Yesterday found me in Israel listening—not for the first time—to a briefing on Israel Defense Forces (IDF) targeting practices during Operation Protective Edge, the 2014 military operation against Hamas in Gaza. I can’t go into the details on the briefing, as it was not on the record, but for those interested in a detailed account of the subject, Section VI of this report from the Israeli Foreign Ministry covers some of the same ground. For present purposes, suffice it to say that the discussion covered the Israeli military’s efforts to avoid civilian casualties in the densest of urban conflict environments against an opponent force that actively uses human shields and both hides among and launches attacks from civilian protected facilities.
At the end of the discussion, I had two brief and radically divergent conversations: Adam Klein of the Center for a New American Security marveled at the extreme disparity between Israeli efforts to avoid civilian harm and the international condemnation of Israeli targeting. On the other hand, Susan Hennessey of Lawfare commented to me that she found the briefing less persuasive. Even accepting Israeli casualty numbers, she observed, the proportion of civilian death was outrageous. One can talk all day about one’s procedures, she said, but if the actual outcome is fundamentally not proportional, then the process itself cannot responsibly be defended by principles of proportionality. The processes in question may constitute a narrow defense against allegations of war crimes, but they don’t present a winning argument more broadly.
The combination of the briefing itself and the reactions prompt the following thoughts on the nature of proportionality—and the incredibly diverse work we are making the concept do for us, both legally and morally. The following is exploratory and fairly tentative, and I welcome thoughts in response.
The Israeli targeting procedures in question are unquestionably impressive. They involve an elaborate series of legal reviews for pre-planned operations and targeting, a system of warnings to civilians (often using cell phone calls) to evacuate targeted facilities, the use of real-time surveillance from drones in many cases, and even the use of so-called “knocks on the roof”—the explosion of small, relatively harmless munitions on buildings to give people a last chance to get out. No other military, including ours, takes such measures as a matter of routine. And they are all designed to make sure that individual operations comply with the principles of distinction and proportionality and minimize civilian harm.
Think of these procedures as the first bit of work we are asking a concept of proportionality to do for us. And it’s doing legal work. This concept of proportionality has a few notable features. It is, first off, incident specific. That is, it doesn’t look at aggregate casualty numbers or damage and ask whether they are in some gestalt sense proportional. It looks, rather, at each specific targeting incident and asks whether the commander in that case used proportional force. Its analysis is also prospective. It doesn’t ask whether the results were good, bad, or tragic; it doesn’t ask either whether the military gains obtained in fact justified the use of force in retrospect. It asks, rather, whether the use of force is proportional from the point of a view of a commander with the information available to him at the time.
In this concept of proportionality—and it is worth noting that this is the sense of proportionality that the LOAC demands—the IDF’s procedures and practices are hard to argue with. I do not mean here that every single Israeli use of force in Gaza during Operation Protective Edge survives this proportionality inquiry. (Some are still being investigated.) I mean, rather, that one cannot reasonably accuse Israel of failing to make genuine efforts both to ensure that uses of force of are proportional and to investigate allegations that they were not.
The trouble is that, at least colloquially, we use the terms “proportionality” and “proportional” to incorporate several other ideas too. These are not ideas that sound in the LOAC. But they are ideas that deeply condition our sense of whether military actions are reasonable. Consider several other proportionalities; I count at least four important ones.
First, there’s proportionality of casualties between the sides. In Operation Protective Edge, according to Israel figures, 2,125 Palestinians died, and—on the other side—72 Israelis (six civilians and 66 soldiers) died. This disparity, a difference of a factor of 30, cries disproportionality to a great many observers. But notice how far conceptually from the LOAC analysis it is. It is an aggregate analysis, after all. It looks not at specific incidents but at the overall damage inflicted on each side in the conflict. What’s more, it’s a retrospective, not a prospective, analysis. It’s asking the proportionality question not from the point of view of the decision-maker who has to decide what sort of force to use but from the point of view of the damage assessment later. What’s more, in the specific case of Gaza conflicts, it is also distorted by Israel’s particularly strong civilian defense systems against Gaza rockets. If it weren’t for the Iron Dome missile defense system, after all, the aggregate casualty numbers on the Israeli side would be far higher than they are. And it’s a little perverse to argue that strong civilian defense renders otherwise appropriate offensive operations disproportionate because one side isn’t losing enough people.
