Happy Flag-Waving Drone Document Dump

ODNI (but not yet I Con the Record) has released its report on the number of drone deaths. The overview is that the US intelligence community is reporting (more on that in a second) far, far fewer drone deaths than credible outside researchers do. (TBIJ, New American, Long War Journal)

The IC numbers are for strikes occurring outside areas of active hostilities, which currently includes Afghanistan, Iraq, and Syria, but might have — the report doesn’t say one way or another — included other places, like Pakistani tribal lands, when these drone strikes happened.

The report acknowledges that this number differs dramatically from these of outside researchers, though it doesn’t include a footnote to permit those who don’t already know the players to compare, which betrays a real lack of confidence in its own analysis. A footnote would also permit readers to see the degree to which NGOs have done granular analysis, as compared to ODNI’s 3 line table.

Plus, it doesn’t acknowledge this discrepancy until after it suggests these other numbers — which I believe are actually more consistent with each other than the IC’s numbers are with them — come from terrorist propaganda, a claim it repeats a second time before the end of the 3-page report.

The large volume of pre- and post-strike data available to the U.S. Government can enable analysts to distinguish combatants from non-combatants, conduct detailed battle damage assessments, and separate reliable reporting from terrorist propaganda or from media reports that may be based on inaccurate information.

In releasing these figures, the U.S. Government acknowledges that there are differences between U.S. Government assessments and reporting from non-governmental organizations. Reports from non-governmental organizations can include both aggregate data regarding non-combatant deaths as well as case studies addressing particular strikes, and generally rely on a combination of media reporting and, in some instances, field research conducted in areas of reported strikes. Although these organizations’ reports of non-combatant deaths resulting from U.S strikes against terrorist targets outside areas of active hostilities vary widely, such reporting generally estimates significantly higher figures for non-combatant deaths than is indicated by U.S. Government information. For instance, for the period between January 20, 2009 and December 31, 2015, non-governmental organizations’ estimates range from more than 200 to slightly more than 900 possible non-combatant deaths outside areas of active hostilities.

[snip]

Finally, non-governmental organizations’ reports of counterterrorism strikes attributed to the U.S. Government—particularly their identification of non-combatant deaths—may be further complicated by the deliberate spread of misinformation by some actors, including terrorist organizations, in local media reports on which some non-governmental estimates rely.

The IC report also suggests that it derives such a low civilian casualty figure by defining belligerent broadly, to include people like drivers and cooks — but don’t you worry, that doesn’t mean that every single military aged male counts as a belligerent (I will check but I suspect the IC’s numbers likely could not be so low without counting some women as belligerents, which might happen if they do things like cook).

Non-combatants are individuals who may not be made the object of attack under applicable international law. The term “non-combatant” does not include an individual who is part of a belligerent party to an armed conflict, an individual who is taking a direct part in hostilities, or an individual who is targetable in the exercise of U.S. national self-defense. Males of military age may be non-combatants; it is not the case that all military-aged males in the vicinity of a target are deemed to be combatants.

[snip]

The U.S. Government draws on all available information (including sensitive intelligence) to determine whether an individual is part of a belligerent party fighting against the United States in an armed conflict; taking a direct part in hostilities against the United States; or otherwise targetable in the exercise of national self-defense. Thus, the U.S. Government may have reliable information that certain individuals are combatants, but are being counted as non-combatants by nongovernmental organizations. For example, further analysis of an individual’s possible membership in an organized armed group may include, among other things: the extent to which an individual performs functions for the benefit of the group that are analogous to those traditionally performed by members of a country’s armed forces; whether that person is carrying out or giving orders to others within the group; or whether that person has undertaken certain acts that reliably connote meaningful integration into the group.

The ACLU is due to get more documents from the precipitating FOIA that may explain better how broadly the government has defined belligerent (remember–these strikes are all in areas outside of active hostilities).

Perhaps the most interesting part of the report is this repeated language:

a summary of information provided to the DNI

The assessed range of non-combatant deaths provided to the DNI

The information that was provided to the DNI

based on the information provided to the DNI

according to information provided to the DNI

That is, the ODNI may be releasing this information. But they’re sure as hell not vouching for it. I find that particularly interesting given that, in May, I had to explain to ODNI that the National Security Letter numbers they were getting (and publishing in transparency reports) from FBI were probably unreliable.

These numbers don’t even, apparently, reflect the kind of rigor that would involve an outside agency reviewing the CIA’s numbers. Instead, the CIA (and presumably, in more limited cases, DOD) provided numbers to ODNI, and ODNI is — as ordered by the President — passing those numbers on.

At least you can be sure this isn’t terrorist propaganda.