3.1 Unruly Civilian Mobilities: The ICTY Prosecutor’s Decision

From 24 March 1999 to 9 June 1999, the North Atlantic Treaty Organization (NATO) conducted a bombing campaign against the Federal Republic of Yugoslavia (FRY) that included more than 38,000 combat sorties, among them 10,484 strike sorties releasing 23,614 air munitions, killing between 400 and 600 civilians.^{Footnote 53} In 1993, the UN Security Council, with assent of key NATO states, had established the ICTY with “the power to prosecute persons responsible for serious violations of international humanitarian law committed in the territory of the former Yugoslavia since 1991.”^{Footnote 54} The architects of the Tribunal had envisioned that it would prosecute acts of violence by different armed forces and paramilitary groups from the states of the former Yugoslavia. Yet by bombing the same territory six years later, some of the states that drafted the treaty brought themselves under ICTY jurisdiction.

On 14 May 1999, while the NATO bombing campaign was still ongoing, the ICTY prosecutor took the unusual step of convening a committee to “assess the allegations” against NATO states in response to “numerous requests.”^{Footnote 55} The committee’s report was issued on 13 June 2000 and recommended no prosecutions.^{Footnote 56} The ICTY prosecutor accepted the report. The committee focused on twenty-one specific incidents with particularly high civilian casualties and reviewed the five “most problematic” incidents in detail.^{Footnote 57} Of these five incidents, one involved a target misidentification (the Chinese Embassy was mistaken for a Yugoslav government building), one involved a target with a disputed military status (the attack on Radio Television Serbia), and the remaining three strikes raised questions about the visual identification of civilians from the air. I will focus on these cases of civilian misidentification.

First, on 12 April 1999, a NATO aircraft launched two laser guided bombs at a railway bridge over the Grdelica gorge just as a passenger train crossed the bridge.^{Footnote 58} The strike was aimed at the bridge, which was classified as a military target. It killed at least ten and wounded at least fifteen passengers of a civilian train that moved through and across the target area.^{Footnote 59} In press conferences, NATO officials stressed that the officer controlling the weapons “caught a flash of movement that came into the screen” and “suddenly the train appeared.”^{Footnote 60} In the committee’s assessment, “The crosshairs remain fixed on the bridge throughout, and it is clear from this footage that the train can be seen moving toward the bridge only as the bomb is in flight: it is only in the course of the bomb’s trajectory that the image of the train becomes visible.”^{Footnote 61} The officers later decide to drop a second bomb on the other end of the bridge, realizing too late that the impact of the first bomb had caused the train to move, resulting in a second hit on the train rather than on the bridge. While General Wesley Clark called the incident an “uncanny accident,”^{Footnote 62} analysts have suggested that the release of the second bomb was in violation of International Humanitarian Law because at that point the presence of civilians in the target area was obvious.^{Footnote 63} In addition, it became apparent that the video of the strike that NATO had shown had been artificially sped up to 3 or even 4.7 times the speed of the original incident. While NATO explained the sped-up version of the video as a result of a “‘technical phenomenon’ rather than human manipulation,” the undisputed effect of the presentation was that the available reaction time appeared shorter than it had actually been.^{Footnote 64} The committee reporting to the ICTY prosecutor, taking note of these issues, decided that the first strike did not warrant an investigation. Regarding the second strike, the report notes that some committee members saw an element of recklessness, but the committee nonetheless decided by a narrow margin not to recommend an investigation.^{Footnote 65} In the judgment of the committee, the civilians, whose presence was unexpected and fleeting, especially when viewed in an accelerated video, were not sufficiently visible to the pilot and the weapons systems officer, rendering their deaths incidental rather than deliberate.

Second, the attack on the Djakovica Convoy occurred on 14 April 1999, when NATO aircraft dropped bombs on a convoy of about 1,000 Albanian refugees. About seventy to seventy-five civilians were killed, and one hundred were injured.^{Footnote 66} After initial denials of responsibility, NATO officials “confirmed that the aircraft had been flying at an altitude of 15,000 feet (approximately five kilometres) and that, in this attack, the pilots had viewed the target with the naked eye rather than aided by technology. The aim of the attack was to destroy Serb military forces, in the area of Djakovica, who had been seen by NATO aircraft setting fire to civilian houses.”^{Footnote 67} Witnesses reported that the refugees had passed a Serb military caravan shortly before the attack. NATO deflected responsibility for the deaths of the Albanian refugees: “NATO regrets any harm to innocent civilians, and reminds that the circumstances in which this accident occurred are wholly the responsibility of (Yugoslav) President (Slobodan) Milosevic and his policies”^{Footnote 68} that had turned the people who were killed into refugees. The committee reporting to the ICTY prosecutor was convinced that “civilians were not deliberately targeted in this incident.”^{Footnote 69}

