Drone Strikes Gone Wrong: Fixing a Strategic Problem
By Andrew Milburn
Watching the chaotic scenes in Kabul airport this last August, it is difficult to make sense of the manner in which Washington pulled the plug on a two-decade Coalition effort leaving our allies non-plussed and our partners to the mercy of a vengeful enemy. Less than three weeks later, these images came again to mind during the testimony of Secretary of Defense Lloyd Austin and two of his four-star generals before the Senate Armed Services Committee. Nothing in that testimony, however, brought a sense of closure. Instead, repeated attempts at justification, and ultimately – a collective refusal to take responsibility – only rubbed salt in the wound.
As we wait for the investigations and inquiries to play out, however, I want to focus here on critical lessons from a single incident. It was the last offensive action taken by the United States in a 20-year war – a drone strike that failed to hit its target, killing instead several civilians. The mistake was no isolated incident but part of a pattern that has implications not just for US counter-terrorism strategy but for US foreign policy going ahead. I discuss here what the problem is, why it matters, and how to fix it.
Stumbling Behind the News Cycle
Instead of admitting that this might be the case, however, the Pentagon reacted to emerging evidence with a sluggish response that lacked transparency. For days after the Aug. 29 strike, Pentagon officials asserted that it had been conducted correctly, despite 10 civilians being killed, including seven children. Meanwhile, news organizations raised doubts about the official version of events, reporting that the driver of the targeted vehicle was a longtime employee at an American humanitarian organization and pointing to a lack of evidence to support the Pentagon’s claim that the vehicle contained explosives.
Sadly, the New York Times revelations were not enough to prevent General Milley from calling the strike “righteous” with an enthusiasm that might seem inappropriate even if he had been correct. Instead, the strike was far from righteous. It was a completely avoidable mistake, propelled by flaws in organizational culture to a level where it would undermine US credibility on the global stage.
A Cultural Faultline
Let’s forget the troubling circumstances for a moment and take a clear-eyed look at what this tragic curtain call teaches us about the pitfalls of conducting counter-terrorism by drone, because if United States policy involves withdrawing from volatile regions around the globe, this may have to be our modus operandi going ahead.
The good news is that after the initial faltering, the Pentagon has been open about what happened. Cynics might suggest that this is merely because irrefutable evidence was made publicly available by an iconic US media enterprise, the New York Times.
The incident reveals a cultural faultline between the US military and organizations such as the media, the UN, and the various non-profit groups that work in combat zones. Such organizations tend to see the Pentagon as being purposefully deceitful and, perhaps criminally negligent when it comes to loss of human life that is not American.
I see instead an ill-informed, stumbling bureaucracy whose leaders genuinely want to do the right thing, but whose actions only continue to confirm negative perceptions around the world. And in a global landscape where the United States can no longer rely on its standing but must instead compete for influence, these perceptions matter.
Looking for Fundamental Flaws not mere Anomalies
The Pentagon has announced that there will be an investigation into the 29 August strike. I don’t intend to speculate about the outcome of this investigation, but enough is already known to indicate that flawed procedures used for targeting drone in this particular strike, align with patterns of the past. The investigative work carried out by the New York Times investigative team, offers us these insights, in contrast to similar strikes of the past which left in their wake tragedy without resolution.
The Times investigation is particularly valuable because Government investigations tend to focus only on perceived anomalies, rather than fundamental flaws in accepted procedures. It is especially important now to eliminate those flaws that consistently result in civilian casualties because President Joe Biden himself initially held up this drone strike as an example of success in his so-called “over-the-horizon” strategy to respond to terrorist threats from afar.
To be clear, airstrikes don’t comprise a strategy at all. But since I am limited by word count, I want to focus here on mitigating the most adverse effects of a policy that I believe in any case is doomed to fail.
Why Civilians Die
I need to begin by explaining that there are three reasons why civilians are killed in air strikes. Misidentification – as appears to be the case in this latest incident -- is the first. The second is when civilians are in the target area, or subsequently move there, unobserved. Given time, drone operators have methods for mitigating this risk. The third reason - not a factor in this case – is when civilians are known to be in the area, but the risk to their lives is judged to be justified by the importance of the target. There is another ascetic term used to describe this justification: military necessity. The pressure to rush to a strike is more likely to cause error in all three cases.
