The Testimony of Drones

Drones do more than see the world: they kill, terrorize, surveil, record, search, map, deliver and fertilize, to name just some of their more prominent uses. Drones can do all this because they are knowledge machines. Remotely piloted and equipped with sensors, drones transmit visual, geographic, diagnostic, and multispectral information back to a controller, where it can be analysed and acted upon. Drones perceive the world through a complex mix of sensors and receivers, monitoring their environment, responding and generating data. The knowledge work of drones has datafication at its centre: the transformation of sensory perception into computational information, structured and delimited by the technical capacities of hardware, software and network. But drone knowledges are also shaped by their operators, by the places and atmospheres they navigate, by the imaginaries of militaries, corporations, technologists and policymakers.

All this means that drones are not simply observers of events, but active participants in their making. They move the air around them, react to the force of wind, decode signals and encode new ones, detect and respond to objects and bodies, capture worldly phenomena and transform it into data. How they make knowledge matters because the drone can act on what it knows. That action takes place along a continuum that extends from minor adjustments of position to the application of lethal force. It might hover in place, zoom in closer, even kill.

Consider the MQ9-Reaper, the weaponised drone of choice for the US military, which can stay aloft at an altitude of 7.5 km for up to fourteen hours and deploy optical and infrared sensors to surveil people and places with the potential to unleash lethal force. Operated from half a world away at Creech Airforce Base outside Las Vegas, Nevada, the Reaper's data streams flow into the secure networks of the US military and out to command points across the globe. Working out of an air-conditioned Connex shipping container, its pilot and sensor operator respond to requests from commanders, ground troops, image analysts and even political leaders; the drone crew instructs the machine to move the air, sense its environment, track bodies and vehicles, zoom in on distinct elements, and launch its lethal Hellfire missiles. From the ground, the drone is a speck against the sky, an incessant buzz, a lethal shadow.

If the drone is the crucial nexus within this distributed knowledge machine, can it testify to what it knows and perceives?

In the simplest terms, a drone is an unmanned aerial vehicle or UAV. But more accurately, a drone is an unmanned aerial system, or UAS, because the vehicle alone is useless. At a minimum, it requires a controller, a network and a signal to operate. While the popular conception of the drone is aerial, the category can be broadened to include all remote vehicles that operate on the ground, underground and underwater. Soon, tiny nano drones might even operate inside the body to combat disease. To be more precise still, a drone is a remotely piloted remote sensor: as it flies it senses its environment and communicates the data it collects. It changes that environment too, displacing air, generating sound waves, sending signals.

But even this does not go far enough. Drones are not simply technical systems, but hybrid collections of human and nonhuman agents, data and processes, signals and atmospheres, environments and possibilities. ‘Drones are not idle machines hovering above;’ Lisa Parks and Caren Kaplan write in the introduction to their 2017 collection of scholarly essays Life in the Age of Drone Warfare, they ‘are loaded with certain assumptions and ideologies’. Drones are clusters of cultural, social and technical imaginings and fantasies: they coalesce fears, desires, attachments and fictions, as well as strategies, policies, technical possibilities, and experimental tendencies. It could be said that the drone is a socio-cultural figure, an icon or signifier that connotes a host of contradictory propositions, from prosthetic vision to encroaching surveillance to techno-salvation to autonomous slaughter.

If torture was the paradigmatic form of violence of George W. Bush's Global War on Terror, drones embody Barack Obama's rebranded Overseas Contingency Operations. Drones promise technocratic war, war without risk to blood or treasure in which the precise projection of power can preempt distant threats before they materialise. By combining the power of the aerial view with big data analysis, the militarised drone offers the neocolonial state war without politics. Persistent surveillance, algorithmic analysis, signals interception, military lawyers in the loop of the kill-chain, no images of bodies or body bags -- drone war promises to slide away from public view, resisting scrutiny and accountability. Demanding that the drone bear witness offers one means of dragging drone war into the light.

Commenting on the historical co-evolution of technologies of watching and war, the philosopher Paul Virilio wrote that ‘from the original watchtower through the anchored balloon to the reconnaissance aircraft and remote-sensing satellites, one and the same function has been indefinitely repeated, the eye's function being the function of a weapon’. While the sniper rifle makes distant killing proximate, it retains the intimacy of a human staring down the scope and pulling the trigger. By contrast, guided cruise missiles split the view between weapon and eye via remote control, leaving the human to coax the missile along its programmed flightpath. The drone places eye and weapon in a more equal relation: its utility for persistent surveillance amplifies its lethal capacity. Watching and warfighting collide, with the human eye geographically dislocated from their point of convergence.

