Cops, cameras, and the policing of ethics

In this article, we explore how cameras are used in policing in the United States. We outline the trajectory of key new media technologies, arguing that cameras and social media together generate the ambient surveillance through which graphic violence is now routinely captured and circulated. Drawing on the work of Michel Foucault, we identify and examine intersections between video footage and police subjectivity in case studies of recruit training at the Washington state Basic Law Enforcement Academy and the Seattle Police Department’s body-worn camera project. We analyze these cases in relation to the major arguments for and against initiatives to increase police use of cameras, outlining what we see as techno-optimistic and techno-pessimistic positions. Drawing on the pragmatism of John Dewey, we argue for a third position that calls for field-based inquiry into the specific co-production of socio-techno subjectivities.


Introduction
"What are you doing? Get off of me!" yells a girl. An officer is struggling to secure her hands. Another teenager thrusts herself between the scuffling pair, sweeping her friend to techno-optimistic and techno-pessimistic positions. Drawing on the pragmatism of John Dewey, we argue for a third position that calls for field-based inquiry into the specific coproduction of socio-techno subjectivities.

Cycles of crisis
Beyond the city limits of Seattle, videos circulated of other killings, increasing in density on a timeline of public attention until there were no spaces between the individual tragedies. In retrospect, the shooting of Oscar Grant by a California Bay Area transit officer in 2009, Eric Garner's last breath on a sidewalk after being put in a chokehold by a New York police officer in 2014, and the death of Michael Brown, killed by a Ferguson, Missouri police officer a few months later, are brackets on a phase of change. By 2015, video deaths came too thick and heavy for each alone to capture the same degree of mainstream attention, although sites for acknowledging and tallying them had proliferated. One significant maker and marker of this shift was the Black Lives Matter (BLM) movement, founded around the midpoint of the period by three black women in reaction to the killing of 17-year-old Trayvon Martin by a neighborhood watch volunteer in Florida in 2012, and his acquittal in 2013. #BlackLivesMatter intended, in the words of one of its founders, Alicia Garza (2014), to be "an ideological and political intervention in a world where Black lives are systematically and intentionally targeted for demise", and rapidly developed into the decentralized, nonhierarchical chapters of the BLM National Network. Counting and naming the women, men, and children who died, disproportionately of color, they signified the loss of individuals through their relation to shared systemic violence and oppression.
Nothing indicates that fatal encounters between the police and members of the public actually increased in the years leading up the deluge of video deaths. Because reporting such encounters to the federal government is not mandatory for law enforcement, and public health statistics are flawed (Klinger, 2012), there is perhaps no way to be completely sure, but fatal attacks on officers, which are kept by the National Law Enforcement Officers Memorial Fund (2014), showed no upswing. If anything, despite highs and lows, the decade average of 163 officer deaths per year, between 2004 and 2014, gradually declined. Official Department of Justice statistics on justifiable police homicides (committed by officers)-collected in an FBI database for self-reports by law enforcement-suggested about 400 deaths per year, but from only around 750 agencies (out of nearly 18,000). The yearly variations were useless for assessing rising or falling rates, since they could simply be more or fewer agencies submitting their numbers. The same problem arose with data on "death by legal intervention" in the National Vital Statistics System, in which underreporting as high as 51 percent made it impossible to assess if changes were in mortality or in reporting (Klinger, 2012: 80). In another marker of this period, however, several nongovernmental efforts to track police killings were started. Relying to varying degrees on newspaper articles and government records, paid researchers and crowdsourcing, the sites "Fatal Encounters", "Killed by the Police", and the "Gun Violence Archive" all began in 2013, and another, Mapping Police Violence, in 2014. These nongovernmental sites documented, retrospectively and in actuality, closer to 1000-1200 police homicides per year.
Thus the number of fatal encounters seemed relatively steady, although ultimately unknown. What had plainly escalated, however, was the quantity of violent images and their circulation, both online through the new activist groups, and in the mainstream media. The sources were usually mobile recording devices-smartphones with video capability and patrol car dash-cams-often substantiated later by footage from security cameras. What seems apparent in retrospect is that the presence of these technologies had passed a certain threshold of saturation in public spaces. And among the many ramifications of the diffuse, ambient surveillance they produce was a new round in the recursive cycle between mediatized crises and police reform initiatives.