Sometimes a rush is unavoidable – but with flaws in the current process, this leads to a huge margin of error. Mistakes of course will happen in any human enterprise, but we should learn from a pattern of similar mistakes in the past to prevent them happening again. Let me explain what I mean by this –in terms that aren’t too technical.
Deliberate vs Dynamic
There are two types of airstrikes - deliberate and dynamic. Deliberate are planned long in advance, have numerous checks, use a pattern of life analysis, and have a relatively low incident of civilian casualties. Dynamic strikes are usually conducted when the attacker has a small window of opportunity to engage a target. In dynamic targeting there is rarely time to conduct the full process of checks and balances that mitigate risk to civilians. Because of this margin for error, dynamic targeting amidst a civilian population is a method usually employed only in cases of imminent threat or for high value targets. Even so, the one feature of targeting that dynamic strikes should never bypass is positive identification – based on credible intelligence.
Given the circumstances, the use of time sensitive targeting for the 29 August strike was certainly justifiable – but the process was flawed. Imminence does not justify pursuing a procedure that is more likely than not to result in catastrophic error – which was to conduct a drone strike on the basis of very little information. Notice that I don’t use the term “intelligence” because uncorroborated information does not earn that title.
Reasonable Certainty?
The hunt for ISIS-K operatives reportedly planning an attack on Kabul Airport in the wake of a deadly suicide bombing at Abbey Gate, was conducted by MQ-9 Reaper drones flying wide race-track patterns above the city.
Based on the information that a white sedan could be used in the next attack, one of the drones began following a Toyota Corolla of the same color in one of the poorer parts of the city. Through the drone’s high-resolution cameras, the strike cell watched the vehicle as it made various stops and observed several men pack it with large bundles, which they believed to be explosives. From this pattern a narrative emerged – one that was quickly accepted by all in the chain of command. However, what the watchers believed was an ISIS-K operative turned out to be one Zemari Ahmadi, a longtime employee of a US aid organization, and what they believed to be a safe house turned out to be a family home, full of children. The explosives were water bottles.
After the strike, General Mackenzie said the decision to strike a white Toyota Corolla sedan, after having tracked it for about eight hours, was made in an “earnest belief” — based on a standard of “reasonable certainty” — that it posed an imminent threat to American forces at Kabul airport. Having been involved in similar situations, I know that these are tough calls to make. I have no doubt that all involved in the strike process were devastated by the result and it’s not my intention to rub salt in their wounds but do believe it’s worth unravelling this account, which we have to assume is accurate.
Anyone who has spent time in Afghanistan, indeed the Middle East, must be driven to question how “reasonable certainty” could be ascribed to a target description that loosely matched one in five vehicles on Afghan roads. Mackenzie’s statement reveals an opinion based more on wishful thinking than sound intelligence. When combined with high rank and a dominant personality, confirmation bias compounds initial errors, and causes other more junior but important participants in the process reluctant to challenge the views of a commander. The result is often what followed in this case – misidentification and tragic error.
From McKenzie’s own account, the intelligence was at best tenuous and despite 8 hours of surveillance by full motion video, there was apparently no attempt to confirm it from other sources. There may have been good reason for this – but McKenzie left this question on the table as though it wasn’t important enough to answer. Given that much time, it is unusual for an experienced commander to conduct a strike based on such a thin veneer of intelligence without corroboration.
A Rush to Kill
I can only imagine that there was intense pressure up and down the chain of command to make this strike happen. At the policy level, to make good on the President’s steely eyed vow to hunt down the killers of 13 US servicemen, to show resolve and competence as a counterweight to the shambolic scenes at Kabul airport which showed the exact opposite. And for those lower down the chain, it was a more visceral motive – the desire for vengeance, a need to dispel for a moment the pall of humiliation that weighed heavily on all of us who had served in that war. These were all understandable motives certainly – but emotion is corrosive and has no place in the targeting process. If left unchecked, it will cause a rush to failure. This is especially true when employing drones.