Third, the attack on Koriša Village took place on 14 May 1999. NATO aircraft dropped ten bombs on the village, killing as many as eighty-seven civilians and wounding another sixty persons, most of them refugees.^{Footnote 70} According to the committee, “the primary target in this attack was asserted by NATO to be a Serbian military camp and Command Post located near the village of Koriša. It appears that the refugees were near the attacked object.”^{Footnote 71} NATO officials insisted that they had hit a legitimate military target and that they “knew nothing of the presence of civilians and that none were observed immediately prior to the attack.”^{Footnote 72} Witnesses described that the village did not have military personnel or objects aside from a few soldiers who had been billeted there.^{Footnote 73} NATO officials, in contrast, suggested that “displaced Kosovar civilians were forcibly concentrated within a military camp in the village of Koriša as human shields and that Yugoslav military forces may thus be at least partially responsible for the deaths there.”^{Footnote 74} In the immediate aftermath of the attack, a NATO spokesperson said “If there are civilians at a military target, we didn’t put them there.”^{Footnote 75} The other party in the conflict was blamed for unexpected civilian presences and mobilities. The committee determined that, on the basis of the “credible information available,” there was not sufficient evidence “to show that a crime within the jurisdiction of the Tribunal has been committed” in this case.^{Footnote 76}

How did visual regimes and visual technologies figure in the committee’s assessment of the cases? Since a crime within ICTY jurisdiction would require intent, these cases hinged on the committee’s review of what the pilots saw, could have seen, or should have seen before they released the bombs. The committee heavily relied on NATO’s own descriptions and assessments of the strikes, quoting NATO press releases verbatim at length, and allowed its own viewing practices to be guided and instructed by the concepts and evaluations that NATO officers had used. Two specific arguments explaining civilian casualties stand out. First, the presence of civilians is consistently characterized as unexpected. In all three cases, civilians—many of them displaced persons or refugees—moved through or near a target area that had been neatly mapped out in advance as a military target. When civilians appear in spaces “known” through maps and aerial views of landscapes and military installations, their presences can be portrayed as surprising.

Second, the use of high-end technologies has become the military’s equivalent to claims of due diligence in law regardless of how these visual technologies were used. For most of the bombing campaign, NATO planes flew at a very high altitude in order to evade the Yugoslav air defences.^{Footnote 77} In the committee’s view, this was not a problem: “The 15,000 feet minimum altitude adopted for part of the campaign may have meant the target could not be verified with the naked eye. However, it appears that with the use of modern technology, the obligation to distinguish was effectively carried out in the vast majority of cases during the bombing campaign.”^{Footnote 78} Yet in at least one of the cases (Djakovica), the pilots did not rely on visual technologies for the identification of civilians. In the committee’s assessment, the availability of these technologies—regardless of their specific usage—suffices for the judgment that NATO took adequate precautions.

3.2 Legal and Other Spaces: Banković

After the ICTY Prosecutor announced her decision to not pursue charges for any of the NATO air strikes, victims of the bombing of Radio Television Serbia (RTS) filed a complaint against Belgium and sixteen other signatories of the European Convention of Human Rights that had participated in the NATO mission.^{Footnote 79} On 23 April 1999, NATO had bombed the central studios of RTS, a state-owned broadcaster, killing between ten and seventeen people.^{Footnote 80} Two legal questions about the legality of the strike were central: First, was this a legitimate target? NATO considered the TV and radio channels appropriate targets since they broadcast war propaganda,^{Footnote 81} but this was disputed.^{Footnote 82} Second, were the civilian casualties disproportionate? The committee reporting to the ICTY Prosecutor found the number of deaths to be “unfortunately high” but stated that they “do not appear to be clearly disproportionate.”^{Footnote 83}

On 12 December 2001, the ECtHR ruled that the application was inadmissible because the acts had not occurred within the jurisdiction of the European Convention of Human Rights (ECHR).^{Footnote 84} Article 1 of the ECHR declares: “The High Contracting Parties shall secure to everyone within their jurisdiction the rights and freedoms defined in Section I of [the] Convention.” Does “within their jurisdiction” refer to state territory or to the reach of state action? The Court had previously held that contracting states are required to secure the Convention rights for persons under their control even when acting extraterritorially as long as they exercise “effective overall control.”^{Footnote 85} But does dropping bombs constitute effective control?