Episodes of massive civil unrest marked the United States in the 1960s. 2 A national commission established by President Johnson to investigate the causes of these "race riots" reported in 1968 that "police practices" were both a trigger and the most deeply held underlying grievance (Kerner, 1968). Despite this, the commission noted, "[i]n several cities, the principal official response has been to train and equip the police with more sophisticated weapons" (Kerner, 1968: 8). A month after the report was released, Dr Martin Luther King, Jr was assassinated and the "Holy Week Uprising" exploded in cities around the country. At the end of that summer, the "nationally televised 1968 Democratic National Convention in Chicago and the riots that ensued perhaps for the first time portrayed the police as institutionally unaccountable" to the nation as a whole (Greene, 2000: 307). In response, over the next couple of decades a series of reforms and different kinds of training were tested, including carrying over Police Community Relations programs from the 1940s and 1950s, "team policing" (Greene, 2000: 308), and problem-oriented policing (Goldstein, 1979).
In 1991, crisis and reform began again when an unarmed black motorist named Rodney King was beaten by Los Angeles Police Department officers. The late-night scene was captured by neighborhood resident George Holliday on his new camcorder, and after trying to give the footage to the police department, which was uninterested, he contacted local media. With the "CNN effect" of nonstop coverage, what was then "the most explicit and shocking news footage of police brutality ever to be seen on television" aired around the world (Skolnick and Fyfe, 1993: 3). When the officers were acquitted in 1992, the city erupted for six days: 53 people were killed, over 2000 injured, and more than 11,000 were arrested. This too received nonstop coverage. Around the country, other protests occurred on a smaller, but sometimes longer, scale, and as before, the United States turned its attention to policing models and technology. In 1994, President Clinton offered sweeping incentives and nearly $9 billion dollars of the Violent Crime Control and Law Enforcement Act's funds for police to adopt "community policing" in their agencies, hire officers, and acquire new technologies. Within five years, 90 percent of police forces in the United States claimed to be doing community policing, and threequarters of local officers used in-field computers (Hickman and Reaves, 2003). In 2000, following police racial profiling scandals, the Office of Community Oriented Policing Services, established through a provision in the 1994 Act, began an incentive program for patrol car cameras (Westphal, 2004).
In 2014, heavily armed and armored police deployed against the crowds in Ferguson, MO mourning the death of Michael Brown, and as in decades before, the clashes were reported nationally and internationally. This time, the story also spread online, and became part of the larger narrative about black lives and police violence. The only images of the fatal encounter, however, were of Brown's body left lying in the street. Witness accounts of how the shooting had occurred-if his hands had been up in surrender or not-contradicted each other and the officer's statement. In the context of contentious debates over what had happened, the idea of equipping officers with small, digital bodyworn cameras (BWCs) emerged as a consensus technological solution for activists and policymakers.
In December of 2014, President Obama proposed $263 million for police training and body cameras, through a partnership program in which states would match what they received in federal funding. The actual congressional appropriations budget for FY2016 whittled this down to $25 million for the cameras, but states were already working on their own funding, and the number of pilot programs skyrocketed at agencies around the country (Nirenberg et al., 2015). Larger law enforcement agencies could and did move faster than smaller ones, so that in 2016 over 95 percent of major cities and countries (70 agencies) reported that they were using body cameras or had plans to (Maciag, 2016). Concerns were raised by nongovernmental organizations and activist groups over potential invasions of privacy, manipulation of the devices by wayward officers, and the fact that surveillance, particularly of vulnerable populations, would increase. Yet, proponents were optimistic. They hoped that videos would resolve discrepancies between witnesses, give lie to false claims, and protect officers from spurious accusations.

Ambient surveillance and mediated subjectivity
As these cycles of crisis illustrate, violent incidents between members of the public and the police are not new. What changed in the first decade of the 21st century is that cameras and social media became prevalent enough among cell phone-wielding citizens and in official use by police departments, other parts of government, and businesses, to generate ambient surveillance, and thereby a greatly intensified "mediated visibility" (Thompson, 2005: 31). A "loosely coordinated camera infrastructure", as Sandhu and Haggerty (2015: 1) put it, turned policing into a high visibility occupation.
To put the situation in historical context, public video surveillance in the USA began in the 1970s to counter crime, and was used by state and local agencies before the events of 11 September 2001, at which point it gained prominence as an anti-terrorism measure (Yesil, 2006). Into this surveillance milieu, relatively well controlled by police, the first consumer cell phones equipped with cameras arrived in 2002 (Hill, 2013). Joined with increasingly available and functional wireless capacity, sending phone photos (and, slightly later, videos) rapidly went mainstream. Within five years of the introduction of the iPhone, in 2007, nearly 60 percent of 25-34-year-olds, and 35 percent of all adults, across household income ranges, had camera-equipped smartphones (Smith, 2011). Likewise, and indeed, relatedly, from 2005 to 2011 the percentage of youth and adults using social media rose dramatically from 8 to 65 percent of the US population (Madden and Zickuhr, 2011). The capacity of these media for interactivity (Jensen, 1998), particularly commenting on online posts, articles, photos, and videos, played a significant role in creating an extended sociality within the affective domains of various social media, from subreddits to the Twittersphere to Facebook.
