attacker to "take all feasible precautions in the choice of means and methods of attack with a view to avoiding, and in any event minimizing, incidental loss of civilian life, injury to civilians and damage to civilian objects…" This is somewhat tangential to Farley’s points, but it’s relevant to the broader discussion about accidental civilian casualties. It’s possible that an attacker identifies possible collateral damage or harm to civilians, takes all feasible precautions to avoid that harm, yet the operation nonetheless accidentally inflicts collateral damage. For example, the attacker identifies a lawful target inside the western room of a compound. In the eastern room of the compound are a group of civilians. The attacker acts reasonably and designates an aim-point at the westernmost wall of the compound and uses a munition likely to kill the target yet believed to be small enough to prevent harm to the civilians in the eastern room. The attacker is mistaken in his calculations, and accidental harm to civilians occurs. This would be accidental collateral damage. Also note that no harm was anticipated in this scenario because precautions eliminated the anticipated harm, therefore we never conducted a proportionality analysis.
Thus, the problem with a proportionality only analysis like Farley’s, and the problem with his correction of [Ben], is that his correction skips distinction and precautions, jumping right to proportionality. The proportionality rule and the distinction rule deal with different concepts, proportionality deals with whether to attack a military objective, not whether an object is in fact a military objective. So let’s turn now to Farley’s proportionality argument: "Deaths due to collateral damage are not unexpected or absent intent; in fact for a proper proportionality analysis, a commander must assess the reasonably likely number of civilians who will be killed in a strike and balance that against the military advantage gained through the strike. . . . It is thus essential that the commander be able to anticipate these deaths."
The middle part of this paragraph (underlined) is an adequate albeit incomplete description of proportionality. However, the first part is troubling insofar as it suggests that all collateral damage is expected, intentional and was deemed proportional. As the examples above illustrate, not all collateral damage is expected and it certainly isn’t all intentional. Moreover, as I pointed out here at Lawfare, in practice, proportionality balancing was not the primary explanatory variable in at least one category of strikes involving harm to civilians; in fact a failure to positively identify a target (part of distinction) was oftentimes the cause of accidental killings of civilians. In other words, mistakes do happen and some of those mistakes don't have anything to do with a proportionality analysis. As discussed above, distinction is the point in a targeting analysis when an attacker assesses whether the object of an attack is a lawful target. If after the distinction analysis no civilian deaths are anticipated, no precautions are necessary, and there is no proportionality analysis to conduct. As Gary Solis notes in The Law of Armed Conflict, "If an enemy sniper is spotted in a remote and isolated desert hiding spot, he could be lawfully targeted and killed with a 2,000 pound bomb or a B-52 bomber's load of 2,000 pound bombs. It would be a gross waste of munitions, but there would be no proportionality issue because there are no civilians or civilian objects with which to be concerned."
So when Farley writes, "That said, it may be appropriate to describe those deaths that exceed this value—the deaths of civilians who could not reasonably have been anticipated—as accidental," he shows how conflating concepts and analyzing out of sequence can lead to a confusing analysis. Standing alone, the second part of his sentence (underlined) is correct; it is certainly the case that unanticipated harm to civilians would be deemed accidental. However that is in no way dependent upon the first part of his sentence, where he makes reference to proportionality balancing. Not anticipating the deaths of civilians would be a distinction problem, not a proportionality problem. The takeaway here is that the decision as to whether something is a military objective (distinction), the precautions taken, and an analysis of proportionality are sequential. Skipping straight to proportionality risks confusion on a blog (although commentators do it all the time). In real life, it risks accepting a targeting decision that fails to minimize collateral damage. As Ian Henderson notes, taking precautions, “…comes before the obligation not to cause excessive collateral damage (the proportionality principle)…[i]n other words, there is a requirement to minimize collateral damage and not merely to cause no more than proportional collateral damage.”