The applicants submitted that they “were brought within the jurisdiction of the respondent States by the RTS strike.”^{Footnote 86} They argued that extraterritorial obligations to secure Convention rights should be “proportionate to the level of control in fact exercised.”^{Footnote 87} The effect should be that aerial bombardment would not imply an obligation to “secure the full range of Convention rights,” but that it would require the states carrying out the attack to observe the right to life and physical integrity.^{Footnote 88} The applicants also argued that the respondent states in fact exercised control: not of the territory on the ground, but of the territorial airspace.^{Footnote 89}

The respondent states argued that jurisdiction “generally entails some form of structured relationship normally existing over a period of time.”^{Footnote 90} Jurisdiction is configured as based on temporally stable control of ground-level territory. The states also cautioned that a reading of jurisdiction departing from the territorial model would “risk undermining significantly the States’ participation” in “military missions all over the world,”^{Footnote 91} hinting at the connections between extraterritorial use of force and restrictions on extraterritorial jurisdiction. Insofar as they had control of the air space, the governments argue, such control could not be “equated with … territorial control.”^{Footnote 92}

The Court largely sided with the respondents, declaring, “the jurisdictional competence of a State is primarily territorial.”^{Footnote 93} Acts “performed, or producing effects” outside the territories of the Contracting States have been found to “constitute an exercise of jurisdiction” only “in exceptional cases.”^{Footnote 94} While the occupation of a territory with ground troops would constitute “effective control of the relevant territory and its inhabitants,”^{Footnote 95} the NATO presence in the former Yugoslav airspace did not meet this standard. The judgment expresses unease at extending the reach of the ECHR: “The Court considers that the applicants’ submission is tantamount to arguing that anyone adversely affected by an act imputable to a Contracting State, wherever in the world the act may have been committed or its consequences felt, is thereby brought within the jurisdiction of that State for the purpose of Article 1 of the Convention.”^{Footnote 96} In short: ECHR signatories would have to respect the human rights of everyone with whom they interact—a result that appears compatible with the theory that human rights are universal and inalienable. Yet this regional human rights court expressed unease about the universalism of human rights and instead drew sharp boundaries: “the Convention is a multi-lateral treaty operating … in an essentially regional context and notably in the legal space (espace juridique) of the Contracting States. The FRY clearly does not fall within this legal space. The Convention was not designed to be applied throughout the world, even in respect of the conduct of Contracting States.”^{Footnote 97} The ruling has been criticized for drawing territorial borders around putatively universal human rights, thereby confirming the European tendency to grant right to those within, but not those outside, Europe.^{Footnote 98}

The work of jurisdiction, as Mariana Valverde argues, is to sort “governance processes, knowledges, and powers into their proper slots as if by magic,” enabling different modalities and rationalities of governance to co-exist without “overtly clashing.”^{Footnote 99} The distinction between metropolitan and other spaces has long been a feature of European law. Indeed, Art. 56 ECHR, on which Banković court relied when it rejected jurisdiction for extraterritorial aerial violence, has been described as a “colonial relic” originating in “British imperial constitutional arrangements.”^{Footnote 100} The clause might well be a relic, but it functions smoothly in the ECtHR “chronotope”^{Footnote 101} or spatio-temporal ordering of jurisdiction that separates the legal space of European human rights from the space in which NATO commits airborne violence. While “the entire purpose for NATO’s involvement in Yugoslavia was to defend the human rights of people in this foreign land,”^{Footnote 102} the human rights mode of governance is not extended to FRY while it is under bombardment. Air strikes have a specific spatio-temporal order: the plane enters the airspace but does not touch the ground of the state in which its targets are located. The only point of contact between the state sending the plane and the victim is the violent encounter between the bomb and the victim. The ECtHR’s chronotope of jurisdiction is not simply spatially exclusionary, but, in an act of legal anaesthesia, it specifically excludes Western airborne violence from juridical consideration.