In sum, the 2000s were a critical time of transition in the USA. The indices of use then continued to rise, but already in a hand's span of years, much of the population acquired a portable telephone that could be used to take pictures and record encounters; and also began to have an online social existence, through which digital archives were shared. Ambient surveillance was a side-effect, a perhaps predictable one, of the convergence in these two historical trajectories of computing and media technologies (Manovich, 2001: 44).
A significant ramification was an increase in what Mann et al. (2002) named "sousveillance". Rather than the overview position of surveillance, this was watching from "below" (sous, in the French), including people recording and posting their encounters with police. As these audiovisual archives of public interactions were poured into the trenches of new media, they entered multidirectional circulations. Some documented incidents of police violence gained enough momentum and attention to receive coverage by traditional media (still caught up in the 24-hour news cycle that had been so important in the 1980s), to greatly magnifying effect.
Police departments too began to feel that they needed to use these new media, for public relations, or community outreach, or because their criminal investigations took them to cyberspace. Indeed, the International Association of Chiefs of Police began conducting a survey of US law enforcement's use of social media in 2009. In the 2015 survey, 96.4 percent of responding agencies said they used social media, although not quite half of those agencies reported "goals and/or outcomes" (IACP, 2015: 18).
Beyond instilling in the police a semi-articulated sense that they needed to adapt to these changes in US life, sousveillance, and ambient surveillance more generally, made officers the subjects of what Goldsmith (2010), following Thompson, calls a "new visibility". Thompson frames this visibility in contrast to the characteristics of the much earlier surveillance analyzed by Foucault of Jeremy Bentham's Panopticon, in which a watcher can observe the inmates of an institution, who in turn know only that they may always be seen, but not when. Thompson (2005: 40-41) writes: Whereas the Panopticon renders many people visible to a few, the media enable a few people to be visible to many: thanks to the media, it is primarily those who exercise power, rather than those over whom power is exercised, who are subjected to a new kind of visibility.
Goldsmith goes on to differentiate between the primary visibility that occurs with direct, first hand observation, and secondary visibility that comes from photographs, narratives, and videos, and the rise of networked sites through which to share them.
This new, secondary visibility has two corollaries that we wish to emphasize. One is that online, audiovisual archives of an event have different temporal features than inperson involvement, or seeing a print photograph or news report on the nightly news. The graphic spectacle of a Seattle police officer closed-fist punching a young woman was directly witnessed by the crowd of observers, yet the crucial moments of the altercation lasted less than 10 seconds. The circulating clip, which would long outlive the event it documented, created a new temporality for the incident. The pull and influence of a video generally spike quickly, then trail off, but attention can resurge at a much later date. Some images-even some montages that have been debunked-do so almost cyclically.
There is a stochastic periodicity to online circulation, a stop-and-go version of virality which is quite at odds with the suggestion of flow that circulation might otherwise suggest. This temporality impacts the subjects of the camera-enabled police gaze, such as the teenager, while expanding the "new visibility" of the officer.
These are the circumstances of the second corollary, which is that the photos and videos of officers floating around cyberspace are also put to use in policing. Here we are interested in the ways they are instrumentalized in what Foucault called subjectivation (assujettissement). 3 Becoming a subject is double-edged, Faubion (2001) has observed. At once it is "those processes through which individuals might make themselves into subjects" and also those "through which individuals are labeled or made into subjects […] the domain of 'subjection,' or of what Louis Althusser called 'interpellation' (Althusser 1971)" (Faubion, 2001: 12). This includes the disciplines and intersubjective socialization through which police are shaped and governed, but preserves a degree of individual, reflexive freedom in those processes. Freedom therein is not a right, but a practice.
As we explore next, police training at the Washington State Criminal Justice Training Commission is organized around subjectivation, and in this sense it is a space of ethical practice. Specifically, it is a milieu in which people are "invited or incited to recognize their moral obligations" (Foucault, 1997: 264) and also establish their own relationship to those obligations, ideals, values, and norms (Faubion, 2011: 24;Foucault, 1985: 27). This is not an endorsement of the moral obligations that hold sway or the techniques used to fashion the recruits, which, as the fieldwork presented here indicates, tend to systemically contribute to violence. What is important, if those conditions are to be addressed and for the discussion here, is that training is explicitly enacted as pedagogical and reflexive subject making, and, thus, is in the domain of ethics. To this end, we examine the role that the audiovisual artifacts of new media technologies play in the ensemble of experiences that summon the recruits to transform themselves into police subjects.
