My student Lina Shaikhouni and I have a new Foreign Policy piece in which we make “The Case Against the Case Against Drones,” to paraphrase Stephanie.
Therein, we argue that although there are many good reasons why drones shouldn’t be used as they are being used, or where they are being used, the arguments against drones per se are often based on misconceptions or assumptions for which we don’t have good data.
In particular, we point out:
1) drones are not killer robots
2) there’s no evidence that “video game” warfare makes war more likely
3) drones don’t violate humanitarian law
3) we don’t have good data on civilian casualties
We also argue that the focus on drones per se is a distraction from what are actually more profound and wider issues:
… Those who oppose the way drones are used should shift focus to one of the big normative problems touched by the drone issue: whether truly autonomous weapons should be permitted in combat, how to accurately track the civilian cost of different weapons platforms, and whether targeted killings — by drones or SEAL teams — are lawful means to combat global terrorism… Focusing on the drones themselves misses this bigger picture.
What you see above is a visualization of the data on which our argument is based, created with help from Dan Glaun (click here for clearer picture). We ran a search for “drone warfare” on DailyOpEd.com, which indexes opinion articles and letters to the editor, and identified 29 hits that dealt with drones specifically (as opposed to just mentioning them in the context of arguments about other things). We loaded these documents into a tool called DiscoverText and used it to render a tag cloud of the most frequently used words across the set of op-eds.
We then drilled into the text itself, studied it closely, and used a consensus coding scheme to tag specific instances of the types of misconceptions we identified across all the op-eds. Any statistical or numerical references in the article are based on this coding. While we don’t single out any author in particular as intending to make any of these particular arguments, the aggregation of passages across many op-eds adds up to a pattern in the broader debate. Specific quotations emblematic of the misconceptions we noticed have been high-lighted in the info-graphic above.
Charli – I’m currently coauthoring a chapter in a forthcoming book on the ethics of drone warfare, so I read your and Lisa’s piece with interest. One thing that struck me is that you’re slightly misrepresenting the “drones make war more likely” position.
The argument, as I understand it, is not dependent on the individual-level psychological distancing/”Nintendo effect” – that’s the argument for why drones make atrocities more likely. The jus ad bellum argument, rather, is the claim that drones make aggressor casualties less likely, so policymakers are more likely to initiate war because the cost/benefit analysis is shifted in some pretty fundamental ways. Thoughts?
OK fair enough – we did conflate those two things and should have developed each separately. Still, I’m not sure the casualty aversion argument is any better than the social distancing one at explaining the likelihood of war, since (correct me if I’m wrong) the bloodlessness of interstate wars has been increasing since the establishment of the Geneva Conventions but nonetheless the propensity for interstate war remains at historic lows. Unless you have other suggestions of how to parse this or know of someone who has. I admit I haven’t thought it through fully yet.
Thanks for the reply, Charli – I should say at the outset that I’m fan both of this piece and your work in general, so all criticism/argument is intended as constructively as possible.
You’re right about the numbers since the Geneva Conventions, but the explanations as to why might support the claim that there’s something unique about the way that drones affect the push to war. Take the following explanation: interstate war has become more bloodless because great power war has been shown to be too costly as a result of the world wars and nuclear weaponry. If great powers could fight over contested territory using principally remote weaponry, the cost/benefit ratio might change to the detriment of the civilians where the fighting is taking place.
One also doesn’t need to accept a realist explanation of decreasing casualty rates for this argument to work. Take explanations like the proliferation of liberal democracies (where leaders are responsible to their citizens for casualties in war) or a shift in global just war norms. In the case of the former, the argument’s pretty intuitive: advanced liberal democracies are the most likely states to have sophisticated drone technology due to technological advantage and their citizens doesn’t care about dead drones. The normative explanation is a bit harder, but one could imagine the normative context being altered a bit by material capacity to conduct war with only limited casualties to one side.
Obviously there are many more explanations that I could have considered, but the point comes across. For citations on this sort of issue, I would look at Chris Coker’s work on the future of war and, for a more philosophical perspective, this article (https://philpapers.org/rec/KILRWT). I disagree pretty strongly with her normative conclusions, but I do think the thought experiment at the heart of the paper is interesting and relevant to the current discussion.
