Drone trauma: violent mediation and remote warfare

A media theory of drone war

In simple terms, drones consist of a sensor-equipped vehicle that operates under its own power but is (for now) controlled remotely, although with significant assistance from on-board software. As a media technology, the drone is often understood as a ‘vision machine’ (Stahl, 2018: 67) that operates within a distinct ‘scopic regime’ (Grayson and Mawdsley, 2019; Gregory, 2011), but the sensors of military drones extend well beyond optical vision and their capacity to process sensor data prior to any human analysis, alongside their embedding in larger socio-technical infrastructures, indicates the need to understand them as operating in other sensory modalities. Drones are better conceived as media apparatuses, techno-political assemblages ‘that gather and produce subjects, objects, discourses, politics, terrains and especially, atmospheres or airspaces’ (Kaplan, 2020: 2). As such, ‘drones do not simply float above – they rewrite and re-form life on earth in a most material way’ (Parks, 2018: 146). But for all their technical complexity, drones are still animated by the performative entanglement of human, media and machine (Chandler, 2020; Fish and Richardson, 2021). To say that drones are media apparatuses points not only to their capacity to represent and communicate data, but also to the constitutive role within drone systems that of mediation, defined as a vital process of informational and energetic transformation between materials, systems and subjects (Cubitt, 2017).

Militarised drones are martial media, with their structures and logics direct towards lethal surveillance (Hall Kindervater, 2016). While the capacity to launch lethal strikes is the most direct, kinetic violence of the drone system, its principal role is the identification and production of enemies through its surveillant capacities. This production of enemies (Packer and Reeves, 2020) is embedded within the logic of pre-emption that shapes both contemporary warfare and the automated software systems that enable its operation (Andrejevic, 2019). Within the drone system, the pre-emptive identification of foes requires the pre-designation of potential targets within a necropolitical logic that renders entire populations killable (Mbembe, 2019). Despite technocratic rhetorics of precision, proportionality and objectivity, drone violence is always political and questions of responsibility and the path to justice should never be far from view. Drone systems such as the US military’s Reaper or Israel’s Hermes and Heron drones flown over Gaza achieve this through violent politics at the level of mediation itself, enacted through sensors, networks, algorithms, interfaces and finally, in the guidance systems and explosive force of guided missiles. For all their technical complexity, drones don’t mark a radical break with prior forms of martial violence but are preceded by other forms of aerial vision and violence (Kaplan, 2018), embedded in colonial histories in the places they operate (Ashraf and Shamas, 2020) and operated in conjunction with other practices of martial and state violence, often with the complicity of the states above which they are deployed (Tahir, 2021).

If drone systems can be understood as media assemblages that produce violence within and through their processes of mediation, then the question is how their medial dynamics (sensing, storing, transmitting, analysing, interfacing and so on) shape the traumas produced through those systems. Extending the ambit of media trauma beyond representation, signification and performance, Pinchevski (2019) argues that we can ‘understand the traumatic as something that is made manifest through media technological rendering’ (p. 2). Media are not simply containers for trauma, but shape and are shaped by its workings: trauma is not ahistorical but bound up with the media within which it is manifested and materialised. I want to push this logic a step further, to argue that in techno-scientific warfare we encounter trauma within technical systems, such that the trauma they produce is shaped by the forms, logics and processes that animate martial media assemblages such as the drone. Drone traumas are thus bound up with what I call violent mediation: material processes that cut, target, exclude, define, categorise or classify in injurious or harmful ways. It is this violent mediation that enables the distention of trauma through time, ensuring that it has already occurred within the drone system well before it is felt on the ground.

In this essay, I follow processes of violent mediation through six distinct media materialities within the drone apparatus: sensors, networks, algorithms, interfaces, atmospheres and missiles. While each of these interventions is necessarily brief, tracing mediation throughout the system from sensing to striking is vital to make sense of how drone trauma both reproduces familiar forms of media trauma but also intensifies and reworks them through its own distinct media logics. These logics distribute and disperse causality, recast individuals and communities as datalogical dividuals and embed a radical, machinic contingency into daily life. This essay thus aims to extend the themes of this Crosscurrents section to ask how the media technologies that increasingly saturate war are reshaping the trauma it brings to those caught up in its violence.