Again, what seems to be the problem here is that Farley assumes a proportionality analysis is appropriate even if no harm to civilians is expected. This does not comport with Article 51.5(b) which enjoins "an attack which may be expected to cause incidental loss of civilian life, injury to civilians, damage to civilian objects, or a combination thereof, which would be excessive in relation to the concrete and direct military advantage expected." (emphasis mine) Absent an expectation of harm to civilians, one never gets to the "would be excessive in relation to" clause because if no collateral damage is expected, how could zero be excessive? In other words, there's simply no proportionality balancing to conduct if the attacker does not anticipate harm to civilians in the first place. Mike Schmitt helpfully explains the differences in the law at play here:
[distinction] seeks to preclude attacks in which protected persons or objects are themselves targets, or where the attack is made with culpable disregard for the civilian consequences, proportionality operates in scenarios in which incidental injury and collateral damage are the foreseeable, albeit undesired, result of attack on a legitimate target…
With the first tier of discrimination analysis [distinction], the question is: 'May I lawfully target an object or person?' With proportionality, an additional query must occur: Even if I conclude that targeting the person or object is unlawful, may I nevertheless knowingly cause him or it injury or damage in my attack on a legitimate objective?
For those who are interested, these sequential targeting concepts can be illustrated graphically and the best illustration appears on page 185 of Corn et.al.’s The Law of Armed Conflict: An Operational Approach.
Having set forth the legal analysis as I see it, let’s briefly address the original subject of the debate. Where along the way might we find accidental deaths occurring? For clarity I thought it might be helpful to list some of the ways accidental deaths might occur. I hinted at a few of them in my discussion above, but let’s list a few more possibilities. This is not a comprehensive list, but I think it is a list of plausible scenarios where accidental deaths may occur. For what it's worth, based on one CENTCOM data set I gathered (planned strikes) these scenarios or variants of them would account for 92 percent of civilian casualties, 70 percent from getting the wrong person and 22 percent due to errant munitions. So here is a list of possible causes of accidental killings:
- Intelligence indicates that Anwar al-Wittes (a lawful target) will be alone on a deserted road in his SUV when in reality four members of his family are accompanying him.
- Intelligence indicates that Anwar al-Wittes will be alone on a deserted road in his SUV, but it's not Anwar al-Wittes it's Ben Wittes.
- Intelligence indicates that Anwar al-Wittes will be alone on a deserted road but as the strike is occurring an unexpected and unanticipated group of civilians enter the blast radius of the weapon.
- Intelligence failed to indicate that Anwar al-Wittes' SUV was laden with explosives, causing a secondary explosion that could not have been anticipated.
- After the initial strike on Anwar al-Wittes the attacker cannot reliably determine he has been neutralized, re-strikes the target but fails to account for the unexpected aid workers rushing to his rescue. (kinda like 3).
- An electronic tagging device was meant to be placed on Anwar al-Wittes' blue SUV by a reliable person, but is instead placed on the wrong blue SUV and the munition hits the right target, improperly identified (similar to 2).
- The munition is inaccurate (bad guidance system, bad laser designation, fin falls off, incorrect fusing, etc.).
- Anwar al-Wittes is in a building, however the attacker is unable to measure the force necessary to ensure they kill him while minimizing harm to collateral structures and persons. They reasonably assume (based on ISR, and local building standards) that the building is brick, when really it is thin corrugated metal. They use a small munition, but the walls of the building are too thin to stop the blast, debris, and fragmentation and civilians nearby are killed who wouldn't have died if the building were brick.
- Other variations of 7 and 8 involving inaccurate targeting due to weather or atmospheric conditions and unexpected blast, debris, and fragmentation patterns due to terrain, weather, vegetation, buildings, etc.
- Many of the above apply to so called “signature strikes” but with the added problem that mistakes are easier to make when merely identifying an individual as directly participating in hostilities (DPH) (tough to gauge from an ISR platform alone). It becomes even more
- error prone if the attacker is using targeting criteria other than DPH such as membership alone.All of this is a long way of saying that one should be careful about assuming proportionality balancing, not mistakes, is the driving cause of collateral damage in airstrikes. Collateral damage can in fact be unexpected and without intent. To suggest otherwise and attribute all collateral damage to proportionality balancing is a frequent error made by commentators, but it’s one that needs to be addressed when it surfaces.