If, as you argue, drones are not “killer robots,” and we stipulate your points 2 & 3, the issue really becomes #4 — collateral damage. Drones are attractive to COIN planners because they can loiter with no risk to friendly personnel (and generally low signature) while the “reconnaissance/strike complex” sifts through the data to identify a target. But so long as you have high-value targets embedded within a civilian population you will have an exceptionally high risk of collateral damage that will almost surely dilute the value of the strike, regardless of whether you employ a drone or an F-16. It’s hard to draw the line on what constitutes “proportional force” in targeted killing.Â
And wrt drones, you’re probably aware that DOD is also studying the question of whether a drone operator in Nevada is a “legal combatant” and can be attacked under the laws of war. If a covert al-Q fighter sneaks into the US to kill a drone operator (even one who is off-duty), isn’t this a legitimate act of war? Extend the argument to include civilians working for defense contractors involved in the design, manufacture, & support of combat drones and their associated technical systems. I think we’re just scratching the surface of the impact of advanced technology in warfare, particularly COIN.
Yes I’ve heard this argument percolating in the intlaw community and am not yet sure where I stand on it. What’s your take?
There’s actually quite a bit of interesting nuance in the debate over the “legal problems” we referred to in that section – problems that have to do not with drones per se but with how and where they are used. At Lawfare Blog (see below), these are being discussed in much more depth, and Robert Cheney wishes we’d have parsed the UN Charter and lawful combatancy issues more.Â
https://www.lawfareblog.com/2011/06/unsound-criticism-of-the-legality-of-drone-strikes-at-foreign-policy/
In terms of IHL, we chose to simply focus on whether the weapons as such would be problematic under Article 36 of AP 1 to the Geneva Conventions – they wouldn’t, but that doesn’t mean there aren’t broader legal issues associated with targeted killing and asymmetric wars, and the drones debate certainly touches on them. The argument against targeted killings is based on the assumption that we use a human rights / criminal justice frame instead of a war law frame. There’s an open debate about which is appropriate.Â
We entered the age of “total warfare” several decades ago, I believe. Air Chief Marshal “Bomber” Harris certainly believed that industrial workers in Hitler’s Germany were legitimate targets, and never pretended that killing workers and destroying their homes
was a byproduct of attempting to hit factories. Not too hard to
extend that argument to the present day, so yes, I think people and facilities in the drone infrastructure here in the USA cannot immune from attack under laws of war that we recognize.Â
What’s This?TranslateWhat’s This?Return
DECEMBER 21, 2010
There Were More Drone Strikes — and Far Fewer Civilians Killed
By Peter Bergen and Katherine
Tiedemann
In the
first 11-and-a-half months of 2010, U.S. President Barack Obama’s administration authorized more
than twice as many drone strikes, 113, in northwest Pakistan as it did in 2009 —
itself a year in which there were more drone strikes than during George W.
Bush’s entire time in office.
Given
the evident importance of the program to U.S. policy toward Pakistan, it
is necessary to ask what we know about the drone strikes, where they happen,
and whom they are killing.
Based on
updated reports from news organizations with deep and aggressive reporting
capabilities in Pakistan (the New York
Times, Washington Post, and Wall Street Journal), accounts by major
news services and networks (the Associated Press, Reuters, Agence
France-Presse, CNN, and the BBC), and reports in the leading English-language
newspapers in Pakistan (the Daily Times,
Dawn, the Express Tribune, and the News), as well as those from Geo TV, the largest independent Pakistani television
network, we have been maintaining a transparent database and interactive map that
tracks every reported drone strike since 2004.
Counting
drone strikes and fatalities is an art, not a science, as it’s not possible to
differentiate precisely between militants and nonmilitants because militants
live among the population and do not wear uniforms, and because government
sources have the incentive to claim that only militants were killed, while
militants often assert the opposite.
Still,
we’ve been able to discern some surprising trends. A frequent criticism of the
drones program is that the strikes kill too many civilians. In the busiest year
of the program, September 2010 was the busiest month, with 22 strikes reported
amid news about potential Mumbai-style attacks in Europe, followed by October
2010 with 15 strikes reported and November 2010 with 14. But even as the number
of reported strikes has skyrocketed — with one every three days in 2010,
compared with one a week last year and one every 11 days in 2008 — the
percentage of nonmilitants killed by the attacks has plummeted.
Pakistani
government officials estimate that more than 700 civilians were killed by the
drone strikes last year, but a U.S. government official asserted a year ago that
“just over 20” civilians and “more than 400” fighters had
been killed in less than two years. U.S. officials continue to claim
(anonymously, of course) that only 1 or 2 percent of those killed by the
strikes are civilians, and other estimates of civilian deaths range from a high
of 98 percent down to 10 percent of the total fatalities.