Sensors

Before it is a weapon, the drone is a sensor. Sensor capacities and capabilities vary but typically military drones are equipped with radar, infrared and optical imaging. Most carry Synthetic Aperture Radar (SAR), which can be used to produce two-dimensional images and three-dimensional models of environments. Optical cameras function like their televisual cousins, making them of limited use in darkness or when environments are built up, thick with foliage or marked by ravines and other features that provide cover. That’s when thermographic sensors become essential, as they detect heat using the infrared spectrum, visualising warm bodies and engines in warm reds, yellows and whites, and cold metal and water in blues and greens. For drones like the Predator or Reaper which can be equipped with lethal missiles and provide direct support for other air or ground operations, the sensor package also includes laser range-finding and targetting for on-board missiles and target-painting systems to direct the fire of other forces.

For those targetted, the mediation of trauma comes before its experience. In war, drone aesthetics – sensing and sense-making – are inseparable from the potential for violence, and from the necropolitical order that designates populations able to be killed (Mbembe, 2019). SAR reproduces environments to make them legible and actionable; optical imaging generates the illusion of omniscience; thermographic sensing translates the basic energetic processes of vitality into informational domains through which they can be targetted. These sensor assemblages are always in the act of pre-emption. If the necropolitics of strategic doctrines of pre-emption and man hunting depend on the designation of death worlds, the machine vision imaging capacities of drones make those worlds operable and actionable: they produce a kind of virtual trauma, already present in the potential of a lethal strike, pre-emptively haunting the populations that must live underneath them. Even without access to those technics, the radical disruption of daily life in Waziristan, Pakistan or in Gaza, Palestine, by the pedagogies of aerial violence produce an intensive relation between the sensory transformation of life into not-yet-death and the living of life on the ground.

No matter the modality of sensor, its output is dehumanising: an array of pixels, a second-order representation of the transference of light and electro-magnetic forces from sensor to visual image. Typical sensors provide a ‘soda straw’ view, while Wide Area Motion Imagery (WAMI) systems like the ARGUS-IS have (as far as we know) failed to work in operation (Michel, 2019). These sensing media thus reproduce a blindness to widespread violence and trauma, such as in the Federally Administered Tribal Areas of Pakistan, where the state has engaged in campaigns of disappearance, torture, reprisal and suppression outside the gaze of much of both Western and Pakistani media (Tahir, 2021). At every level, then, from technical process to field of vision to where the gaze of media turns, the sensors of militarised drones produced and are themselves produced by conditions of actual and virtual trauma in those subject to its violence. Drone traumas are thus already latent within the sensors themselves.

Networks

Drones are necessarily networked entities. Signal animates the network, travelling between a drone in the skies above Syria, say, to an orbital satellite, to Ramstein airbase in Germany, to optical fibres that run under the Atlantic and then across the United States, with deviations and digressions along the way to split off duplicates to command centres and other sites, to the ground control stations where operators and analysts interface directly with the system. Drone mediations are thus always distributed and dispersed, as well as vertical and violent. In fact, central to the violent mediations of drone systems and the traumas they engender is this networked architecture. Signal distorts and transforms as it moves through media, but more importantly the distances and transductions involved produce significant latency, as much as 2–6 seconds depending on estimates. Sensor operators view images that are not just on the cusp of live-ness but that have already passed into pastness. Here, then, is a strange medial relation to trauma, a rupturing of experience that is always latent, always arriving after the event even as it pre-empts, produces and records traumatic violence.

