Negligence has been defined as “the fact of not giving enough care or attention to someone or something.” When we consider the Aug. 29 drone strike in Kabul and the Department of Defense (DOD) response to that strike, we see the Aug. 29 strike serves as an exemplar of DOD’s overall approach to civilian harm over the past two decades: not giving sufficient care or attention. The results are what we saw in that courtyard in Kabul: tragedies for innocent civilians that are simply passed off as another mistake.
If this were the only or a rare time such a mistake occurred, then the DOD response could be reasonably understood. But that is far from the case. There have been thousands of incidents where U.S. attacks have harmed civilians over the past two decades. As a U.S. advisor, I myself have analyzed about two thousand incidents since 2007, and there were many more in the years prior. Conservatively, that alone translates into, on average, every week since 9/11 at least one such tragedy . What could be seen as reasonable mistakes in isolation add up to a lack of care when they reach thousands. Thus the incident in Kabul serves as a window to reveal an overall pattern of negligence with regard to DOD and civilian harm.
What does this overall negligence look like in practice? It includes four elements which I’ll discuss in turn:
- solutions were feasible but not pursued;
- lessons were identified but not learned and implemented;
- a striking lack of leadership existed within the DOD; and
- a failure to assign resources to address the issue of civilian harm persisted throughout.
Solutions were feasible but not pursued. At times, civilian harm is described as an unavoidable element of war, and a result of the fog of war. While civilian harm can never be eliminated entirely, the past decade has demonstrated that we can do better: practical steps to mitigate civilian harm can reduce the risk to civilians from military operations. My work with U.S. and international forces in Afghanistan and elsewhere showed that understanding the causes of civilian harm was crucial for mitigating risks to civilians. And the more specific we can be in determining those causes of harm, the more effective we can be in reducing it.
For example, DOD officials in the recent press briefing after the investigation of the Aug. 29 strike mentioned “confirmation bias,” and that is often a factor in these strikes. But we can also be more comprehensive in understanding the kinds of cognitive biases and other factors that can lead to civilian harm. To this end, I recently created a framework for civilian harm causes (below) that outlines various causes of civilian harm, which captures thousands of real-world incidents. This more detailed and broader understanding can help us to better identify why such harm occurs and what can be done to prevent it.
So, what more could have been done in the case of the Aug. 29 strike? Consider two examples.
First, the targeted individual was an aid worker, and visited an office of a humanitarian organization funded by USAID. The location of this office should be available in systems used for humanitarian notification and for deconfliction (e.g., for implementation of the No Strike List). While the pattern of life for the strike considered the individual’s visit to a location associated with a potential threat, operators do not appear to have considered other information available that could have shown the driver was a civilian (and associated with a humanitarian organization). This is a blind spot we have seen before in U.S. military targeting: targeting processes tend to focus on intelligence with threat information but do not consult information about the civilian population, humanitarian organizations, and critical infrastructure. Fixing this blind spot would go a long way toward reducing the cognitive biases that can lead to strikes such as this one.
Second, the family—including children—were apparently outside the frame of the full motion video during the strike. This is not a new problem. We saw a pattern of this sort of problem in Afghanistan and developed several ways to address it. This happened primarily with drones, as operators tended to zoom in for the engagement. That was solved by having the drone operators zoom out right before the engagement and look for transient civilians in the area. Another solution was to have another view of the target, either by another surveillance platform or a person on the ground, as overwatch to detect transient civilians outside of the field of view.
Another issue is one of information sharing. Drones, in particular, involve complex coordination and command mechanisms. In an assessment of civilian harm in airstrikes in Afghanistan in 2010 and 2011, I observed that drones were ten times more likely to cause civilian harm compared to manned aircraft, and a major reason was the propensity of failures in communicating essential details. All too frequently in these incidents, there was someone with information that, if provided to the individual approving the strike, would have stopped the engagement.