The Thing about Drones: A Dangerous Misconception
The number of drone strikes escalated sharply over a decade ago during the Obama administration. This preference was particularly pronounced for strikes in places such as Yemen or Pakistan – foreign and supposedly sovereign countries far beyond the reach of US ground forces. One reason for this trend was that using unmanned platforms did not involve risk to pilots. But there was also an implicit but clearly evident feeling that by distancing human beings from the act, we could somehow make the whole killing thing less distasteful -- even when American citizens were the target. This was around the same time when Presidential statements gave rise to the misconception that drones are more precise, less dangerous to non-combatants.
This remains a common misconception born from an overreliance on technology combined with a ton of wishful thinking. The truth is that drones are thirty times more likely to cause civilian casualties than manned aircraft.
By placing a human in the loop, with observation of the target, manned aircraft offer another opportunity to corroborate information. This advantage probably would not have made a difference in the Kabul strike since it is unlikely that a pilot would have seen anything more than the drone’s cameras. But it isn’t just the remote-control aspect of drones that makes them inherently more dangerous to civilians. During a drone strike, the process for analyzing and sharing key information is usually distributed among people who are not only geographically separated but working in different capacities, as if in silos. Sometimes one person will have critical information that obviates the validity of the target but because he is separated from the people controlling and directing the strike – it doesn’t reach them in time. Communication flow between these silos takes time – and breaks down when racing against the clock or countering other pressures such as confirmation bias.
In a recent tweet, Marc Garlasco, the former Chief of High Value Targeting on the Joint Staff, and now a leading analyst on non-combatant casualty mitigation, commented on an observation gleaned from two decades of experience with drone strikes: “I found that there was a pattern: in a set of cases, the fact that civilians were present in the target area was known by someone in the group of operators and analysts, but that information didn’t reach the commander making the decision to fire.”
All of these factors indicate that the process of dynamic targeting using drones should include rapidly executable checks that counter the obfuscating effects of confirmation bias and distribution. But there is much more that can and should be done.
Fixing the Problem
As DOD commissions two separate investigations into the situation and various Congressional committees conduct their own probes, it’s going to be important to maintain unity of effort based on the correct approach. So – from policy makers down to those directly involved in the process of hunting and killing by drone – here are some guidelines intended to keep the US military on the right side of the counter-terrorism equation.
Recognize that there is a Problem
How bad is the problem of mistaking civilians for valid targets? Well, estimates vary widely. Defense experts acknowledge that the official Department of Defense estimates are probably low, so to get a better perspective of the problem one has to look at those of the various non-profit watch dogs that track such figures.
The Pentagon acknowledges that U.S. airstrikes have killed more than 1,400 civilians across Iraq and Syria, since operations against ISIS began in 2014, but insiders admit that this data doesn’t come from any stringent procedure for collection.
Seth Jones a defense analyst at the Center for Strategic and International Studies, commented recently “Even when I was in government and had access to top-secret information, I do not ever remember seeing an analysis on the costs and benefits of strikes.” It would be hard to find someone better placed to know. Jones was for several years a senior advisor to US Special Operations Forces in Afghanistan, and subsequently worked in the Pentagon, as the Commander of US SOCOM’s liaison to the Assistant Secretary of Defense for Special Operations.
Airwars, a reputable London-based nonprofit watchdog that tracks civilian casualties in war zones, estimates about 29,000 innocent people have been killed by US-led air strikes. It’s important to note that this figure reflects civilian casualties from both manned aircraft and drones across all theaters.
Even these last numbers pale in comparison to civilian casualties in wars that the United States has fought in the past – or to those inflicted by countries such as Russia and Saudi Arabia in recent conflicts. But do we really want to set the bar that low? If we have the means to narrow that margin for error why not take it – and do so transparently? For the US military, seizing the high ground should be as valuable a moral imperative as it is a tactical one.