In drone warfare, information always lags. Pilots respond to navigation and climatic data that is already past, sending signals to turn or rise that arrive later still. Decisions to kill are based on information that will have been outdated by the time the missile launches. As Pinchevski (2019) observes, the belatedness of trauma is bound up with the belatedness of recorded media (p. 17). In the fictional film Eye in the Sky (2014), these distended moments are captured in the return to the kill zone of a young girl who has been selling bread in the street outside a house at which a Hellfire has been launched. The networked architectures of the system entrench this latency, binding the very workings of the system to a rupturing of time, a discorrelation between world and image. No wonder, then, that the traumas of drone warfare circulate in just such a disconnect: that between the sky above and the potential for death to have already arrived. As a 13-year-old Pakistani boy named Zubair told the US Congress: ‘I no longer love blue skies. In fact, I now prefer grey skies. The drones do not fly when the skies are grey’ (Abad-Santos, 2013). As lethal media, drones fuse sensing and killing – but they do so only through a fracturing of time, by embedding that disjuncture into the workings of the system and carrying that through to the experience of violence on the ground.

Algorithms

Decision-making within militarised drone systems is frequently dependent upon or subordinated to algorithmic systems. Some large drones now carry on-board image analysis, typically using machine learning to pre-process sensor data and transmit only that which is most like to be actionable to human operators and analysts. Under Barack Obama, US drone operations were dependent upon the GILGAMESH data collection system, designed to intercept cell phone network data and channel it into the SKYNET system for analysis. Such algorithms are vital agents in producing the ‘signature’ of threat through the analysis of ‘patterns of life’ determined from cell phone metadata. The targets produced by these algorithmic processes are not individuals but dividuals, aggregations of data that obtain meaning and coherence only within the data system itself but are otherwise stripped of context and history (Amoore, 2013). Distributed through the system by computational processes that are blackboxed from operators and entirely inaccessible for those subject to their determinations, decisions to undertake lethal strikes are filtered through the technoscientific apparatus of sensor-network-algorithm and so rendered inaccessible.

This inaccessibility shares much with the familiar unrepresentability of trauma, its constitutive refusal to be known or accounted for. If the printed word and the moving image share a certain capacity to capture the dynamics of trauma’s desire and refusal to be known, computational systems such as SKYNET or on-board artificial intelligences packages like Agile Condor introduce a medial agency of a different order. SKYNET uses random forest algorithms, a type of machine learning system that generates a host of outcome ‘trees’, a process which ‘takes place as a calculation amid incalculability, mobilising chance and the splitting of agency, sometimes with lethal effects on human life’ (Amoore, 2020: 127). In doing so, drone war’s algorithmic systems enfold the production of trauma through the decision to kill into a zone of indistinction and inscrutability, in which chance, error, probability and potential collide in ways that cannot be legible and knowable to humans. For the operators of drone systems, there might well be a certain release in this – a capacity to defer responsibility to the system. This certainly appears as a critical component in the technocratic rhetoric deployed by Obama, which sought to keep drone war at arm’s length from politics. But for those on the ground who are wounded and have their loved ones killed, or live with the possibility of that occurrence, the insertion of an algorithmic layer of computation finds its lived resonance in the desperate efforts to shift the forms, rituals and actions of daily life to avoid death from above. Drone trauma thus entails the incorporation of becoming-dividual into the rhythms of life.

Interfaces

Interfaces are the necessary connective elements that allow signal, for instance, to pass from drone to satellite to receiver. Military drone histories reveal the ad-hoc nature of network interoperability, the problems of signal and connector types, of format and transcoding (Arkin, 2015). All these interfaces and the necessary translations and conversions that take place through them point to the aggregated nature of drone media, and to the futility of attempting to think of drones as coherent media objects and the necessity of understanding them as unstable assemblages. Drone operators pilot the uninhabited aerial vehicle and direct its sensors through more overt and familiar interfaces: controllers modelled after those of gaming consoles; keyboards; and screens that display information (sensory image, flight data) and provide interactive surfaces for navigation, surveillance and targetting.

