Centcom Report on the Kunduz Hospital Attack: Accounting for a Tragedy of Errors

The U.S. military Central Command (Centcom) report on the attack on the Medecins Sans Frontieres (MSF) hospital at Kunduz, Afghanistan demonstrates how a perfect storm of mistakes and mishaps can lead to tragic results. While military rules of engagement (ROE) seek to cabin the use of lethal force, aggressive commanders in the heat of battle can fail to heed that guidance. Moreover, even the best technology can fail. The errors made during this distressing episode killed at least thirty civilians, and led to no corresponding military advantage. From the comfort of hindsight, it is tempting to assert that such grievous harm should surely result in criminal prosecutions. However, a careful look at the relevant law significantly narrows prosecution options. That said, the Centcom summary of the decision to forego criminal charges relied on an inaccurate premise: that intent is a necessary element of a war crime. That inaccurate premise may undermine global confidence in Centcom’s process for determining accountability.

To understand these dual points, it is useful to first consider the laws of armed conflict (LOAC) and provisions of the Uniform Code of Military Justice (UCMJ). First and foremost, LOAC includes the principle of distinction, which requires a commander to take reasonable steps to ascertain that the intended target of an attack is a combatant or a military objective, not a civilian or a protected site such as a hospital (as long as such sites have not been commandeered by an adversary for military purposes). Second is the principle of proportionality, which bars “excessive” harm to civilians, given the military advantage that a commander anticipates from an attack. Third is the principle of precautions in attack, which requires a commander to take all “feasible” precautions to minimize harm to civilians, even when an attack would otherwise comply with the principles of distinction and proportionality.

There is some daylight between violation of these principles and criminal liability. The Rome Statute of the International Criminal Court (ICC), for example, does not address violations of the principle of precautions in attack. However, the Rome Statute is not the last word; LOAC is a body of customary international law (CIL), and a wholesale failure to take even rudimentary precautions might well violate CIL. Contrary to Centcom’s position, LOAC does not require intent as an element in the commission of war crimes. Often knowledge will be sufficient. See Rome Statute article 8(2)(b)(iv) (criminalizing attacks committed with the knowledge that harm to civilians would be “clearly excessive”). Indeed, recklessness—an act or omission engaged in despite awareness of a substantial risk of harm—may well suffice.

The UCMJ does not contain offenses that correspond exactly to violations of the LOAC principles. Nevertheless, crimes outlined in the UCMJ in many instances entail elements that lend themselves to the prosecution of LOAC violations. For example, murder is a UCMJ offense. Murder (article 118) requires intent, or, at the very least, “wanton disregard” for human life. But intent is not a prerequisite for all UCMJ charges. For example, another UCMJ charge is manslaughter (article 119), which requires “culpable negligence” that is the proximate cause of death. Culpable negligence amounts to recklessness—knowing disregard for a substantial risk. Article 134 of the UCMJ includes the offense of negligent homicide, which requires both proximate cause and the lesser showing of “simple negligence.” (The latter is closer to a tort standard of lack of reasonable care in discerning and addressing a risk.)

Consider the facts in the Kunduz MSF attack against this legal background. The attack occurred at night, when visual identification of targets is more difficult. Moreover, the attack occurred at the tail end of four days of heavy fighting. These facts do not excuse violations of LOAC, but they do highlight the need to consider decisions from a commander’s perspective at the time.

Two mishaps that were clearly not criminal in nature occurred early on in the night mission, and those mistakes paved the way for the tragedy to follow in Kunduz. First, the attacking aircraft—an AC-130U with extensive firepower on board—took off early because of a report of U.S. troops being attacked. Due to the haste of the aircraft’s departure, there was no time to upload the “No-Strike List” (NSL) to the aircraft’s computers. That list would have provided definitive coordinates on the MSF facility’s location. Second, well before the attack on the MSF facility began, the U.S. aircraft’s satellite radio—its data link—failed. While the aircraft still had radio contact with ground forces, it could not send or receive emails or upload data, such as the NSL.