Early drone development gave few hints of this lethal future. They were targets, not weapons. Drones take their name not from the droning noise associated with both their fixed wing and quadcopter varieties, but from the Queen Bee, a remote-controlled plane exhibited by the British in 1935 at the Second Naval Conference in London. Running with the insect analogy, US Navy observers adopted the term 'drone' for a target plane without a mind of its own on board. Even though this development project stuttered and stalled (as would a series of similar initiatives by the US military over the next fifty years), the name stuck and now describes everything from multimillion dollar militarised machines to aerial agricultural tools to small airborne selfie cameras.

Drone development in the US military moved forward in fits and starts, hampered by technological limitations, cultural reticence and bureaucratic infighting. While innovative engineering was crucial to the eventual adoption of the Predator and its kin, the more fundamental technological transformation was the rapid proliferation of networked computation in the US military and accompanying changes to strategy, tactics and culture. Set in motion by the end of the Cold War and accelerated by 9/11 and the war on terror, the Revolution in Military Affairs saw the US shift focus to integrated technologies and mobile fighting capacities. Computation, satellites, geographic information systems and networked infrastructure turned the world itself into a mediatised zone of potential war. The purest distillation of this is the ‘kill box,’ a term used by the US military to describe a zone in which lethal force is preauthorised. Kill boxes are three-dimensional volumes designated over a defined geographic area, up to a specific height and for a defined period of time. During the 1991 Gulf War, the first kill box was an area as big as New York City but today kill boxes can be tightly defined in space and time, mapped volumetrically into the air and below the ground. In his influential book Drone Theory (2014), the philosopher Gregoire Chamayou calls this a ‘temporary autonomous zone of slaughter,’ the violent manifestation of a new form of global war built around a logic of manhunting.

Despite claims that the aerial view would enable total situational awareness of the battlefield, the view through its optical camera is often described as ‘looking through a soda straw.’ At an optimal surveillance altitude of 5,000 feet, figures are flattened and difficult to distinguish. Non-optical sensors such as thermographic or infrared cameras require expert decoding. Major drone missions can involve over seventy analysts combing various data feeds, along with the pilot and sensor operator responsible for crewing the drone.

Aerial platforms dedicated to surveillance rather than killing present even greater processing challenges. Designed to be attached as a sensor payload to certain large drones, the Autonomous Real-Time Ground Ubiquitous Surveillance Imaging System, or ARGUS-IS, combines 368 overlapping lenses into the equivalent of a 1.8-billion-pixel camera to provide high resolution full motion video of up 10 square miles at a ground resolution of six inches from an altitude of 20,000 feet. Analysts can create video windows, track vehicles, generate 3D models and access location-specific archives. All that capability requires half a million gigabytes to be transmitted per minute, an impossible number even on the most advanced global networks. Maximising the power of the ARGUS-IS knowledge machine far exceeds the processing capacity of its operators, which is why the Defense Advanced Research Projects Agency (DARPA) now seeks partnerships with Silicon Valley to develop artificial intelligences capable of analysis across multiple spectrums.

In popular imagining, drone warfare is a surgical form of violence carried out against known and targeted individuals or groups. The 2016 film Eye in the Sky captures this succinctly: three terrorist leaders on a most wanted list are identified, tracked, targeted and killed from a loitering Reaper, with political interventions and legal considerations at every stage. In practice, 'personality strikes' against known, identified and verified targets are in the minority. 'Signature strikes' that identify potential targets based on 'pattern of life analysis' predominate US drone operations. Signature strikes target unknown persons, flagged as threats by an algorithmic system that uses signals interception to track movements, calls, texts, meetings, and other activities. Emergent within the data as potential threats based on correlation and calculation, the victims of signature strikes might never be identified. Because there is no feedback mechanism to ensure accuracy, because the threat is never actualised, their deaths become proof positive that the system works.

In war, the drone is both that which witnesses violence and what must be witnessed. It is witness and perpetrator. If the function of the eye is the function of the weapon, must testimony be sought from those eyewitnesses to drone violence found either on the ground, in the control room or accessing video feeds from elsewhere in the network? Or might the drone itself be called upon to testify? Might the knowledge machine bear witness to its own violence?