Unlike the recording media that co-constituted the traumas of the 20th century, mimesis and signification – those bedrocks of representation – are only secondary to the images displayed on the drone control screens. These images are what Farocki (2004) called ‘operative’, or images in which their representational content is superseded by the function they serve within the operation of the computational system. While not strictly images produced for machines, as Farocki formulated the term, drone control screens produce an interfacial relation that binds operators to apparatus – and, crucially, to those at the other end of the sensor array. On weaponised drones, the interface layers a targetting reticule over data rendered into pixels legible to human eyes. Range, flight time, locking options and other such information accompany its movement. These targetting capacities, yoked to the aerial flight of the drone, invert the dynamics of war: violence can be visited wherever the target locks based on the designation of a dividual as threat, rather than through the entry or actions of a person or collective within a geographic space designated as a zone of war. If recording media detach experience from time through the capacity to recall at will, then the computational interfaces of drone warfare disaggregate experience from place and introduce a radical contingency. For those bodies transcoded into infrared heat and tracked by the targetting reticule, trauma is already in production within the system. It is an operative outcome, a computational consequence of the targetting interface itself.

Atmospheres

Drones are atmospheric technologies, designed to loiter thousands of metres above the surface of the earth. Air is the medium of drone signals, whether of the optical or electromagnetic variety. But air is also the medium for drone sound, a whirr that cuts through the hum of daily life and grinds against the mind: an aural wound. In a report conducted by Stanford law clinic, one man describes the sound of the drones as ‘a wave of terror’ that sweeps through the community. Another links their buzz both to the permanent affective state of fear and to the strain placed on communal gatherings. ‘When we’re sitting together to have a meeting, we’re scared there might be a strike’, he tells the researchers. ‘When you can hear the drone circling in the sky, you think it might strike you. We’re always scared. We always have this fear in our head’ (Stanford Law School and NYU School of Law, 2012: 81). As Mohammad Kausar, father of three, says, ‘drones are always on my mind. It makes it difficult to sleep. They are like a mosquito. Even when you don’t see them, you can hear them, you know they are there’ (p. 83). If the everyday disruptions and anxieties of life under drones is most present in their aural intrusion, then can we really exclude the sonic landscape from the wounding of drones?

Drone mediations that travel through air are not immaterial but felt bodily in the vibrations transduced from motor to ear and skin. To live under drones, then, is to live within range of their atmospheric mediations. To say that drones produce an atmosphere of fear is a material claim: affective intensities course through the air as it is shaped by the drone, materialising in bodies as vibrations. Following Anderson (2009), we can ask how such atmospheres ‘envelope’ and ‘“press” upon life’ but also how we might ‘attend to the collective affects “in which we live?”’ (p. 77). The affects of drone war’s atmospheres are themselves traumatic, rupturing intensities that keep the body off kilter, on the verge of trauma (Atkinson and Richardson, 2013). Atmospheres actualise the contingency of drone traumas, that paradoxical relation between not-yet-arrived and already-too-late. Just as drone mediations cannot be confined to the movement of information within technical networks, so too the mediations of drone trauma extend beyond the technics and computation. Drones produce traumatic atmospheres through violent mediations of air itself.

Missiles

Originally designed as an air-to-ground armoured tank buster, the AGM-114 Hellfire was repurposed for drones in the aftermath of 9/11: it was cheap, widely available and just small enough that four could be strapped to the wings of a Predator. But the problem was that it didn’t do enough damage, its explosive charge was packed into the nose and built to tear through armoured plating. The solution was quick, dirty and brutally effective: engineers wrapped the Hellfire in a blast fragmentation sleeve that would come apart in the heat of the explosion, hurling molten shrapnel through bodies and buildings with a kill radius of 15 m and a wounding radius of 50 m. For all its supposed precision, the armed drone would become an indiscriminate killer within a define zone of lethality. Unleashed on weddings, funerals, councils and convoys of women and children it rips through bodies. Its wounds can maim. Fragments of the casings, often with readily identifiable markings, are presented as evidence of the occurrence of a strike: they are its remainder, a material instantiation of the absent perpetrator. Once the missile is in the air, trauma can rarely be escaped.

