I have received several interesting responses to my post earlier this week on different types of proportionality and Israel's 2014 Operation Protective Edge in Gaza. Two, in particular, offer useful additional information I want to share with readers.
First, a correspondent made me aware of an important appendix to the Israeli government report to which I linked, a standalone section on the topic of Palestinian fatality figures. Beyond alleging deficiencies in the statistics and methodologies employed by other actors, and alleging efforts by Hamas to fudge or fabricate the numbers, the appendix gives important information on how the IDF measures civilian casualties and some of the standards it used to arrive at the numbers I reported in my post: 936 militants killed, along with 761 civilians, and 428 persons whose status is "undetermined."
The appendix, as a preliminary matter, offers some additional information on the combatant casualties: "Out of the number of militants, 631 (67% of the militants killed) were affiliated with Hamas, 201 (22% of the militants killed) were affiliated with Palestinian Islamic Jihad, and 104 (11% of the militants killed) were affiliated with other terror organisations."
The IDF categorizes those it kills as civilians, the document goes on, either because "there was no indication they were involved in hostilities or because they were assumed to be uninvolved based upon their age and gender." This number includes "includes 369 children under the age of 15 (16% of total fatalities), 284 women (13% of total fatalities), and 108 men (5% of total fatalities)."
Note here that civilian is the default category. Classifying a killed person as a civilian requires no evidence. Rather, it seems to be what the IDF does when it cannot show that the person in question was either a combatant or a civilian directly participating in hostilities--and when the person was either a woman or a child.
This is actually a conservative approach. It's pretty easy to imagine situations, after all, in which someone died as a combatant but left "no indication" of that fact for investigators to notice. Nor is Hamas shy about boasting that it has sometimes used women for combat functions (though this isn't its norm):
— Al Qassam Brigades (@qassamsms) November 23, 2015
Nor is it coy, for that matter, about training children—including female children—for combat. It is thus at least possible that Israel is undercounting combatants—either because some of the women or children killed were engaged in activity that made them legitimate targets or because some men may have been combatants but "there was no indication" available of that fact.
More importantly, the appendix also offers an important clarification on the "undetermined" category: "the IDF is still trying to make an accurate determination as to whether an additional 428 males between the ages of 16-50 (20% of total fatalities and almost all of the unclassified fatalities) were involved or uninvolved in the hostilities. Based on the IDF’s past experience, it is highly probable that in the upcoming months, new information will surface demonstrating that some of these individuals were involved in combat against Israel in the 2014 Gaza Conflict." The importance here is twofold—first, that these are all fighting-age males and second, that the IDF's efforts to categorize them are ongoing. This latter point is particularly significant if one agrees with my suggestion that the civilian casualty rate in Protective Edge might provide a useful analytic benchmark against which to measure future IDF asymmetric engagements in urban settings. If so, it suggests that the rate I reported—between 36 percent and 56 percent of those killed were civilians—is likely to be fluid and specifically likely move downward as the Israeli categorization efforts continue. To the extent we use Protective Edge as a kind of benchmark, in other words, it is important to understand that it may not be an especially stable one.
A second corrective comes from the estimable Chris Jenks of Southern Methodist University's Dedman School of Law, who writes in with the following:
On your proportionality post, [I'm] not sure if this is a quibble or disagreement. But saying proportionality analysis is incident-specific is correct insofar as you are referring to it. But it's misleading in that there is a debate on that point. Imagine a military campaign in which a number of units take action. One of the units is conducting a feint or demonstration in the north. The whole point is to make the enemy think the friendly force is attacking north, not south, or that the friendly force is larger than it is. Meanwhile, on another part of the battlefield, perhaps a ways away in time and space, friendly forces conduct what is the main attack. The military advantage in securing the city in the north which is the feint is in and of itself not high. If the feint is considered for proportionality by itself, incident specific analysis, it may not fare well. The friendly forces don't care much about the city in the north. The whole point of the feint was to set the conditions for main effort success in the south.
The US view is that the proportionality is or can be more broad than incident specific when there is a coordinated military operation.
I take Jenks's point but I'm not sure his point and mine are at odds. His point, after all, goes to the question of how broadly or narrowly we define the incident to which I was referring when I described incident specific analysis. After all, we don't consider every time a weapon gets fired to be a separate incident. Chris is making the common-sense point that sometimes military activities that seem quite separate have to be considered jointly where coordinated military makes separate proportionality analysis irrational. In other words, treat two tightly interrelated actions as one incident to the extent one cannot meaningfully analyze their proportionality in abstraction from one another. That seems right to me.
It also doesn't conflict at all with my point that we never say that the LOAC proportionality inquiry allows the aggregation of all "casualty numbers or damage and ask whether they are in some gestalt sense proportional." There's a pretty big difference between saying that two incidents are intertwined enough to be treated as one and saying that all incidents should be treated as one for proportionality analysis. We just don't do the latter.
That said, I appreciate Chris's clarification of the margins of the principle as I articulated it.