4.1 Jurisdiction: War and the Limits of German Law

Can individual victims of German military action abroad sue the state for compensation? In the Kunduz case, all three courts ruled that international law does not establish an individual right to redress for violations of the law of armed conflict. The trial court and the appellate court found that the clause in the German Civil Code that allows individual compensation for wrongdoing by state officials would apply to the Kunduz air strike, but both courts established that there was no relevant wrongdoing. In 2016, the BGH ruled that the civil compensation clause would not apply to military action abroad. The Supreme Court referred to its 2003 decision that had denied compensation to Greek victims of 1944 Nazi atrocities. In the Distomo judgment, the court had established that at the time of the atrocities, “war was considered a state of exception in international law” that allowed for “the collective use of force,” thereby “suspending the legal order that is in force in times of peace.”^{Footnote 108} This was a debatable assessment of international law in 1944. The Supreme Court had the option of recognizing changes in the international legal conception of war since the 1940s or referring to new obligations arising from the 1949 Constitution. Nonetheless, in Kunduz, the Court relied on the Distomo case in its judgment that the civil compensation clause was not applicable to war since the drafters of the 1900 Civil Code had not contemplated this possibility.^{Footnote 109} In addition, it found that military action is different from “normal administrative action,” for which the law was designed.^{Footnote 110} The Supreme Court places the Kunduz air strike beyond the territorial and functional jurisdiction of German courts. Here, the “machinery of jurisdiction”^{Footnote 111} works to keep the normal administrative functioning of the state legally separate from the extraordinary military action abroad, thereby insulating the state from compensation claims arising from the actions of the Armed Forces and maintaining a strict separation between spaces of rights at home and spaces of military violence abroad.

4.2 Reading Dots

The German courts that asserted jurisdiction over the case had to decide whether German soldiers had violated International Humanitarian Law (IHL). Did they expect civilians near the trucks? Should a reasonable military commander have known or expected the presence of civilians? The Attorney General as well as the civil courts found that Colonel Klein did not know and could not have been expected to know that civilians were present in the target area. The Federal Supreme Court summarizes the prevalent approach to sight, knowledge, and care: “Whatever a decision maker ‘doesn’t see’ despite careful and conscientious examination of the available sources of information, they ‘don’t need to see’ and take into consideration.”^{Footnote 112} How was this work of carefully not seeing civilians accomplished?^{Footnote 113}

In a first step, the decisions provide a specific narrative of the events leading up to the air strike. The Notice emphasizes that the German Forces in Kunduz had been attacked almost daily since May 2009.^{Footnote 114} In this narrative frame of constant danger, specific places are coded as hostile. The sandbank on which the trucks were stuck is presented as a location “frequently used by Taliban.”^{Footnote 115} Against the backdrop of a string of Taliban attacks using vehicles, the fuel trucks are understood to be bombs in the making. The trucks, the sandbank, and the landscape are primed as hostile and dangerous. These qualities will transfer onto persons seen in these spaces.

Moreover, the decisions adopt NATO’s expansive vocabulary for non-civilians, including “Taliban” and “hostile persons.”^{Footnote 116} The informant’s assessment that all the persons near the river were Taliban is based on this expansive and imprecise category that includes political supporters of the Taliban who don’t participate in armed conflict and who would not be legitimate targets under IHL.^{Footnote 117} Seeing civilians requires not only taking notice of persons, but also making judgments about their status. The shifts in the vocabularies towards more expansive concepts of non-civilians directly impact the ability of soldiers to see people as civilians. For example, the local informant was never asked about the definition of “Taliban;” his (translated) unquestioned vernacular vocabulary became the basis for the air strike.^{Footnote 118}

In a next step, the Notice and the courts validate Commander Klein’s stated assumption that the persons on the ground were “insurgents” and therefore legitimate targets. The source of this assumption of hostility is not clear, but the record shows that Klein and his colleagues imagined that the people close to the trucks would be “involved” in the insurgency. Klein is quoted with a number of reasons why he did not expect civilians, particularly children, at the site: it would be unusual for civilians to leave their houses in the night during Ramadan, and the Taliban would claim the space around the sandbank as theirs, making it unsafe for civilians to be in this area.^{Footnote 119} Once this assumption of hostility is established, any additional information can be interpreted to confirm rather than challenge it.

