The debate over civilian casualties from drone strikes continues
The Jamestown Foundation's Terrorism Monitor released a study last week claiming to shed new light on the accuracy of the CIA's drone strike campaign in northwest Pakistan. The authors of the Jamestown study compiled a database of statistics on civilian and militant casualties from drone strikes, relying on both Western and English-language Pakistani news reports about drone strikes -- a methodology broadly similar to the one used by both The Long War Journal and by The New America Foundation. The study found that only 68 casualties (4.95%) could be clearly identified as civilian, and concluded that "the [drone] strikes have not only been impressively accurate, but have achieved and maintained a greater proportionality than either ground operations in the area or targeting campaigns elsewhere."
Charli Carpenter, an associate political science professor at University of Massachusetts - Amherst who specializes in human security, has offered a thought-provoking critique of the Jamestown study. First, Carpenter questions the study's assumption that "all children under 13 and women" should be classified as civilians:
Numerous scholars, myself included, have shown how misleading it is to assume all women are civilians and all men and older boys are combatants; and to build this gendered stereotype into one's dataset immediately prejudices the data in favor of finding fewer civilian deaths.
On the one hand, this is a valid critique -- clearly, there have been teenage-aged males and older men killed who are not militants, and it is entirely possible that some women killed in drone attacks may have been involved in supporting militant activities (which is why it's better to track all confirmed civilian casualties regardless of age or sex).
However, it should be understood that some assumptions are defensible when conducting a study of civilian casualties from "covert" airstrikes in an extremely inaccessible, isolated, and remote region. Given that Taliban factions immediately surround and cut off access to the area in the aftermath of each attack, objective identification of each casualty by an independent source is impossible. Scholars attempting to establish an accurate estimate of civilian casualties are forced to rely on either often thinly detailed Pakistani and Western press accounts or the Taliban's own propaganda, which predictably claims that virtually all casualties are civilians. So while somewhat arbitrary classifications may be problematic, given the restrictions facing researchers I think they are also a defensible method of estimating (note: different than counting) civilian casualties.
Carpenter also takes issue with the study's categorization of "suspected militants," explaining that this phrase implies that:
[T]he US is carrying out a mass murder campaign against individuals suspected of committed crimes, in the absence of any sort of effort to determine whether or not they are actually guilty. In short, what renders these individuals putative "legitimate targets" appears to be nothing more than the suspicions of those with their fingers on the trigger. Oh, and possessing testicles.
Sarcasm aside, Carpenter's claim is undoubtedly driven by the substantial secrecy about the CIA's process for selecting targets for drone strikes. Yet it is disingenuous to say that there is an "absence of any sort of effort" on the part of the CIA to determine if the people they are killing are "actually guilty." Alleging that CIA officials are carrying out a "mass murder campaign" against Pakistanis on little more than "suspicions" is no less absurd than assuming every suspected militant is a terrorist.
A reasonable person would assume that the CIA is at least attempting to verify, corroborate, and ensure that the intelligence that feeds these strikes is accurate, and then determining that the intelligence is convincing enough to carry out an armed strike. Since we don't know how the CIA chooses targets, we have no way of gauging the accuracy or validity of that intelligence or decision-making process, but insinuating that the CIA is flying drones around Pakistan killing people on a whim requires a substantial amount of bad faith.
Finally, Carpenter accurately notes that the study is actually measuring the concept of "distinction" or "discrimination," and not the concept of "proportionality."
The distinction principle measures the ability to hit combatants while minimizing the costs to civilians. Proportionality measures the overall good of an attack relative to its overall negative side-effects.
From a human security perspective, I would argue the appropriate measure for an analysis of proportionality would not be the number of civilian death to combatant deaths, but rather the number of civilian deaths by drone strikes to some estimate of the number of Pakistani civilians who will not now die as a result of militant activity.
I leave it to the number crunchers to figure out how to calculate this.
While I don't think such a number is quantifiable, an even better metric would also take into account the lives of Afghan civilians and citizens of Western countries who would have perished in additional terror attacks organized by Taliban and Al Qaeda leadership in Pakistan.
For reference, since the start of the intensified CIA drone campaign in Pakistan in mid-2008, well over 7,000 Pakistani civilians have been killed in terrorist attacks (detailed data here)-- and that's with the pressure of drone strikes. The highest scholarly estimate of the total number of Pakistani civilians killed by drone strikes (the New America Foundation's "high" estimate) is 535.
How many more Pakistani, Afghan, and Western civilians would have been killed in the absence of these strikes? Would the additional deaths be higher or lower than the number of civilians killed by drone strikes? Your guess is as good as mine.