And let’s not forget that there are pragmatic reasons for doing so. The “all’s fair in war” argument backfires when a civilian population seeks revenge against a foreign military that treats it with ruthless disdain. Being subjected to apparently random acts of devastating violence tends to nurture a desire for revenge.
Such an approach only serves to make terrorist organizations more violent and resilient while ensuring that they have no shortage of recruits from among a populace whose only knowledge of the United States, is the occasional bolt from the blue that claims innocent lives. Allegations of civilian deaths from drone strikes are widely covered by the regional media outlets and become a rallying cry for our enemies.
For those still unconvinced that the US should care about killing civilians overseas, a recent example offers sobering lessons.
A decade of US drone strikes in Pakistan which killed hundreds, maybe thousands of Pakistani civilians is attributed by experts as being a leading cause of anti-US feeling in that country. It would be hard to argue that alienating the population of an entire nation comprising some 226 million Muslims, a nation armed, incidentally, with nuclear weapons, could be anything but inimical to US interests. And as it happens now, the United States could really use having a better relationship with that particular country.
Accepting then that there is a problem, how do we solve it. I am going to focus here on improving the manner in which the US military conducts investigations. Although these occur by definition “right of bang”, fixing the problem of who conducts them and how will enable Washington to identify patterns and prevent future incidents.
Widen the Focus and Composition of Investigating Teams
An inherent flaw of the way that investigations are currently conducted is that they tend to focus on a single incident, which leads them to look for anomalies or departures from accepted procedures instead of examining that incident in its wider historical context. Fixing this will entail changing the mandate of such investigations, to ensure that they do the hard work to identify patterns. But it will also entail modifying the procedures for selecting the members of these teams together to ensure that they encompass a wider pool of expertise while providing depth and continuity of experience.
One method of doing this might be to establish a pool of experts to form fly-away teams after an incident. This pool should include non-DOD civilians to pre-empt perceptions of bias and opacity.
Many, if not most, investigations do not visit the scene of the strike or talk to witnesses. There are a number of good reasons for this, derived from understandable concerns about force protection and operational security. But these obstacles can be overcome if DOD is willing to find creative solutions. One solution would be to ensure that all those experts assigned to the pool of investigators are granted the appropriate level of clearance. There are probably also a number of non-DOD personnel with subject matter expertise who already have such clearances.
There are options available to ensure an equally thorough process for those locations which pose too great a risk for US personnel to visit. All members of the pool of experts should be trained in how to conduct remote investigations. If the New York Times has this capability –the US Government should too.
Another work-around should involve seeking help from organizations outside DOD – from the media, NGOs and organizations such as the UN. There are other reasons why this would prove helpful. Such contacts have an added advantage, because they can provide information to assist an investigative team beyond the intelligence that led to the strike – intelligence that may, as we have seen, been flawed in the first place. But making this coordination common practice will entail a concerted effort to change organizational culture.
Fix the Culture
Dr. Colin Jackson is Chairman of Strategic and Operational Research department at the US Naval War College, a former high-level Pentagon official, and renowned expert on counterinsurgency. When questioned in a recent interview about what makes a military unit good at counter-insurgency operations, he replied “Intellectual humility” before going on to define this concept as a willingness to embrace the views of outside agencies. In my experience, however, intellectual humility is hardly a main-stay of US military culture.
One indicator: military officers tend to treat members of the media with a curious mix of condescension and barely concealed hostility. Understanding that transparency serves our cause better than obfuscation, the US military should try instead to find common ground, and a desire to discover the truth should be central to that common ground. How refreshing it would have been to see DOD reach out to the New York Times investigative team in the aftermath of the 29 August strike, instead of engaging in a battle of the narrative – a battle in which the Pentagon rapidly found themselves on the losing side.
The supercilious attitude that military officials display towards the media is also an implicit theme of their dealings with Non-Governmental Organizations. Whereas reporters are seen as being inherently inimical to US national interests, NGO members are Granola Eating Tree Huggers – despite the fact that these same people routinely incur far greater risk than US military personnel who are accustomed to operating only within a cocoon of force protection.