According
to our estimates, the nonmilitant fatality rate since 2004 is approximately 25
percent, and in 2010, the figure has been more like 6 percent — an improvement that
is likely the result of increased numbers of U.S. spies in Pakistan’s tribal
areas, better targeting, more intelligence cooperation with the Pakistani
military, and smaller missiles.
Under
the Obama administration, approximately 80 percent of those reported killed by
drone strikes have been militants; under the Bush administration, it was closer
to 55 percent. The majority of those killed appear to be lower or midlevel
militants; of the some 1,260 militants reported killed in the strikes since
2004, only 36, or around 2 percent, have been leaders of al Qaeda, the Taliban,
or other militant groups.
Deciphering
the true nonmilitant fatality rate from the drone strikes is important as a
practical as well as a moral matter. If the true nonmilitant fatality rate
were more widely known in Pakistan, the program might be less unpopular there.
Those targeted in the strikes, after all, are thought to have carried out or
planned attacks not only in Afghanistan and the West, but also in Pakistan,
where more
than 4,000 people have been killed in militant attacks since the Red Mosque
incident in July 2007.
North
Waziristan, home to a hornet’s nest of militants affiliated with the Haqqani
insurgent network, al Qaeda, Tehrik-i-Taliban Pakistan, and other local
militants, has been the target of about 90 percent of the reported strikes this
year. Given that the region is the source of at least
half of the attacks in Afghanistan, this is unsurprising. Before the
Pakistani military began a much-anticipated offensive in South Waziristan in
October 2009, some 60 percent of the strikes that year took place there;
since then, only nine of the 122 reported strikes have occurred in that
southernmost tribal area, suggesting a high degree of Pakistani-American
coordination. Bolstering this point are reports
that the CIA and Pakistan’s intelligence service, the ISI, have conducted more
than 100 joint operations in the last year and a half, including the arrest of
the Taliban’s second in command, Mullah Baradar, in early 2010.
Another
common refrain from critics of the drones program is that it violates
Pakistan’s territorial sovereignty, one of the possible reasons the strikes are
so unpopular among those
who live where the strikes most frequently occur. But despite official public
protest, in private Pakistani officials are supportive of the program, and in
fact much of the targeting intelligence appears to come from Pakistani sources.
The
drone strikes are a tactic, not a strategy. But given Pakistan’s negative
reaction both to a Special Forces raid
in South Waziristan in early September 2008 and to NATO helicopter
strikes in Kurram in late September 2010, plus the Pakistani military’s
unwillingness
to mount major offensives in North Waziristan, it’s not clear what additional
options the United States has. Militants
continue
to seek to attack the West from the tribal areas, despite the drone program’s
escalation. U.S. officials are reportedly even seeking
the expansion of the program into Quetta, where the leadership of the Taliban
is believed to be based. So despite the controversies about civilians,
sovereignty, and strategy, the strikes are still, as CIA chief Leon Panetta commented
in May 2009, “the only game in town.” For that reason, it’s worthwhile for
everyone to understand exactly what the rules of the game are and continue to
be.
Peter Bergen, author of The
Longest War, is the director of the New America Foundation’s National
Security Studies Program, where Katherine Tiedemann, a doctoral student in
political science at George Washington University, is a research fellow.
Â
Peter, thanks for your comment. I’m not quite sure how to interpret it.Â
Regarding your data itself, I think your dataset is one of the best available, but I have concerns about how you are coding ‘civilian’ versus ‘militant.’ I’ve written exhaustively before about the questions I have regarding your coding methods (see links below), and would welcome a response to my queries about them. https://duckofminerva.blogspot.com/2011/01/actually-we-dont-know-how-many.htmlIf you’re drawing attention to your research showing that as drone strikes have increased, reported civilian casualties as a percentage of all casualties have decreased (according to your data), you may be asserting that we pay too little attention in our piece to the potential upside for civilians of drones. We don’t dwell on it, but I think that hypothesis is implicit in our civilians and IHL sections. I find it highly plausible that drones may actually cause fewer civilian deaths than manned aircraft or ground troops in similar situations, and so to the extent that your data bears this out, we are on the same page.However, without good comparative data – not only on drone strikes but on collateral damage rates from other strikes – I’m not sure what to make of either hypothesis. Our argument is not that your data is wrong, but that it’s only one of wildly conflicting estimates and the world could do with a standardized mechanism for counting incidental casualties not just of drones strikes but all military ops.Â
As a meer diligent observer of Foreign Affairs, I found the article to be informative and substative. The crucial point, being as discribed, should be the relevant question to be answered. Who will decide who is a target, and who is to decide the answer, and based upon what/which/or whatever. What are the criteria? It truely is the extrajududicial and expediant dispatch of “judgement”, that needs to be discerned. Thank you for bringing that question to our attention.
you may find a paper from Pat Johnston instructive: https://patrickjohnston.info/research.htmlThere is a similar study on Comparative Strategy published last year that sees that drone strikes reduce insurgency.Â
Thank you – I’m grateful for this and have not seen it before. However, the link is not working – if you come back by please repost!