Missiles such as the Hellfire are the kinetic delivery devices of lethal media networks. Equipped with a sensor that allows it to remain fixed on a laser-tagged target, the launched Hellfire transforms the datalogical and potential violence of the drone apparatus into kinetic brutality. Trauma unfolds in its aftermath, as trauma entangles many if not all survivors of martial violence. In photographs of drone strike aftermaths, such as those taken by the Pakistani photojournalist Noor Behram, Hellfire fragments are often held in the hands of survivors, testifying to the event that has taken place. They stand in for the trauma to come, themselves fragments of a whole undone in the explosion of their arrival. The Hellfire – strapped to the wings of a Reaper, in flight, in fragments – materialises the starkest distinction of drone trauma from other forms of mediated trauma: this is a media designed to produce trauma, as well as one shaping and shaped by trauma. In the blast of the Hellfire, the datalogical dividual is returned to flesh, now torn and broken and those that survive are marked by having come into contact with the datafied system, with its networked mediations, through the explosive force of its most violent mediation.

A trauma theory of the drone

The trauma produced by militarised drones shares much that is familiar: belated arrival, rupturing of experience, resistance to narrative and representation and debilitating capacity to upend the present with the force of an event that refuses to become part of the past. But the trajectory I have charted here from sensor to missile also reveals that drone traumas possess distinct characteristics that are intimate to the specific technics, forms and processes of violent mediation that animate this decidedly martial media assemblage. Drone traumas begin before and operate independent of any kinetic, explosive event: like the media apparatus itself, drone traumas are pre-emptive, already in the making prior to their arrival in the subject. This pre-emptive quality is different from the trauma that lurks in the future of the survivor of gun violence because it is already in mediated process, already percolating within technical systems that possess a nonhuman agency. Sensors, networks, algorithms and interfaces distribute data and decisions, dispersing responsibility such that the traumatic object – the drone – is doubly withdrawn, not just hovering at the limits of perceptibility but too large, distributed and complex to meaningfully trace and make sense of. Within drone systems augmented by algorithmic threat detection, such as those of the US and Israel, targets are always already fractured from their lived existence. Produced as dividuals by machinic data analysis, the targets of drone violence reckon with this in the radical contingency that plagues life under drones: trauma’s ever-present potential within the media apparatus of the drone.

As with so many aspects of drone warfare, there are ways in which all these dimensions of associated trauma have antecedents in earlier forms of martial violence. What was called shellshock in WWI, for instance, was a cataclysm from the air, and there is little practical difference if Hellfire missiles are launched from unmanned Reapers or crewed Apache gunships. But like drone warfare writ large, tracing these emergent properties matters because contemporary warfare is undergoing significant transformation. Increasingly autonomously weapons systems also mean more media-saturated warfare, with sensors, algorithms, networks and interfaces shaping more of the material media (atmospheres, terrains, buildings, oceans) through which bullets, missiles and other munitions travel. Swarming drones, for example, widen the gap between targetted and target, potentially eradicating notions of responsibility for the other. What will it mean for trauma if the media that produce it and shape how it is transmitted don’t ‘record’ in any ways that are at all analogous to the printed page, the cinematic camera or even the smartphone? There are crucial questions here too for the future of media and trauma beyond war. Machine learning and other algorithmic systems are transposable between contexts. Techniques developed in war are already applied in policing, expertise flows from Google and Microsoft into defense. So much media is already computational, but what that machinic malleability means for our own subjectivity is still an open question. If the drone traumas already manifest in those who live with their violence, then reckoning with how the unstable, evolving media assemblage of the drone shapes and is shaped by that trauma will remain an ongoing, always incomplete task. But the task of reckoning with that trauma must begin with an insistence on the politics of drone warfare and a refusal of the disavowals by states, militaries and politicians that diminish or elide its traumas.