Automated Violence: Who Dies for Politics?

Stock photo of a drone with a camera

If you're not already familiar with the incredible Heather Cox Richardson's Letters from an American blog/mailing list, and you have any interest in American politics, current events, and history, you are definitely missing out. She recounts current events with the context and references of a historian, while making both the news and the historical analysis accessible to non-historians. Well worth joining the read, it's reliably the most interesting thing in my inbox.

A few weeks ago, in the wake of the American withdrawal from Afghanistan, she discussed the tragic drone strike of August 29th, which killed 10 civilians including an aid worker and 7 children. After a New York Times investigation of surveillance video challenged the assertions by the Department of Defense that the car which was struck was carrying explosives intended for a strike on the Kabul airport. (The later analysis, according to the NY Times, suggests that it was actually carrying water bottles, and the driver's only interaction with the ISIS-K fighters had been a brief stop at a suspected safe-house where he may have been offering humanitarian aid to the beleaguered residents of a Kabul which was at the time in chaos during the American withdrawal.)

This kind of mistake is not a new one, nor is it the only ethical concern with armed drones. In the last 20 years, armed drones or over-the-horizon shipboard missiles have been responsible for controversial assassinations, attacks on hospitals and wedding parties, and many many deaths.

How many deaths? The number of civilian deaths is difficult to count, because identifying someone as a civilian or a "militant" in an urban combat situation with an unorganized militia or terrorist organization is frequently controversial, but a 2012 study by Columbia University estimated between 72 and 155 civilian deaths in Pakistan alone in just 2011. A study by the Bureau for Investigative Journalism and a similar study by New America(1), show between 910 and 2200 civilians killed between 2004 and 2020 in four countries (the New America data adds Libya, with an additional 637-930 civilian and 1867-2482 total deaths). In addition, they note at least 10,000 total casualties and as many as 20,000.

Country	Minimum Strikes	Civilians Dead	Children Dead	Total Dead	Injured	Total Casualties	Source
Afghanistan	13,072	300–909	66–184	4,126–10,076	658–1,769	4784–11,845	BIJ via Wikipedia
Pakistan	430	424-969	172-207	2,515 - 4,026	1,162-1,749	3,677–5,775​	BIJ via Wikipedia
Somalia	202	12-97	1-13	1,197-1,410	55-101	1,252-1,511	BIJ via Wikipedia
Yemen	336	174-225	44-50	1,020	155-303	1175-1692	BIJ via Wikipedia
Libya	4606	​637-930	not available	1867-2482	not available	1867-2482+	New America
Total	18,646+	1547-3130	283-454	10725-19014	2030-3922	​12755-22936+

Over the 20 year history of counter-terrorism over-the-horizon US drone strikes, the precision and accuracy has improved - but the decision-making process is still lacking. We can pinpoint targets using complex vision systems and behavior analysis, but it is difficult and prone to error. Beyond that, the potential targets are rarely observed and targeted *while* engaged in illegal activity, so targeting individuals relies on human intelligence (HumInt, a.k.a. traditional networks of spies and informants) to identify persons who might be part of terrorist organizations planning an attack. In a large number of cases, targets of drone strikes, even "militant" targets, were identified, targeted, and killed without any proof that they were planning acts of terrorism. None were given trials or opportunities to defend their actions - as "probable enemy combatants", the chain of command over the drone is their judge, jury, and executioner.

Not only is this deeply asymmetric with how white, American terrorists are treated (a threat identified as more dangerous to American security than jihadis by many sources), it violates the rules of Just Warfare and the Geneva Conventions by failing to conclusively identify combatants. Placing the weapons over-the-horizon places the decisionmaking authority over life and death in the hands of either automated systems (which are only as good as the information they can perceive and the biases they have been programmed with) or with distant operators, limited by what they can see and what they've been told about the targets, and then forced to reckon with the consequences. A New Statesman article recounts:

The effects of military drones go beyond the casualties they create. Pilots are required to confirm deaths and identities – meaning they witness, close-up, the impact of their work. It’s no surprise, therefore, that drone pilots experience a far greater degree of post-traumatic stress disorder than other “distance warriors” – stress that may be amplified by the isolated experience of drone operatives, who can’t discuss their work with family or friends.

This undermines the stated goal of drone warfare, its most important selling point: that it would reduce injury and trauma for American soldiers. It is clear to me that there is still significant injury here -- the responsibility of heavy civilian casualties, and the "dump shock" discordance of logging out and returning within seconds from surveying a destroyed home in Afghanistan to your comfortable office in Colorado Springs or onboard a ship miles offshore, is potentially just as damaging or even more than the potential threat of person-to-person urban warfare.(3)

In conclusion, the civilian casualties and the moral and traumatic consequences to the operators of armed drones make their use, especially in complex counterterrorism and urban warfare situations unadvisable and unethical. While the goal of reducing "our" casualties is well-intentioned, the results bespeak a lack of respect for human life in general, which is unacceptable, In addition, the mental health effects of drone operation and potential errors is a significant addition to the already underpowered military mental health system, and replacing soldiers with drones does not provide the "easy fix" for protecting our protectors that it was advertised to provide.

(1) I was not familiar with this news source until I found them searching for statistics, but Media Bias Fact Check rates them as biased towards left-center policy positions but highly factual. The same applies to the Bureau of Investigative Journalism.

(2) The New America methodology was very interesting reading. Attempting to gather data from potentially biased and often incomplete news stories is difficult, and official sources for casualty counts have been unavailable for most of the history of US drone warfare. When the US did, briefly, publish official casualty counts in 2016, the Bureau of Investigative Journalism noted that the official count of civilian casualties was a fraction of their own count, although the total reported death counts were very similar, suggesting a difference in how various people were classified as civilian or militant but verifying the accuracy of the BIJ's ongoing total counts. US official drone warfare casualty counts were discontinued in 2017.

(3) The civilian death toll for boots-on-the-ground urban warfare in the War on Terror have also been substantial - a rethinking of the entire Rules of Engagement for counterterrorism work is clearly needed to move that field of military thinking in line with ethical principles and Just Warfare principles.