Another form of non-LOAC proportionality is what we might call incident-specific retroactivity. For example, a commander fires under circumstances that may be prospectively reasonable but nonetheless end in disaster, and people look at the incident after the fact and declare the use of force disproportionate based on the results. This sort of thing happened a lot in Operation Protective Edge. (It happens a lot in US operations like drone strikes too.) And these incidents generate particular anger when Israeli review mechanisms then turn around and, analyzing them from the point of view of the prospective commander, declare them legally unproblematic. For example, one of the most frequently cited incidents in Operation Protective Edge took place on July 16, 2014, when Israeli forces killed four children on a Gaza beach. Yet after reviewing the incident the Israeli Military Advocate General (MAG) issued a statement closing the matter, saying:
After reviewing the investigation's findings, the MAG found that the attack process in question accorded with Israeli domestic law and international law requirements. The decision to attack was taken by the competent authorities, and the attack was aimed at figures who were understood to be militants from Hamas's Naval Forces, who had gathered in order to prepare to carry out military activities against the IDF. At the time that the decision was made, the attack was not, according to the assessment of the operational entities, expected to result in any collateral damage to civilians or to civilian property. Moreover, the attack was carried out while undertaking several precautionary measures, which aimed to prevent any harm to civilians. Such measures included, inter alia, the choice of a munition which was not expected to cause any harm to civilians, and the deployment of real time visual surveillance. The MAG found that the professional discretion exercised by all the commanders involved in the incident had not been unreasonable under the circumstances. However, it became clear after the fact that the identification of the figures as militants from Hamas's Naval Forces, was in error. Nonetheless, the tragic outcome of the incident does not affect the legality of the attack ex post facto.
A third distinct form of proportionality analysis is the one to which Susan Hennessey was responding: the aggregate ratio of civilians versus military casualties resulting on one side. According to Israeli figures, of the Palestinians killed in Operation Protective Edge, 936 were militants, 761 were civilians, and 428 involved persons whose status is undetermined. UN and Palestinian figures make the picture look far worse, but let’s take the Israeli figures at face value for purposes of argument even if one does not believe—as I do—that they are likely closer to the truth. Assuming they are correct, between 44 percent and 64 percent of those killed by the Israeli side were militants—and thus between 36 percent and 56 percent were civilians. Again, this is a retrospective analysis and an aggregate one. It’s altogether unrelated to the demands of LOAC proportionality. On the other hand, think for a moment like a newspaper editor instead of a lawyer. Is any editor going to correct a reporter who writes a sentence like, say, “The civilian casualties in Operation Protective Edge, which ranged somewhere between 36 and 56 percent of those killed by Israeli forces, have been criticized as disproportionate by many observers”? I wouldn’t.
Still a fourth type of proportionality is the proportion of power between the parties. That is, it seems to many people inherently disproportionate for a wealthy country with advanced weaponry, drones, and highly-sophisticated surveillance capacity to do battle with a desperately poor population armed with light weapons and homemade mortars and rockets.
None of these proportionalities are the ones that the LOAC actually requires. But it seems to me a big mistake to ignore them simply because they are not the proportionality to which the laws of war specifically demand attention. The reason is that each reflects some genuine moral instinct that conditions the emotional and intellectual responses to a conflict by huge numbers of people. Some of them may not be remediable. Are you really going to stop killing the enemy because they are not killing enough of your side, for example? And some of them may fail on close moral inspection. For example, if the weaker side launches hostilities, is the stronger side obliged not to defend itself? (The ICJ has actually come close to saying this in denying that Israel has a right of self-defense against Palestinian terrorism.)
But some of these visions of proportionality, in my opinion anyway, require serious moral attention. In particular, I don’t think we can blithely brush aside the relative rate of civilian versus military casualties simply by arguing that each specific use of force followed defensible, even laudable, policies aimed at civilian protection. Imagine, for example, that Israel had used the same systems as it did and 95 percent of those killed had been civilians. I don’t think anyone would argue that that would be acceptable. What about 80 percent? There’s some threshold of relative civilian harm above which we would have to conclude that processes producing such a level of civilian harm must inherently be deficient and that the LOAC—to the extent its vision of proportionality does not reflect this—is itself too lenient.
I don’t know what that number is, and I’m confident that the low end of Israeli civilian casualty estimates (36 percent) is well below it. On the other hand, if we assume that all of those the Israelis describe as “undetermined” were, in fact, civilians, and that the civilian casualty rate is thus actually 56 percent, the question becomes a lot harder.
There’s a final complicating moral factor on this point: The rate of civilian death in this conflict is largely a function of Hamas’s intentional decision to hide among civilians and use civilian protected buildings and facilities (mosques, hospitals, ambulances, UN facilities, schools, etc.) for military purposes. That fact clearly changes the acceptable proportion of aggregate civilian-to-military casualties. After all, in a conventional military-to-military struggle on an open battlefield, we would not tolerate anything like even the low-end rate of civilian casualties reported in Protective Edge. Indeed, even a tiny percentage of civilian casualties might shock the conscience under such circumstances.
The trouble is that while we have a moral instinct towards this form of proportionality and also a moral instinct that the behavior of the enemy to some degree loosens the demands of this sort of proportionality, I at least have no clear numerical sense of what this sort of proportionality looks like. That is, what’s an acceptable rate of civilian casualties in a conflict like this one? And what’s the rate that should make people assume that something must be wrong?
I have no strong instinct on this point—except perhaps one: the delta here matters. While things are quiet now along the Gaza border, there will be another round of conflict there. It seems to me highly significant whether the civilian-to-militant casualty ration in Israeli operations rises, falls, or stays the same. That is, there is probably no numerically correct answer to the question of what the rate should be. But we know within a 20 percent range what it has been as this particular military with its particular targeting procedures and compliance systems confronts this particular opponent with its particular tactics. And that gives analysts a useful baseline against which to compare IDF performance in the future.