It is difficult to know what transpired in the Aug. 29 strike, but the failures appear to involve both of these problems: transient civilians outside of the frame of view, and a failure to communicate essential details. The Inspector General who conducted the investigation said that members of the strike cell did not alert others that they were about to take the strike. Doing so “would have allowed folks to provide additional context and information that might have been helpful to maybe delay the strike for a few minutes to do additional clearing of the target area, but the information sharing laterally and certainly outside the strike cell was not optimal,” Lt. Gen. Sami D. Said stated. CNN reported (and the New York Times confirmed) that the CIA had eyes on the target, saw the children, but was not able to get the information to the drone operators in time.
Overall, there are practical solutions that can be applied both to the circumstances of this specific strike and to the broader set of civilian harm incidents, but these solutions do not appear to have been implemented. There is even room for creative thinking about how to use emerging technologies to better protect civilians, but this avenue has thus far been largely unexplored.
Lessons were identified but not learned and implemented. Why haven’t such solutions been implemented? One reason is that DOD (and the U.S. government overall) does not learn effectively from its past mistakes. DOD lacks an institutional memory regarding civilian harm incidents and causes. This amnesia can be seen in Lt. Gen. Said’s comment that the conditions of the Aug. 29 strike were “a very unique strike.” But the problems seen in that strike were seen many times before. For example, the problem of transient civilians outside of a drone’s field of view was solved in Afghanistan during the ISAF campaign, but more recently there have been many incidents in Syria and Afghanistan where this problem appears to have recurred. The pattern re-emerged but steps were not taken to mitigate it, and so the problem was sadly seen again in the Aug. 29 strike. DOD hasn’t learned from such lessons.
Another example is the general problem of misidentification—targeting and engaging civilians in the mistaken belief that they are combatants. In a Joint Staff report I authored in 2013, I identified the problem of misidentification as a significant challenge that DOD needs to address. As I wrote back then:
Misidentification is often a result of a self-defense engagement where PID [positive identification] was based on perceived hostile intent. In other words, friendly forces engaged individuals who appeared to be behaving in ways that were suspicious or threatening, leading the friendly forces to believe the individuals were hostile; however, post-engagement BDA revealed the individuals were peaceful civilians.
In Afghanistan, we saw that half of all civilian harm incidents were caused by misidentification. Not only was this recommendation not implemented, but in a later 2018 Joint Staff review on civilian harm, the report asserted there was no such problem—an assertion based largely on interviews with military personnel. Sarah Holewinski and I wrote a dissenting footnote in that report to point out that data and studies contradict that position. And in the Aug. 29 strike, a civilian was once again targeted because he was mistakenly believed to be a combatant.
One particular challenge with misidentifications is that they are difficult for the U.S. military to detect even after the strike, especially with cognitive biases present. Indeed, Lt. Gen. Said stated that not only was “confirmation bias” a cause of the mistaken strike, but also a cause of the mistaken assessment of secondary explosions after the strike. It is not surprising that it took media reporting, most notably the New York Times video investigation, to discover that ten civilians, including seven children, were killed in the Aug. 29 Strike. This is another lesson we have seen over and over again; it is not an isolated mistake but rather part of a systematic pattern. There are other such patterns, identified through my and others’ analysis of thousands of civilian harm incidents, and many standing recommendations to address them that have not been implemented. And so, these problems persist and the civilian toll from U.S. operations continues to grow.
These lessons should inform the ongoing U.S. government review of its counterterrorism policy. Such a learning process was mandated in the 2013 Presidential Policy Guidance, and again in Section 4 of the 2016 Executive Order on civilian casualties (I initially drafted that section, hoping to address this chronic learning problem), but neither of those policy commitments has been acted on to date. Therefore, our counterterrorism campaign continues to be uninformed by these past lessons, with higher civilian harm as one of the costs.
Perhaps learning could be strengthened by having investigations and assessments both outside the chain of command (e.g., the USAF conducted the investigation of the Aug. 29 strike) and including experts on civilian harm. On the latter point, I have never seen an investigation or assessment of a specific civilian harm include an expert on civilian harm. For the Aug. 29 strike, Sarah Holewinski and I (the two authors of the dissenting footnote on misidentifications mentioned above) contacted General McKenzie, offering to help inform the investigation with past lessons, but our offer was ignored.