Witnessing is one of the most significant points of intervention into violent conflict because it asserts significance, meaning and responsibility. To bear witness to events is to become responsible for communicating what happened; to be placed under an injunction to act. One might refuse the burden or reject the imperatives of justice, but witnessing is nonetheless a fundamentally relational experience. It forges a bond. If the witness should choose to speak, testimony can take many forms. Pericles’ Funeral Oration is perhaps the earliest war testimony in European history. It summons the authority of what Pericles has seen or learned directly from others; it seeks to give account for the war to the dead. It tightens the fabric of the polity by creating a shared account of what happened and its meaning.

Fast forward to Crimea, the First World War and the brutal birth of industrial warfare, and witnessing is no longer the preserve of rulers and religious figures but of people forced into war. Testimony takes a different shape: photographs, letters, poetry. After the Second World War and the horror of the Holocaust, the trials at Nuremberg and Tokyo demand a new mode of narrative testimony. Under the aegis of international humanitarian law, the expansion of human rights leads to what the political theorist Michal Givoni calls an ‘era of becoming witness,’ in which the very foundations of political subjectivity became connected to the imperative to give testimony to political violence and injustice. Grappling with the impossible scale and sheer objectification of the Holocaust, novels and films become equally crucial testimonies.

At the same time, witnessing and media become enmeshed. In the Vietnam War, photojournalism brings the war home to America and incites widespread opposition. Operation Desert Storm in 1991 remade the witnessing of war for the cable television era and established practices that would be followed in Afghanistan and Iraq a decade later: embedded journalists, controlled briefings, packaged video. Witnessing in, by and through mass media reached a peak with 9/11, an act of violence made world-historical by its ubiquitous, real-time media presence. Smartphones, social networks and encryption have made media witnessing more democratic, enabling the viral circulation of image testimonies to potentially global audiences. Footage of Black and Brown death is far from an antidote to the enduring violence of white supremacy, but it can ignite protests for justice that force issues and events to greater political prominence. State efforts to control signals and networks speaks in no small part to the capacity of witnessing to spark political action against and in excess of the state. Protests and riots in the wake of the murder of George Floyd are the latest irruption of the body politic against the state political machine.

Just as smartphones give citizen witnessing an unprecedented potency, so too do drones mark a break in the witnessing of war. An eye in the sky that watches and kills is also an eye that witnesses, but the human operator is displaced. Unlike the scope of the rifle, the optical targeting system of the drone is highly technical. With a two-second lag as signals bounce between satellites, the drone displaces the human in time as well as space. Equipped with a sensor payload capable of perceiving across multiple spectrums and encoding that data into signals for distribution via a global network of control rooms, access points and secure facilities, the drone expands the number of potential witnesses exponentially. In a famous 2011 photograph, President Obama and his senior advisors huddle around a screen in the White House, witnessing the killing of Osama bin Laden through drone sensors and helmet cams. More often, witnesses through the feed of the drone are located in more prosaic surrounds: a shipping container repurposed as a ground control station; an open plan office crowded with computer screens.

In the countries in which drone warfare is now waged -- Afghanistan, Pakistan, Iraq, Yemen, Somalia, Syria, Palestine, to name those most affected by American, Russian, British and Israeli drone operations -- the witnessing of violence is visceral and immediate: ruined bodies, torn and scorched flesh, broken buildings. By sharp contrast, publics in the US and UK see very little, and in Australia, a step removed from killing in the global apparatus, we see even less. Even if the violence of drone warfare made its way into the mediascape more frequently, it would still remain at a distinct remove. The people killed by drones are Black, Brown and almost always Muslim: no drone strikes are launched against white or Christian populations.

Within militaries, pilots, sensor operators and image analysts are bound by codes of silence. The occasional whistleblower can reveal much, but the information is partial and can swiftly become outdated. On top of this, black boxed technologies such as thermographic cameras and automated image classifiers are in turn black boxed by military secrecy. This containment and enclosure makes the most intimate witness to drone violence utterly inaccessible: the drone itself.

I didn't expect to write about drones, but my PhD research on torture and the book that followed led me there. Drones loitered, a whir on the periphery of text and context. Soon enough I couldn't escape their insistent buzz, the way drones coalesce so much of contemporary life, technology and politics. Yet despite their pervasive presence, one of the problems in writing about drones is that they are elusive subjects.

Hunting for militarised drones, I spent three days in June 2019 at the Israel Defense and Homeland Security Convention, ISDEF. Compared to the enormous biannual events in Paris, Istanbul and Singapore, ISDEF is a modest affair but in the world of drones Israel is a leading player and its major manufacturers have been a presence at previous conventions. I had travelled to Tel Aviv hoping to come into contact with that rare beast: the large-scale militarised drone. But exhibitors complained quietly about the size and organization of the event. Israel Aerospace Industries, Aeronautics Defence Systems and the other big players had stayed away; only small drone and counter-drone technologies for security and policing were to be found. For the visitors to the convention, it seemed drones had become passé. Booths stacked with facial recognition tools and machine learning threat detection software attracted larger crowds; a sign of the times and of the normalisation of drones.