In contrast, body cameras and their footage are conceptualized as external to ethical field practice. Government, media, and activist discourses present the problem as one of ensuring accountability on one side, and privacy on the other. This is to be done through the technologies themselves (control over on/off buttons, automatic uploading to the cloud) and with laws governing their operation (with punishment for improper use). The ethical capacity of the decision-making subject is perhaps not irrelevant but simply not addressed in the debate over governance. Of course body cameras are not external to ethics-they do enter into the reflexive practice of considerable freedom that officers possess-and this omission has its own toxic subjectivational effects.
Both training and policing are understood here as possible ethical domains, which is to say that the actors in them could be expected to recognize and respond to obligations to cultivate themselves in relation to specific standards. At the academy, this is foregrounded while, we argue, in the police camera debates this is ignored. That is not to say that new media technologies or the diffuse surveillance they produce have but one, simple effect in each case. Rather, our analysis is specifically of how ethics are conceptualized and implemented at these two key intersections of policing and new media, drawing from the anthropological research to which we next turn.

Videos, training, and the practice of ethics
The Washington State Criminal Justice Training Commission (WSCJTC) is located on the forested former property of the Pentecostal Community Chapel and Bible Training Center, about 20 minutes by car from downtown Seattle. Thirty-six of the church's former acres are now home to the Basic Law Enforcement Academy (BLEA), the training program for all law enforcement agencies in Washington, except the State Patrol (which runs its own academy). The former chapel, with plush rose-colored seats for 2500 in curved rows, now hosts graduations ceremonies for the recruits. Paved walkways connect the different areas, and along one there is a waist-high row of signs that list selected virtues: loyal; ethical; honorable; selfless; courageous; dedicated; professional; confident. Here the virtues are literally guideposts on the path that recruits must walk, to be actively internalized in becoming a good officer. The signs were gifted by a previous class, each of which selects and undertakes a project before they leave (see Figure 1).
BLEA is a four-month training program that provides 720 hours of fundamental education in law enforcement procedure: a normal week of coursework could include fitness, firearms training, aspects of patrol procedure, and criminal law. Cohorts are composed of students from all over the state, most of whom live on campus for the duration of the training in the church's old dormitory or in one of the newer additions. 4 After completing the program with satisfactory marks, recruits then return to their home departments to work with field training officers, and some amount of training specific to their agencies, before starting as independent officers.
The group we accompanied for three months would graduate in 2012. With them, we attended selected lecture courses and "mocks" (problem-based simulation activities), and in parallel we conducted semi-structured interviews, and had other, more informal conversations with recruits, instructors, and administrators. We choose classes on their recommendations, as those which would address the situations in which officers might be video recorded, or might themselves record use-of-force. These included Patrol Procedure, Traffic Stops, Criminal Procedure, Crisis Intervention, Criminal Law, and Tactical Thinking. We also attended the lecture-based introductory firearms and defensive tactics training classes.
We asked the cohort about their home department's use of in-car or body-worn cameras. Although it was already clear that the role of cameras in policing would increase, at that time in mid-2012 only a few Washington local police departments were using video (or audio) recording devices. The Seattle Police Department was ahead of the trend, however. By far the largest force, it was fully outfitted with car cameras, and in December 2014, would begin testing body-cams with a dozen officers in their East Precinct. (Washington State Patrol, the second largest force, also used in-car cameras at the time.) At the academy in 2012, however, little in the classroom, or in the informal teaching that occurs throughout the day, had to do with the ambient surveillance into which the cohort would graduate. In part this was because, as several instructors remarked, training with cameras typically occurs at a departmental level, after recruits have graduated, when they begin to learn agency-specific policies and practices. The sheer quantity of material and hours of practical exercises packed in the typical academy curriculum preclude much emphasis on what will end up being locally set procedures for dash-or bodycams. Nor was much use made of cameras themselves, to record mocks as an aid to analyzing performance, for example, although some mentioned this as a future goal.
If, a few times, instructors mentioned that we live in a world where everyone has recording devices, it was with a nod to the researchers present. They made clear that any discussion of cell phone videos that captured police violence did not aim to make recruits more aware or attentive to the contemporary reality of surveillance. Rather, as one instructor put it, "we [officers] should be doing the same, professional job" whether on or off camera. That is, they aimed to train the recruits such that if they were recorded, the camera would but capture them behaving rightly. In-car cameras are helpful, he added, because they counter false accusations of officer misconduct. Specific training about being recorded in public was not only unnecessary, it could suggest that as officers, they would have things to hide.