Martina Kolanoski’s incisive ethnography of the trial court hearing offers insights into the legal politics of interpreting the video images in court.^{Footnote 120} Instead of asking survivors or soldiers to testify in order to reconstruct the events unfolding throughout the night, the court relied on a review of the video footage taken from the F-15 plane as the main piece of evidence. The video, portions of which have become public in subsequent years, shows a rendering of the landscape through night vision technology that picks up sources of heat instead of depicting the shape of persons and objects on the ground. The sandbank and the shore appear in light grey; the river is dark. Crowds of people, recognizable only as small dark dots, are near two larger objects (the trucks). Some of the dark dots move, others stay in place. The center of the video is marked with crosshairs; we are watching from the perspective of the bomber. The crosshairs flicker, a large drawn elliptical shape appears on the video, and suddenly we see the impact of the bombs: a rapidly expanding dark cloud morphing into a mushroom cloud and a fireball that blocks the view of the people and the trucks on the ground.^{Footnote 121}

During the hearing, Kolanoski reports, “defendants and plaintiffs agreed on one central point: the people who were targeted had been visible only as ‘infrared dots.’”^{Footnote 122} The visual technology itself was operating at the limits of sight, representation, and resolution. The humans on the ground were at the “threshold of detectability” because their size and heat map “approximates the recording ability” of the chosen visual technologies.^{Footnote 123} As a result, they could only be identified as generic humans moving in the vicinity of the trucks, not as people with distinct physical attributes, carrying specific kinds of military or civilian objects, or appearing to be of a certain age or gender. As the trial court stated: “since the infrared images show sources of heat, individual persons would only appear as dots. As the examination of the video has demonstrated, it is impossible to see if one of the persons shown as a dot was carrying a weapon.”^{Footnote 124}

How would the courts read the dots? While the complainants “displayed a preference for seeing civilians in the dots,”^{Footnote 125} the defendants marshalled expert witnesses who insisted that the infrared dots could represent either civilians or combatants: “I cannot recognize military conduct in the classical sense, but also no typical civilian conduct.”^{Footnote 126} In the view of the trial court as well as the appellate court, the uncoordinated movements of the people near the sandbank did not necessarily indicate civilian presences because such patterns are also compatible with combatant behaviour: “Taliban fighters do not normally have formal military training,” which is why the lack of “coordinated conduct” resembling that of military forces is not surprising.^{Footnote 127} The indeterminacy of the images is used to argue that the dots could represent civilians or non-civilians. On the basis of the previously established assumption of hostility and a “racially inflected regime of visuality,” the courts superimpose “stereotypical images” of Afghans as likely militants on the video footage.^{Footnote 128}

In the legal reconstruction of the sights and decisions of that night, the courts allow Klein to not only suspect but assume that the persons near the trucks were combatants. In the legal proceedings, the epistemic authority of the video was limited: instead of creating doubt about the combatant status of the people on the ground, it was viewed as compatible with what was already imagined or “known”: that the persons near the trucks were Taliban. The paranoid imaginary of the local population as hostile “interpret[s] in advance”^{Footnote 129} the visual evidence of the video feed and erases any doubt that might give rise to additional legal obligations. Here, the bodies of Afghans appearing as black dots on a video screen are understood to be hostile even when such hostility cannot be confirmed or denied using the video alone. Superimposing the indeterminacy of the video upon the assumption of hostility, the Supreme Court attested that “the presence of civilians in the target area was objectively not recognizable” to the commander.^{Footnote 130} All legal decisions echo the Notice in its finding that “there is no evidence that Commander Klein expected civilian casualties when he ordered the strike.”^{Footnote 131} It is worth recalling that Article 50 of the Additional Protocol I to the Geneva Conventions requires that “in case of doubt whether a person is a civilian, that person shall be considered to be a civilian.” Yet the Attorney General established that Klein did not actually have any “doubt” about the status of the persons: he presumed that they were legitimate targets.

Once the courts established that Klein did not see the people on the ground as civilians, they had to adjudicate whether a “reasonable military commander” should have foreseen the presence of civilians in this case.^{Footnote 132} In the court proceedings, participants therefore produced the abstract “category of the military viewer.”^{Footnote 133} The law’s insistence on the figure of the “reasonable person” or, more specifically here, the “reasonable military commander” can mask engrained gendered and raced assumptions about who embodies reasonableness and which assessments are to be accepted as reasonable. Yet instead of using the legal technique of the “reasonable person” as a foil to assess Klein’s actions, the courts collapse the actual Commander Klein with the abstract “reasonable military commander.” As if inhabiting a caricature of law’s identification of rationality with the fears of white men feeling threatened by brown men, the decisions do not distinguish between what Klein saw and thought he knew and what a “reasonable military commander” should have seen and known. As a result, Klein’s assumptions, fears, and actions appear reasonable, and he walks away as the innocent legal embodiment of reasonableness.