A few months later, I drove to an annex of the Smithsonian Air and Space Museum in Chantilly, Virginia, some 45 minutes outside Washington, DC. The first Predator to launch a missile in anger was meant to be on temporary display, relocated from the main museum on the Mall. When I found it, the drone was only half assembled, its wingless body inert on a trolley. I sat and stared at the sensor ball, suspended below the bulbous satellite transponder locate where the cockpit should be. This strange reflective eye. But prone like this, without vitality, the drone refused to give up its secrets.

All technology is like this to an extent: it has a material existence that cannot be reduced to the frameworks that we have for understanding it, even with scientific expertise at our disposal. The philosopher Gilbert Simondon teaches us that technology has its own mode of existence distinct from the human, its own completeness that belongs all to its own gathering into technical unity. Demanding that the drone speak might be an anthropomorphising step too far, but it is clear to me that we must consider the drone's material, discursive and affective relations to events as opening onto a kind of nonhuman testimony.

Less burdened by the refusal of blackboxes to open, contemporary art offers powerful testimonies to drone war and the uneasy, ambivalent atmosphere that gathers around drone themselves. Unlike the anti-war art of earlier forms of violence, art historian Thomas Stubblefield argues that condemnation in drone art is rarely explicit. This seeming passivity reflects a deliberate effort to dig into the particular contexts of drone violence in order to reimagine its relations. This is particularly evident in some of the earliest and best-known works in the genre. Each drone in Trevor Paglen's ‘Untitled [Drones]’ (2010) photographic series is a tiny fleck in an expansive, delicately hued sky, just on the very edge of invisibility. The to-scale crime-scene outlines of James Bridle's ‘Drone Shadows’ (2012-2014), painted in thick white on the streetscape of London, New York and other cities, ask that publics far from conflict zones identify with life lived under drones. The politics of such works is worn lightly. At work is something more subtle than fury: a finely tuned insistence on addressing the disquieting illegibility of drone war and calling for an identification with its victims.

Aesthetic responses from those more deeply impacted by drone warfare tend to be more urgent in their politics. Iraqi diasporic artist Wafaa Bilal's ‘...And Counting’ (2010) deals with precisely this illegibility of violence but makes his own body the site legibility. In a 24-hour live performance, the names of Iraqi cities and towns were tattooed on Bilal's back, followed by thousands of small dots in representing civilian deaths while names of the killed were recited in the gallery. Inked in white, the dots glow under ultraviolet, phasing in and out of photographic visibility but ever-present in the skin. While not exclusively focused on drone war, what Ronak Kapadia calls the ‘insurgent aesthetics’ of Bilal's work testify to the inescapably bodily violence of techno-war.

The photojournalism of Noor Behram is similarly insistent on the material and corporeal effects of drone violence, bearing witness to drone strikes throughout Waziristan on the Pakistani border with Afghanistan. Haunted faces of survivors, shattered bodies of victims, broken homes, and fragments of Hellfire missiles; the people and objects documented by Berham refuse to go unseen. Survivors stand in the ruins of their homes, cracked metal from the shaft of a Hellfire held in their hands like the weight of it might break them all over again. These images propose very different material relations between the technical apparatus and bodies on the ground than the precise and algorithmic one proffered by the apparatus itself.

Connecting this material wreckage of drone war to its computational and global positioning systems, the investigative work of the research agency Forensic Architecture possesses both an urgent politics and alluring aesthetic. Nominated for the UK's premier art award the Turner Prize in 2018, Forensic Architecture engages in both human rights advocacy and aesthetic production by using architectural techniques of research, analysis and presentation to investigate political and environmental crimes. Working at what founder Eyal Weizman calls ‘the threshold of detectability,’ the agency's projects seek to expose violence that phases in and out of visibility.

Among the agency's earliest projects, ‘Drone Strike in Miranshah’ (2014) identifies, maps and models the split-second violence of a drone strike on a home in Northern Waziristan in 2012. ‘Drone Strike in Miranshah’ uses mobile phone footage in the aftermath taken in the strike's aftermath and smuggled out of Waziristan to build a computer model of the building which can then be identified using satellite imagery and Google Maps. By carefully translating shrapnel marks from the walls in the footage to the computer model, the exact position of the mid-air explosion of a Hellfire missile could be estimated with a high degree of probability. Sections of wall without shrapnel suggest the location of bodies.