Yet videos which were in one way or another the product of ambient surveillance were important as pedagogic tools. Many of the courses at the WSCJTC operate on a "problem-based learning model" (PBL), meaning that even in lecture situations, heavy emphasis is placed on using "problem" scenarios to think through concepts and develop general skills such as team-work, problem solving or risk management, while learning the more specific technical skills or legal knowledge of law enforcement procedure. Through these experiences, they were given a repertoire of technologies of the self which shaped them while also offering the means for reflexive self-formation. In this way, they were socialized into being a police officer.
Video recordings, while little taught or discussed as a topic, played significant roles in pedagogic methods. Out of 31 classes observed in their entirety, 12 made use of at least one video element, many of these using multiple videos. For the most part, these consisted of reproduced dash-cam footage, or, less often, had been shot on cell phones.
Many had made the nightly news, and what was shown in class had been copied from television, and overdubbed with captioning. They often had a grainy quality and advanced with the slight jerks of dropped frames, as the date and time of recording flickered in a corner. The aesthetic itself was an appeal to their origins; the value of the videos lay in the sure fact that what passed on the screen had really happened, and could happen to those watching.
The subject matter of all of the videos was officer-civilian interaction, primarily traffic stops. A few were used to illustrate situations that had positive outcomes. Afterward, the instructor guided the students in assessing what else the officer could have done to make the positive situation even better. The rest of the videos (28) presented situations where an officer should have acted very differently. These negative-example videos tended to be one of two genres: cellphone videos of officer misconduct; or, grimly, camera-captured death.
Sousveillance footage, from the cellphones of those caught in the mire of officer abuse, had been posted on YouTube or an analogous website, and then, for it to have come to the instructor's notice, been picked up by the mainstream media. In a "Tactical Thinking" class, for example, a video was shown where a Baltimore officer harasses three skateboarding adolescents. He berates them and hurls obscenities before attempting to get one of them in a headlock. The instructor used this video to set up a mock "community discussion" on the events. The video provided a lesson on "emotional intelligence", which he explained as training their responses to be consistent rather than emotional (e.g. excited or angered), even in taxing situations.
The other videos used in class showed the serious injury or killing of a police officer. A description of one such screening reveals its power. With the classroom lights dimmed, a dashboard camera video shows a police officer walking toward a "known troublemaker". He walks with far too little care, we are told by the voice-over. In an instant, the suspect jumps from his car with a rifle, opening fire. The officer is struck; he screams into his radio. Dispatch calls for backup. The officer crawls off the road, out of view of the dash-cam. The suspect pursues, and takes up the whole right side of the frame, pointing his gun downwards, off screen. We hear a few more chilling screams, and a few more shots. Then, only the cursing of the killer as he flees to his car, and the video ends.
The scene and the experience of watching it had a strong impact on all in the room. The recruits experienced the death of one of their own, mediated and yet in another sense, directly. They were shaken (as were we). The instructor looked around gravely, and with comments and questions drew out aspects of what had gone wrong in the officer's last interaction with a civilian.
Later, he observed that the growing together of recruits, and their fusion as a cohort is an important aspect of the training; it is the formation of an individual and collective identity of what it means to be an officer. Although videos of violent, real deaths can be readily found and watched on media-sharing platforms such as Youtube, Instagram, and Twitter, and with increasing frequency on the front page of mainstream newspapers, what happened in the classroom was a deliberate, and distinctive pedagogic technique. The experience of watching was elevated in intensity and drama for the recruits by their identification with the slain, and also further served to structure and solidify their developing understanding of themselves as police officers.
Training in this case is a matter of both pedagogy and active work on the self, and perhaps in some measure also one of violence: a wrenching experience to which those present voluntarily submit, but that proceeds inexorably and wrests transformation from its participants (see Samimian-Darash, 2013). The videos, however, are integrated in multiple forms of questioning and re-enactment that bring them into the domain of ethics. Dash-cam or cell phone footage would be used for recruits in groups to work through the steps of a specific scenario, such as a call for suspicious activity at a residential home, or a mock building search. In these cases, they had watched one outcome of an officer in an interaction or dangerous situation. Working through the scenario, they are inculcated with the importance of not making the same mistakes as that officer. "Mocks" provide live-action playing-out of the incident, with two goals that both depend on the video. One, students are guided toward visual cues so as to recognize, in a split second perhaps even before thought, the dangers revealed, and two, concepts of officer safety and communication, rationally grasped, can become embodied practice.