Leadership not exercised. DOD is the largest bureaucracy in the world (followed by China’s PLA and WalMart), and has the largest budget by far of any military on the planet. Where it puts its people and attention is a matter of leadership priority and focus. It is notable that while the Services, Office of the Secretary of Defense (OSD), and the Joint Chiefs of Staff have leadership positions for a wide variety of issues, there is no leader in DOD working solely to mitigate civilian harm. While there are leaders who have this issue in their portfolio, they have many competing priorities and almost no resources or authority (more on resources in a bit).
There are times when senior leaders get involved in civilian harm issues, such as Chairman of the Joint Chiefs of Staff Joseph Dunford ordering the 2018 Joint Staff Civilian Casualty (CIVCAS) review, but that effort was episodic, and none of the recommendations of that review have been implemented. Perhaps the most successful institutional leadership exhibited on civilian harm was by Admiral Michael Mullen, who created a Joint Staff CIVCAS Working Group in 2009 that included the Service chiefs and others. He himself led the effort for two years, demonstrating the personal importance of this issue to him. After two years, it was led by the Joint Staff J7, who oversees the joint lessons learned process. The working group created significant focus on strengthening efforts to better mitigate civilian harm in U.S. operations in Afghanistan. The 2013 Joint Staff CIVCAS report was written to try and capture the major lessons from that effort (and previous efforts in Iraq) to enable a shift to making institutional changes, but after civilian harm levels dropped in Afghanistan, the imperative for the effort was reduced, the working group was disestablished, and those institutional changes were not made.
Recognizing this historical lack of leadership, Congress mandated that OSD designate a senior official who would be responsible for civilian harm mitigation across the Department. DOD assigned this responsibility to the position of Deputy Undersecretary for Policy (DUSD-P), a position already responsible for many other issues. As a result, this move has had no tangible effect on leadership on this issue within DOD. Having met with several prior DUSD-P officials, I know they were universally well-meaning but not at all familiar with the substance of the issue of civilian harm, and had no ability to dedicate time and energy to work for needed progress. We see this with the Aug. 29 drone strike: while General Kenneth McKenzie himself, as commander of the U.S. Central Command, was involved with the response to the strike, it is notable that the DUSD-P—the senior DOD official responsible for the issue of civilian harm—has not made any statements regarding the incident or DOD’s overall response.
No resources. Another reason why progress has been so challenging is that DOD has expended almost no resources to solving this problem. There are fewer than a handful of individuals working on civilian harm within DOD. Even the 2018 Joint Staff CIVCAS review showed a stinginess of resources, turning to the National Defense University (NDU) to lead the review. NDU had no institutional expertise in civilian harm, but they were a resource that cost nothing to use. In contrast, both RAND and CNA contributed personnel to this effort, bringing in real expertise, but those organizations had to donate those personnel because the Joint Staff was unwilling to fund them. There is much work that can be done, as I point out above regarding feasible solutions and learning, but the resources are simply not there to carry out those tasks. In a Department that has over $700 billion in its FY22 budget, this is a remarkable form of neglect.
There is an easy solution to this problem. Congress should direct an earmark to devote resources to the mitigation of civilian harm. There are strong precedents for this. For example, Congress dedicates over $10M to the implementation of the Leahy law every year, which promotes accountability for foreign security forces when they harm local populations. Should the U.S. government spend any less on preventing the harm of civilians from its own actions? I would argue this should be much higher, but even that amount would be drastically more than what is available now.
End the negligence. The United States government, and DOD in particular, has been nothing short of negligent regarding the issue of civilian harm. We see this in the Aug. 29 drone strike and response, but that is just the tip of the iceberg, the one incident that is seen by the public when there are many, many more that go unseen. As I stated above, on average at least one incident of civilian harm has happened every week since 9/11. This level of civilian harm is not a necessity of war, and certainly can be mitigated. The negligence must be stopped: by pursuing feasible solutions, identifying and implementing lessons, exercising strong leadership, and assigning resources to address the problem of civilian harm. What we have seen so far is negligence, but this negligence can easily be cured with a few decisive actions and commitments of leadership and resources. Americans should expect—we should demand—more from our military and our government.