Sorry,
https://patrickjohnston.info/research.html You have to write him for accessing the article.
:) anon
As a member of ICRAC and coauthor of its Berlin Statement https://www.icrac.co.uk/Expert%20Workshop%20Statement.pdf I have particularly emphasized the autonomous lethal robot question which you zero in on in your point #1.Â
We need to draw a bright red line now and say that we will not allow a future in which machines are set loose to decide when and where and against what or against whom to apply lethal violent force. We must assert, without need of scholarly justification, that it is a human right not to have one’s life or death decided by a machine, no matter how “intelligent.”
However, this is not to say that the use today of remote-controlled killing machines should not be a matter of concern or should not be subject to national and international legal restrictions. In particular, the profligate use of drones to carry out “targeted killings” — a euphemism for a once-banned policy of official assassination — risks creating an increasingly lawless world in which no one is safe from what could become quite insidious lethal technology. As Malcom X put it, “the chickens come home to roost.”
You zero in on this in your point #4, but you argue that the fact that so many of these attacks are being carried out using drones is “irrelevant.” It seems to me that in fact the use of drones serves to shield these actions from scrutiny as extrajudicial assassinations and violations of the national sovereignty of states we are not at war with. The fact that a drone is used becomes the overarching qualitative categorizer of the action, and we have rapidly been conditioned to see the drone weapon as an arrow in the quiver of the President which he is entitled to use when and where he sees fit.
To ask whether drone strikes kill more or fewer innocent people than strikes by manned bombers misses a deeper question: Exactly how did bombing become OK, and do we want to OK a new kind of wanton killing?
Teleoperated weapons also represent a new technological frontier of the arms race, and such weapons should be subject to both qualitative and quantitative arms control limitations in order to prevent this from leading to destabilizing competitions and dangerous preemptive postures.
Given the seriousness of this issue, to quibble with the use of terms like “killer robot” in reference to teleoperated weapons, or to argue that the important issue of autonomous lethal decisionmaking should be isolated from concerns about the use and proliferation of drones today, seems overly pedantic. We need to build broad public engagement with this issue, and I daresay that the public views teleoperated robots as robots and drones as robot weapons. Isn’t that, after all, what your analysis of op-eds reveals?
Furthermore, to insist that teleoperated weapons have nothing to do with autonomous weapons would be to ignore the way the line is progressively blurred as technology advances and autonomous lethal decisionmaking moves stealthily forward.Â
UAVs today are strongly enabled by the onboard computers that allow them to be controlled from halfway around the world despite communications delays, in some cases by point-and-click with a mouse. The planes basically fly and land themselves. Increasingly, automated systems will cue operators and assist in identifying targets. The military is seeking to have multiple drones operated by a single “warfighter,” and talks about moving from “human in the loop” to “human on the loop.”Â
While I believe it is possible to draw a clear line by insisting on the responsibility of an individual human soldier for every individual lethal action, it will be impossible to do this without addressing the “slippery slope” and “grey area” arguments which are increasingly cited as proof that fully automated lethal decisionmaking is inevitable.
“UAVs today are strongly enabled by the onboard computers that allow them to be controlled from halfway around the world despite communications delays, in some cases by point-and-click with a mouse. The planes basically fly and land themselves. Increasingly, automated systems will cue operators and assist in identifying targets.”
Why haven’t you just described, in one way or another, all modern weapons systems — the only difference being the physical distance between (some of) the equipment and the operator?
Clearly not all, but more than just drones. So what? We are concerned about the drift toward automated killing. You are right that this is more general than just “killer robots,” i.e. self-mobile weapons possibly operated at a distance from any human decisionmaker. However, it is of special relevance to such weapons, since the weakness of comms links and the potential for robot-robot combat is often cited as an imperative driver for the move to fully autonomous operation.
Mark, you’re raising some really interesting points. Could you contact me by email to discuss further (I’m having trouble tracking down yours)? I’m writing a book chapter on the autonomous weapons issue, which I understand the humanitarian law folks see as distinct from UAVs, and so have been treating as distinct. But I’m really interested in the development of new normative understandings here so to the extent that you and other anti-AWS advocates see there being a slippery slope between the two I’d be really interested in understanding that view better. Let’s talk.Â