Cops, cameras, and the externalization of ethics
The incident of the officer punching the teenager, which occurred in 2010, was one of those used to request that the United States Department of Justice step in to investigate whether the Seattle Police Department engaged in unconstitutional policing through the use of excessive force or discrimination. The same civil rights leader who had arranged for the teen to apologize supported the request for a Department of Justice review, commenting "[t]wo wrongs don't make a right. What she did was wrong and what he did was wrong" (Miletich, 2010). Drawing on our attendence and conversations at police-community meetings during this period, what was evident was that police in Seattle had understood the incident one way: the officer, who was responding to citizen-reported concerns about dangerous jaywalking in the area outside a high school, had used reasonable force to stop an imminent attack and control the situation. Local activists and civil rights organizations, however, saw a cop punching a teenager in the face, and this, they argued, is what Seattle police do to the city's residents of color, when they are not doing worse.
Thirty-four community and civil rights organizations, together with the American Civil Liberties Union of Washington, demanded that the federal government investigate. The Department of Justice agreed, and its final report alleged unconstitutional policing, with a "pattern or practice of subjecting individuals to excessive force" (Perez and Durkan, 2011: 3). As this process was underway, a Seattle City Councilman began campaigning for police officers to wear body-cams. The technology, he argued, would help restore much-needed trust between cops and the community. Furthermore, he believed that there would be fewer problems, adding confidently, "[p]eople behave differently when they are on camera" (Kaste, 2011). From the beginning, however, local activists raised concerns over when and where the cameras would be used, what would happen with the footage, and who would have access to it. Previously, they had waged a successful legal battle to force Seattle to release a three-year log of squad car videos, so that people could check for their own. From their experience with dash-cams, it was clear that policies and protocols governing any new technology would be important, and a site of contestation.
In 2014, this question was brought to the fore when an anonymous requestor filed massive records demands with 20 different state agencies, including one for the police department's entire database of in-car camera footage. Washington state has broad public transparency laws. Anyone has the right to inspect and copy records of all state and local agencies, without providing a reason and with only a few exceptions. Officials acknowledged that they were obliged to provide the dash-cam footage, but pleaded that since even minutes of video had to be reviewed frame by frame in order to delete protected information, releasing anything meant impossible hours of labor. The police department was, they said, in the infernal position of failing to comply with either the state's disclosure laws or its privacy protections.
A couple of nearby small towns, also receiving the records requests, resolved to give up on body camera programs, but SPD's Chief Operating Officer reached out to the requestor on Twitter. Contending that they shared an interest in disclosing as much information as possible, he proposed a meeting. The requestor turned out to be one Tim Clemans, a self-taught 24-year-old computer programmer, and in a surprising, muchreported move, the police department hired him to figure out how to more efficiently redact their video and release it to the public (Herz, 2015). The department then sponsored a hackathon, ultimately selecting Cleman's proposal for "over-redaction" as the most promising. He set to work refining his code to blur and remove sound from the images automatically (see Figure 2), also experimenting with transforming the images into cartoon-like black and white outlines of the actual footage (see Figure 3).
In 2015, Seattle started a six-month pilot program under the national body camera initiative (eventually deciding in favor of long-term implementation) and launched a YouTube channel (SPDBodyWornVideo) dedicated to showing redacted footage captured by body-worn cameras. Although nearly illegible to the general viewer, someone involved in an incident would be able to identify what was going on and could request the pertinent segment in the original. Formal requests, once approved for release by the Public Disclosure Unit, were also posted in clear video to the channel. Police technology companies moved to incorporate automatic redaction software into the lucrative data management side of their body-cam packages as well, and by 2016, almost all of them offered agencies the ability to mark and blur specific objects or persons throughout a video.
In its pilot program, Seattle tested equipment from two of these companies. Taser's cameras have GPS tracking features and a 30-second capture buffer, preserving audio and video from before the officer even presses the record button. Vievu introduced a holster that automatically starts the camera when officers draw their guns, and a similar alert belt for other items, such as pepper spray or a baton. Another video system maker, Utility, offers to start cameras filming when the officer breaks into a run or is engaged in physical struggle; and a geographic action zone feature, in which devices are triggered to automatically record inside the specified radius of a high-crime or emergency area.
The shared objective is mechanical control over human action, technological insurance that incidents will be documented, even when urgency or willful obstruction enter into play. Correspondingly, SPD's policy manual consists of a set of rules that facilitate the smooth functioning of these technological controls. Officers are instructed to rule out any possibility of equipment malfunction before they begin patrol; training, while required, centers on camera placement and operation, correct use, letting people know they are being recorded, and a bland reference to learning the program objectives.