The reality is that such organizations are comprised of dedicated, intelligent, and courageous men and women with expert knowledge of local conditions which often surpasses in depth and context anything gleaned by military intelligence sources. Even were this not the case, the military only hurts itself by not working to resolve misperceptions that naturally arise between two very different cultures.
Many reputable NGOs and Non-Profit watch dogs see the Pentagon as being purposefully deceitful and – where it comes to the expenditure of human life that is not American – criminally negligent. This latest incident reinforced that perception.
“ The U.S. military’s initial, stubborn insistence that this was a ‘righteous strike’—despite compelling and immediate evidence to the contrary—raises fundamental questions about why U.S. military commands so routinely ignore reports of tragedies from affected communities,” says Chris Woods, director of Airwars. “It can’t be right that civilians only appear to have accountability when the investigative might of U.S. media giants is brought to bear.”
Woods is not completely off-target. Over the course of a 31-year career to include multiple combat deployments I have observed an implicit but nevertheless prevalent belief in the US Military that American life is inherently more important than Afghan, or Iraq or Syrian life. It’s a belief which underlies our policies and attitudes – and saps our credibility. I like to believe that the institution that I served for over 31 years is better than that, but we need to work to make it so.
Make Reparations a Reality
Civilian deaths inevitably leave in their wake a desire for vengeance – perfect fodder for extremist groups. A process that results in rapid and fair reparations for victims will do much to undercut this tendency. In the aftermath of the 29 August strike General McKenzie apologized for the error and said the United States is considering making reparation payments to the family of the victims. Based on the US military’s track record, “considering” may be the culminating point of this process.
The US Congress earmarks $3 Million annually for ex gratia payments to war victims. In practice only a fraction of this amount is disbursed to victims and their families. The reason for this discrepancy is not indicative of a low number of civilian casualties but rather of flaws in the underlying policies and methods of investigation. Reform would be relatively simply. The policy should be re-written so as not require proof in each instance that ex-gratia payment will benefit US policy. And, as discussed, the manner in which investigations are conducted is not as rigorous as it could be.
This would be a good opportunity to make good on McKenzie’s promise, and in so doing examine why such payments often falter. The family concerned has already said with understandable outrage that an apology will not be enough. With the Middle Eastern media continuing to follow this tragedy, the world is indeed watching.
Armed Overwatch: Putting a Human in the Loop
The same technology that enables the United States to lead the world in drone technology should also be applied to fixing their flaws and finding alternate solutions to the problem of undermining extremist organizations. I have written elsewhere about what such solutions might entail – but it is worth here looking through a narrower aperture at preferred platforms from which these strikes are conducted.
We have determined why manned platforms are preferable to drones when it comes to avoiding civilian deaths – but there simply aren’t enough manned aircraft in the US military’s inventory to allow their exclusive employment for strikes. Fortunately, US Special Operations Command is pursuing an “armed overwatch” program that will provide manned platforms that are robust, expeditionary and manufactured in enough quantity to complement drones. Whilst, not a solution in itself, this will give commanders another more reliable tool for target identification.
Machines not Humans are the Automatons
Wearing a uniform should not relegate anyone to being a mere cypher – from four-star generals whose responsibility it is to inform policy makers about the disadvantages and risks of targeted killings to junior drone operators and intelligence analysts who should be encouraged to avoid compartmentalizing information. Speaking up to avoid mistakes is a practice must be inculcated into targeting in the same way as it is into training safety.
Conclusion
The sad finale to US involvement in Afghanistan continues to hang like a pall over Washington’s foreign policy. From this disaster, however, we should glean lessons that will enable the United States to do better going ahead.
The needless killing of civilians on 29 August was a tragic mistake – but like so many such mistakes, not isolated by cause or effect. Only by understanding this context, can we fix a problem that has long term ramifications, and – if corrected -- will continue to plague US Foreign Policy.
“‘At first, there was no support for the Taliban,’ commented Mullah Omari, a Taliban military commander, in Afghanistan’s Wardak Province. ‘It was when the Americans started killing civilians that people started supporting us, giving us food, bullets, and offering men.” Quod Erat Demonstrandum.