Just before Independence Day, an analyst for a defense research agency stated in a media interview that a classified DoD study shows that drones are likelier to cause civilian harm than attacks from manned fighters. Lawrence Lewis, a researcher for the Center for Naval Analyses, says these findings resulted from a  statistical analysis he conducted using classified data from Afghanistan mid 2010-mid 2011 as part of a project funded by DoD’s Joint Center for Operational Analysis.
If true, this would dramatically shift the discussion about the humanitarian impact and value of armed drones. There are all kinds of human security arguments against drones – they make war likelier, they kill too many civilians, they weaken the rule of law; and all kinds of national security arguments in favor of them – they decapitate terror networks, they prevent attacks, they keep troops out of harm’s way. But there is also a human security argument in favor of drones: that when used lawfully they save foreign civilian lives relative to other kinds of strikes because they are precision weapons.
But if drones are likelier to harm civilians than manned attacks (the explanation is that drone pilots lack the training in humanitarian law and civilian protection that manned pilots have) then that goes out the window.
So, is the finding valid?
That’s the problem: no one can review or interpret the results of the study because both the report and the data are classified. So actually it is impossible for journalists to verify what the study finds beyond the declassified executive summary, much less for commentators to evaluate the data, the findings or the causal analysis. There is good reason to think it may be valid since the DOD has little incentive to manufacture liberal estimates of civilian casualties. But all reporters can know at this point is that a DOD analyst has reported such a finding. We don’t know how what data is used, how the study defines “civilian” or “casualty,” how the analysis was conducted or how generalizable it is. We especially do not know how the causal explanation for the finding was arrived at without seeing how the analysis was conducted. And we can’t know any of that if the study remains classified.
We do know one other thing, however. If the study actually exists, then whatever the data shows it means the US Government is sitting on an analysis of body counts it has been denying it keeps, the findings of which should certainly be subject to public scrutiny, since we need to know something about casualty rates in order to have a more informed debate. Perhaps they will be soon, if a pending FOIA request by the ACLU is honored.
Â
Charli: as I noted when the study came out, I wouldn’t be too quick to assume that the DoD is a unified actor here. There are those within the USAF who have *very* strong incentives to discredit drones in favor of strike aircraft, which is exactly the comparison made in the study. There are some real reasons this smells, including the lack of obvious comparability between missions, the fact the study may be based on data involving very different mission profiles than what drones are being used for in, say, Pakistan, etc. Should be interesting if the FOIA request succeeds.
https://www.whiteoliphaunt.com/duckofminerva/2013/07/tiwesdaeg-the-left-hand-of-linkage-5.html
Good point and thanks for referring me back to that linkage post, which I’d missed. My biggest questions are how ‘civilian’ is coded, and I’m also not happy with how the media coverage has misreported the findings, but that’s the subject of a different blog post.
Normally I would agree with interservice rivalry. But, in this case I am not sure. The other JCOA report that has been declassified lately emphasized reducing civilian casualties are part of counterinsurgency strategy, and does not read like a interservice paper. Also, from the executive summary, I do not see pleas to move drones under a different service, which is the result for example of the CIA-Air Force debate over the U-2 during the Cuban Missile Crisis. In my mind, this may simply be a part of the military’s continued recognition, which happened throughout Afghanistan, that civilian casualties are usually counterproductive.
My concerns have more to do with the comparative aspect of the study and how it (1) argues that strike aircraft are more precise and (2) doesn’t seem to apply to a broad range of drone missions. Recall the *intraservice* dynamics. Drones challenge the fighter-jock culture of the USAF.
That makes sense.
One thing to consider might be whether there are differences in how drones are used tactically compared to manned aircraft. In a comparison between, say, manned aircraft called in to support forces already in an engagement versus drones used for ‘signature strikes’, then it might be that difference in civilian deaths would obtain regardless of manned/unmanned. Or it may be that ‘signature strikes’ carried out by manned aircraft would produce even more civilian casualties. Or that you just can’t do them for technical reasons, making the comparison senseless. In all of these situations, there may be a far more significant variable than pilot training determining the difference in casualty rates.