Body-cams are more intimate than in-car cameras, both because they are worn, and because being worn means that they enter private spaces such as the home, but the chastity belt logic underlying their governance is the same. The conduct of ethical conduct is externalized to the technology, and the result is often a recursive cycle of technological control and officer resistance. Half of the car antennas in one Los Angeles patrol division were found to have been sabotaged by officers in 2014, while in Chicago a full 80 percent of the dash-cam videos lacked audio due to what the department itself described in late 2015 as officer error and "intentional destruction" (Balko, 2016). Even before bodycam programs gained mainstream attention, activists, journalists, and the Department of Justice's own investigations were finding that in addition to damaging the equipment, officers would simply not turn on the cameras, manually block the lens, or have mysterious malfunctions (Kindy and Tate, 2015).
Ethics is removed from the conversation just as thoroughly by police efforts to place control entirely in the hands of officers. A long history of abuses (Kindy and Tate, 2015;Perez and Durkan, 2011;Westley, 1953), and the evident need to change the status quo of training and practice is by turns ignored and denied (Balko, 2016). Police chiefs, unions, and professional organizations have waged largely successful campaigns to be able to set their own policy. Of 138 state laws proposed in 2015 for governing cameras and their footage, 118 died in the face of fierce and organized police opposition; 10 of the 20 that were enacted set up legal blocks to public access (Kindy and Tate, 2015). And even while body-cams extend the police gaze, scholars note that officers conduct a "war against cameras", in which threat or force is used to crack down on the "unauthorized stare" of the public (Wall and Linnemann, 2014). The officer malfeasance which occurs is generally protected from consequences by peers and management. Although ambient surveillance is produced by the ubiquity of visual technologies, that does not mean there is ever a total record of a situation, and what is documented is subject to coercion, cannot be made immune to tampering, and is ultimately always partial.
Inasmuch as there is "a repertoire of 'technologies of the self' that belong to or must be fabricated from a collective repository" (Faubion, 2001: 12), that repository is just as diminished by pretending that good behavior is automatic as by framing it as merely a matter of technological and legal control. Without sustained attention to pedagogy and reflexive work on the self, as Chuck Wexler of PERF commented in 2015, "body cameras are just recording the policy and training that already exist" (Murphy, 2015). "Things" will continue to happen, he went on to say, "if we don't change the policy, and training, and tactics, and community trust […] because you're simply video taping existing practices" (Murphy, 2015).

Between techno-optimism and techno-pessimism
Technology, contended activist Gregory Malandrucco in 2012, "presents us with the unforeseen potential to hold public officials accountable for their actions in swift and certain terms" (Brucato, 2015: 48). Just as sanguine is the argument that cameras will produce well-behaved police and publics. These views, like the government investments in police camera programs, are founded on an optimism about the determining capabilities of these technologies. In that police institutions around the country are under criticism spurred by the increasing circulation of videos depicting death at officers' hands, there may be hope for change, although where and how remains undetermined.
It is worth noting that the multiplication of visibility is the realized dream of community policing, which as a doctrine says that a highly visible and socially engaged police force results in less crime and a greater feeling of order and safety (Kelling and Moore, 1988). The combination of the body-worn camera program and YouTube channel in Seattle is in keeping with this logic. The camera documents officers' words and interactions, which are then posted and made accessible to an unknowable and indefinite public. This potentiates surveillance of but also by the officer.
As Ben Brucato (2014) points out, however, "police violence is not a problem because of its invisibility". The violence is already visible, and a longer history suggests that documentation, while not meaningless, does not on its own change the systemic problems that are documented. Counter the argument that people behave better on camera is the fact that the incidents that have come to light have largely done so because they were recorded. Counter an assumption that accountability will ensue is the fact that police rarely face negative consequences even with video evidence. Juries tend to believe officers, and attorneys for the prosecution often need to work with that same police department on the next case.
What gets seen in a video is contingent and subject to certain, often racialized, ways of looking. Following Judith Butler (1993: 17) (and a similar argument made later by Allen Feldman (1994)), in the trial of the officers who beat Rodney King, the viewing of the Holliday video did not simply "speak for itself", but rather was already "fully schematized by racism". The beating of Rodney King was seen as justifiable, because his blackness marked him as "in total control" and "rife with dangerous intention", although he lay helplessly in the street. In this way, videos, like photos, are subject to the viewers' interpretation, and can be contested, edited, and contextualized to suggest multiple, quite opposite truths.
There is, then, a pessimism that goes along with the camera and increased circulation of its images, in which they are considered subservient to an always a priori regime of signs. Body-worn cameras serve to intensify the already powerful racialized gaze of the state, especially on marginalized populations. The specific regime within which the footage is deployed can vary, but the key argument made by Butler, Feldman, and Brucato is that the existent symbolic regime will coopt these images and forcibly control the ways in which they are used, unless, as Brucato (2015: 52) puts it, partisans engage in the political labor of countering these dominant readings.