This communications failure resulted in mutual mistakes about identification of targets by the aircraft crew and ground commanders. Special Operations forces on the ground wanted to target a prison overrun by the Taliban, formerly run by the Afghan government’s National Directorate of Security (NDS). However, because of the lack of a data upload capacity and imprecise descriptions at both ends of communications, the aircraft crew mistook the MSF facility for the former NDS prison. For example, ground forces informed the crew that the intended target had an “arch-shaped gate.” While this description matched many buildings in the area, the crew took it as a match for the MSF facility. The ground force commander (GFC) did not seek clarification, and had no independent visual ability to confirm the crew’s judgment. In other words, the GFC and the aircraft crew were actually labeling entirely different structures as being one and the same. To avoid a tragedy, the crew and ground command would have had to be on “the same page.” Sadly, the crew and ground command were on different pages until the very conclusion of this disturbing narrative.

The dialogue that took place for the hour in which the crew observed the target prior to the attack exhibited confusion by both the crew and the GFC. For example, the crew repeatedly asked the GFC to confirm that the intended target was a “large t-shape building.” The GFC (or his inexperienced subordinate, the Joint Terminal Attack Controller (JTAC)) confirmed this description, which matched many buildings in the vicinity. Unfortunately, the crew and the GFC were still unknowingly talking about two different buildings: the crew “had eyes” on the MSF facility, which was marked with MSF emblems but not the more familiar red cross or red crescent symbols used globally to label medical sites, while the GFC believed that the crew was describing the intended target: the NDS prison held by the Taliban. The JTAC did not help matters with instructions such as “soften the target,” which did not correspond to the situation the crew was viewing in real time.

This communications failure was compounded by the inappropriately aggressive posture of the GFC, whom the military report described as having “willfully violated” the ROE. The US ROE restricted the use of air power, except in response to a hostile act by the Taliban directed at U.S. forces. However, the GFC ordered the attack on the MSF facility although the GFC “could not have reasonably believed” that the attack was justified by an ongoing hostile act. While the GFC asserted to military investigators that he saw what he believed to be an attack on a friendly military convoy, the Centcom report viewed that assertion as inconsistent with other sources, including aircraft video, radio transcripts, and tracking data. Moreover, the military’s report indicated that the GFC, because of distance from the convoy, could not have had the line of sight that he claimed. In sum, even an attack on the “right” target—the NDS facility—would not have been a response to hostile fires, as the ROE required. Because of the GFC’s aggressive posture and the mutual misidentification of the MSF facility as the “right” NDS target, the attack commenced.

The communications failure also figured prominently in U.S. acts and omissions when the attack was in progress. Shortly after the attack started, MSF representatives contacted U.S. commands, imploring them to stop the fires. However, the communications failure and the ground-air misunderstandings severely impeded a timely U.S. response. About 12 minutes into the attack, personnel at the U.S. Special Operations Task Force inquired about the coordinates of the target being engaged. Two minutes later, U.S. personnel contacted the crew and sought to confirm that the attack had not harmed the MSF facility. The crew stated that the MSF facility hadn’t been touched, reporting that the only structures affected were the “T-shaped building” or adjacent structures, which the GFC had earlier mistakenly identified as the NDS prison, although in reality the crew was still describing the MSF hospital. It took several more minutes to ascertain that the building being attacked was in fact the MSF facility. Once that awful realization took hold, the attack ceased. However, the crew had already fired a deadly mix of high-explosive and anti-personnel rounds at the hospital, causing horrific damage and loss of life.

The military’s report details the errors that dogged each step in this appalling episode. Not unlike the critical moments of Ian McEwan’s novel Atonement, missteps and misunderstandings at several junctures paved the way for an unspeakable tragedy. However, the report is correct that these missteps do not rise to the level of war crimes (see Jens David Ohlin’s analysis here, which also urges formulation of a new war crime to cover cascading mistakes like those that led to the MSF attack, and Alex Whiting’s analysis here). The report’s comprehensive account makes clear that until the conclusion of the attack, neither ground commanders nor the air crew knew that the MSF hospital was under attack. That lack of knowledge resulted from a series of mishaps and errors which in combination caused enormous harm. The crew’s lack of access to the no-strike list played a role, as did the failure of its satellite communications system. So did ground personnel’s failure to confirm that the crew had not received the emailed no-strike list sent after the AC-130 had embarked on its ill-fated mission—if ground personnel had known that the NSL email had not been received, they could have conveyed the information by other means, such as real-time radio. The JTAC’s imprecise language also played a role, since the use of phrases not found in U.S. military doctrine, such as “soften the target,” contributed to the crew’s confusion. The GFC’s violation of standing ROE also helped cause the attack. Finally, the mutual misunderstanding of the target’s identity that emerged in the run-up to the attack also impeded efforts to stop the attack once it had started.