Rendered and animated, this architectural investigation becomes both an act of testimony and an aesthetic object. Installed and presented in more than a dozen exhibitions around the world, ‘Drone Strike in Miranshah’ is both art and advocacy. Unlike drone art that aims to establish some kind of identification between the viewer and drone violence, this and other Forensic Architecture investigations are about what counts as evidence, how testimony to unseen violence can take place, and how aesthetics can enable new forms of accountability. This testimony of the drone is reluctant, a product of sustained forensic investigation on the part of Weizman, his team and collaborators.

Such a mode of investigation is not a sustainable practice for the thousands of drone strikes launched by the United States, the UK, Israel, Russia, Turkey and others each year. But it alerts us to the necessity of witnessing that takes the apparatus itself as the site of investigation, that insists that the drone bear witness to its own violence. In ‘Drone Strike in Miranshah’ the aircraft itself is never visible, but rather reveals itself in the reverse engineering of its violence.

Researching and writing on drone war, it is easy to fall into a totalising trap that figures such technologies as an all-encompassing planetary enclosure. While the desire exists to achieve what the US military calls full spectrum dominance, the reality of drone warfare is that it is contingent and error-prone. Take the ARGUS-IS surveillance platform -- the system theoretically capable of offering six-inch resolution over 10 square miles. With its high error rate, astonishing bandwidth and storage demands and the challenges of real-time analysis, the ARGUS-IS is more notional than practical.

But it is not only the ARGUS-IS that plagued by latency, error and data volume. Even the established Reaper and Predator programs can be tripped up by poor weather, by satellite and signal failure, or by simple mistakes. Networked data streams mean that commanders and even political leaders can be looped into active operations. Information flows become tangled, chains of command snarled. Errors of judgment compound up and down the system, amplified by the desire of operators for the kinetic release of lethal force. Drones crash and become totemic trophies. Costs overrun. None of this is new, either to the military industry complex in general or to drones in particular. The history of unmanned aerial vehicles in the US military is one of failure, of programs quietly shelved, undercooked technology and blunted ambitions.

Even now, US drone operations are rife with failures of information, technology, process and judgment that lead to unjustifiable death. While some US drone strikes are justifiable under international law and meet rules around the proportionality and risk of civilian harm, others are not. Many involve flawed data, mistaken analysis, callous disregard for life and racializing and gendering interpretations of what appears on screen. Weddings, tribal councils, convoys of women and children, family homes: the reprehensible bombing of these and other sites breeds fierce resentment. The witnessing of drone war often arises out of these terrible failures. Bodies leak into the media, survivors speak before inquiries, whistleblowers find their voice. But the drone itself, most intimate of witnesses, is rarely called to testify.

Within the emergent field of drone studies, questions of how drones make knowledge and what that knowledge makes possible remain urgent. Grasping the technics of the drone -- not just how the technology works, but the material, discursive, institutional, affective, environmental, atmospheric and processual milieu -- is vital. If drones are knowledge machines, how might that knowledge be interrogated, held to account and, when necessary, dismantled altogether? Asking how drones bear witness and what kinds of witnessing they demand is one line of flight into the drone apparatus.

Testimonies of the drone would begin with the multispectral record of the sensors it deploys, the full-motion video and thermographic imagery. But rather than demote such footage to the status of evidence, the algorithmic and networked architectures of the drone system could be made to account for themselves: their inputs, processes and outputs made public, and their status as active agents in a system of killing made contestable in courts of law, public opinion and politics. Those things that remain unknowable and inaccessible could be presented and recognized as unknowable and inaccessible, rather than hidden from view. The architectures of command and control, the military rules and information practices, the network components and signals: drone testimonies would require that these things be called to account and placed under the injunction of responsibility for killing, terror and surveillance. Doing so would mean expanding the current focus from the drone and pilot to the system itself, on the one hand, and the technical workings of its components on the other.

Assembling drone testimonies means reconceiving discontinuities, ruptures, redactions and absences as traces and markers of the system itself. Drone testimonies will always be partial, incomplete. Composed of proliferating elements that might include aesthetic interventions, survivor and whistleblower testimonies, news media packages, investigative reports, ruined flesh, missile fragments, snippets of imagery, and more besides, the difficulty of producing drone testimonies speaks to the urgency of the task.