In this spirit, we argue for a third position between the optimism of technological determinism and the pessimism of social constructivism, which stresses the coproduction and emergence of socio-techno subjectivities. John Dewey (1948: xxxvi), responding to the pessimism he saw in some interwar and atomic-era philosophy, argued that dealing with technological innovations and how they affect human life requires "a systematic exploration […] of the values belonging to what is genuinely new in the scientific, technology, and political movements of the immediate past and of the present". He was concerned with sloppy critiques of change in science, technology, and politics, affirming instead recognition of the tremendous growth in human capacities that scientific practice has produced, while underscoring that new science emerges within a context of values. A critique of old values and the experimental creation of new ones is, he argued, appropriate for changing terrains. In this lineage, careful attention through pragmatic inquiry to the actual relationships between ethics and politics is key.
As we have shown, one of the strongest moral obligations presented and internalized in training is for recruits to keep themselves and others safe. There are many consequences to this emphasis on what Steve Herbert (2001: 307) has described as "an unrelenting emphasis on officer safety" and a morality that constructs police work in relation to the "'ultimate sacrifice' that officers stand poised to make-to surrender their lives to ensure the well-being of others". Both characteristics often seem absurd or hypocritical to outsiders in the face of daily accounts of unarmed people shot by police, moreover in the historical context of the cycles of crisis provoked by these events. Understanding officer training and the way it deeply ingrains a way of perceiving the world as moral is central, however, to making sense of these police actions.
The strong emotional experience of watching the video is one of the specific sociotechno mechanisms through which recruits today are socialized and by which they shape themselves. Identifying danger and picking the right approach to neutralizing it become ways of doing the right thing. Risk assessment and response are techniques learned with the videos toward that imperative, and they help develop the "us versus them" mentality, an orientation and way of speaking about the uncertainty of encounters, and the importance of "coming home in one piece", documented in the earliest of sociological studies (Westley, 1953). 5 Evidently, then, this did not emerge with new media technologies nor is it inherent in them. Rather, skills in these arenas of performance, which training aims to make second nature, are linked to and framed by the videos and they are thus also possible points of intervention.
Likewise, body camera initiatives, in Seattle or elsewhere, raise a number of questions around how their policies explicitly and inadvertently form officers as certain kinds of socio-techno police subjects. Policing, as Karpiak (2013: 79) has explored, is "a type of sociality that is constituted through practice", and the practices that have emerged for police camera technologies serve to reinforce the existing system rather than changing it. Faubion (2011: 4) points out that "actors are never born ethical subjects", as though becoming an officer were the equivalent of having a certain sized foot, and finding the right shoe. But there are "styles and sizes" of subject positions which are available for self-fashioning and to which one must adapt. These are never wholly determinative nor are they entirely set, and thus it is in part because of the changes that technologies introduce that their governance is important, since in relation to them new subject positions may emerge. The omission of ethics from the debate has the toxic effect of doubling down on the options for being a specific police subject, rather than opening up new possibilities.

Considerations
We find contrasting ethical regimes in academy training and in the debates over police videos and their footage, and have endeavored to sketch some of the resultant ramifications. The mode of subjectivation in training emphasizes the moral obligation of officers to internalize certain precepts and strategies, significantly of risk assessment and response; governance through cameras externalizes ethics such that good behavior on the part of officers (and the public) is instead a matter of surveillance and the threat of later consequences. This reinscribes in policy, and in its implementation, the longstanding "us versus them" mentality, and forms of resistance for which police are known.
How policing changes, and how it does not, under conditions of ambient surveillance is a particular instantiation of the demands that advances in science and technology pose in general. At stake in inquiring into the subject positions that come into existence and how they are produced under these conditions is the question of if human flourishing will increase or not, and for whom. Dewey and Foucault shared a recognition that human growth and development are never free from a value context. Broadly, then, our inquiry into the actual uses and ethical practices of new media technologies is toward increasing capacities of experimentation and critique, without intensifying existing power relations. Such inquiry aspires to both an understanding of how subjects today are constituted, and the recognition of potential for new forms of life. 4. Recruits have already been hired by their agencies, which are responsible for 25 percent of the cost, around $3000 per session including ammunition. The state pays the remainder for mandated agencies, namely, cities, counties, state universities, and Fish and Wildlife, while the cost for non-mandated agencies, such as tribal, parks liquor control, and so on are considerably more, particularly if students require meals and lodging (around $7700 for FY2015). 5. Westley argued that this mentality has been around since the inauguration of the municipal police in the United States in 1844.