Contrary to Centcom’s position, the lack of intent among U.S. personnel is not determinative. As noted above, both CIL and the UCMJ allow liability in intent’s absence. However, the lack of concrete and specific awareness of a substantial risk of harm to civilians would greatly complicate proof of a war crime. While some legitimately worry (see Ryan Vogel’s comment here) that the ICC may step in to file charges against U.S. personnel, the ICC needs to first answer the question, “What did U.S. personnel know, and when did they know it?” U.S. forces as a whole may have had all the information required; but in a large organization, getting that information to the right people is often difficult, particularly in the fog of war. That reality requires both better systems, which Centcom has already implemented, and getting rid of mistake-prone personnel by relieving certain individuals of command and issuing career-ending reprimands, which Centcom has also done. However, the cascading systemic errors in the hospital attack impede the attribution of culpable awareness to one or more specific individuals. Rather, confusion and chaos seemed to have carried the day. That’s a cautionary tale for the U.S. military, but one that makes individual criminal responsibility under LOAC far more difficult to establish.

The organizational context of the errors in the Kunduz episode also undermines UCMJ liability. Any student of tort law knows the centrality of proximate cause. To satisfy this standard, an error or omission must play a substantial role in the harm that occurred. The litany of missteps related in the report make it difficult, if not impossible, to identify individual errors that supply the requisite substantial causation.

However, in articulating its justification for not referring individual cases arising in this matter to courts-martial, Centcom’s reasoning was flawed. Centcom would have been better served by acknowledging that intent was not required, but that awareness of risk was distributed among many organizational components, without full awareness concentrated in one or more individuals who could be charged criminally. Centcom should also have discussed problems in proving proximate cause for particular individuals. Centcom’s reliance on lack of intent was an imprecise surrogate for this more methodical analysis.

Moreover, Centcom did not address another obvious, if ancillary, UCMJ charge that would be consistent with the report: charging the GFC under Article 107 with making false official statements to investigators about the basis for the intended attack on the NDS facility. Case law is clear that statements made to military investigators count as “official” statements under the UCMJ, since they directly relate to the military’s need to ascertain the facts concerning service members’ conduct (United States v. Solis, 46 M.J. 31, 32-33 (CAAF 1997)). The Centcom report indicated that video, radio, and tracking data flatly contradicted the GFC’s claim that there had been an attack on a friendly convoy that under the ROE would have justified a U.S. response.

In addition, while the GFC claimed that he saw an attack, investigators concluded that the GFC’s line of sight made this “impossible.” Indeed, military investigators concluded that the GFC “could not have reasonably believed that a hostile act warranting engagement” under the ROE had occurred. Based on these findings, the report found that the GFC had “willfully violated” the ROE, thus supplying a motive for false statements to investigators to cover up this violation. It’s hardly a big leap to infer deception from a strong motive and the lack of any reasonable basis for the statements made.

That said, pursuing such a charge might not have been worth the candle, and military lawyers may well have so informed Centcom. Prosecutions based solely on false statements are rare. At best, a conviction on such a charge would have led to minor punishment. The discipline meted out here will ensure separation from the military—the functional equivalent of the maximum punishment that could be expected.

In conclusion, the Centcom report painstakingly detailed a parade of errors. That’s one parade in which no member of the armed forces wishes to participate. Moreover, the discipline the military handed out will send a strong message to commanders who value their careers. This resolution lacks the emotional resonance of criminal charges, but in the final analysis may be more effective in promoting constructive change.