Credit...Kenji Aoki for The New York Times

Feature

Should We See Everything a Cop Sees?

Body cameras have been promoted as a solution to police misconduct. But the strange two-year saga of Seattle shows just how complicated total transparency can be.

On his first day on the job in the Seattle Police Department, Mike Wagers was invited to an urgent meeting about transparency. It was July 28, 2014, little more than a week after Eric Garner was killed on Staten Island, less than two weeks before Michael Brown was killed in Ferguson, Mo., and police departments around the country were facing a new era of public scrutiny. Wagers, who has a Ph.D. in criminal justice from Rutgers, was the Seattle department’s new chief operating officer, a 42-year-old civilian in jeans and square-rimmed glasses. He’d left his wife and two kids in Virginia and come alone to Seattle, a city he didn’t know — where it rained but cultural norms, he’d read, didn’t allow you to use an umbrella — because the job was what he called “the chance of a lifetime.” Seattle was the first big-city police department in a decade to have come under what is known as a consent decree — police reform by federal fiat — after a string of violent police actions against black, Latino and Native American people were caught on camera in 2009 and 2010. Wagers and his new boss, Chief Kathleen O’Toole, herself just arrived in Seattle, would use the best new thinking and the best new technology to lead the turnaround. And then Wagers would go home.

O’Toole’s eighth-floor conference room, which has views of City Hall and Elliott Bay and the snow-capped Olympic Mountains, was packed with top police and city officials. All eyes were on a lawyer from the city attorney’s office named Mary Perry. A former naval officer in her 60s, Perry was small and soft-spoken and favored pearls, and she was the city’s unrivaled expert in the seemingly mundane intricacies of the state’s Public Records Act. She was briefing the room on the potential fallout from a landmark case she had just argued and lost before the Washington State Supreme Court. The case stemmed from a series of public-records requests by a reporter for the local television station KOMO, Tracy Vedder, who began filing them after the same high-profile incidents that would lead to the consent decree. She asked for user manuals to the department’s new system of in-car dashboard cameras, then for lists of dashcam recordings, then for some of the recordings themselves. After the department denied every one of them, Vedder sued it for violating the Public Records Act.

The act, which dates to 1972, when governments ran on paper and our modern torrent of electronic data was unimaginable, is one of the strongest in the country. Washington State agencies cannot deny requests for records because the requester is anonymous or the request is too broad, nor can they deny requests simply in order to protect an individual’s privacy; instead, agencies must redact only the details deemed sensitive under state code — for example, some addresses, sometimes the face of a minor — and disclose the rest. Before the court, Perry had argued that a different law, the state’s Privacy Act, which allows departments to withhold recordings until related criminal or civil cases are resolved, should take precedent and the Seattle Police should be allowed to broadly deny Vedder’s requests until the relevant statute of limitations ran out. The court disagreed.

As Perry now told Wagers and the officials gathered in the chief’s conference room, Seattle and other departments across the state were operating in a new reality. Only in cases under actual, pending litigation could the police withhold video footage from the people. This presented several problems. The first was logistical and financial: Seattle Police were sitting on more than 1.5 million individual dashcam and surveillance videos, or about 300,000 hours and 350 terabytes total. Before releasing any footage, someone in the department had to review and redact it in accordance with the specific privacy exemptions the state code did have. The process was manual, a painstaking, frame-by-frame ordeal. By one estimate, 169 people would have to work for a year just to fulfill the department’s existing video requests, and the department added 2,000 new video clips daily. Perry feared that the new flood of data, especially but not exclusively video, could bankrupt Seattle if someone requested it all. “It’s like being on the Titanic,” she later told me, “and you’ve got a teaspoon to bail.”

The second problem was privacy. Dashcam videos were already a concern, but Seattle had also been considering using body cameras. In fact, the Police Department was now preparing a small pilot program involving a dozen officers from a single precinct to test the hardware. Any footage that bodycams gathered, Perry warned, would also be subject to the Public Records Act. The department would have even more video to manage and release, but most worrisome was how fundamentally different, and more intrusive, this video would be. Unlike dashcams, bodycams, which are attached to an officer’s uniform or purpose-built glasses, can go into homes and hotel rooms. Unlike dashcams, their default view during an arrest, or during a simple conversation with a victim or witness or informant, is an intimate close-up. Many people think of body cameras as a tool for police accountability, but the primary subject of their surveillance isn’t the police — it’s the public.

“I believe in open government, I really do,” Perry told me, “but I don’t think people have really wrapped their heads around all the implications.” If the bodycam pilot was deemed a success, and the city expanded the program to the rest of its 850 front-line officers, all of them now walking surveillance cameras, what then?

Wagers left the meeting stunned. “I like to think I can walk into a room, take a complex problem and break it down into its component parts,” he says. “This meeting, I walked out thinking: Wow, this is complicated. This is messy.”

In 2011, three years before Wagers joined the department, a 20-year-old programmer named Tim Clemans set the record for consecutive daily visits to Seattle’s Space Needle. “The main reason I go everyday,” he wrote in an online journal documenting his record attempt, “is because on July 7, 2010, I attempted suicide by jumping off a bridge. The Space Needle changed me from a depressed, shy and lazy kid to a happy, outgoing, disciplined young man who has a reason to live.”

Clemans lived with his parents in a small, suburban house south of Seattle, sleeping in a bedroom decorated with a picture of the Space Shuttle and a free calendar from a local Chinese buffet. His bed was a mattress on the floor, and his electronic devices — computers, radios, keyboards — dominated the room. Clemans, who has sandy blond hair and wide blue eyes and walks with the stooped gait of someone who spends a lot of time in front of a computer, taught himself to code HTML at age 8. Home-schooled as a teenager to escape bullying, he was editing web pages for a hospital, where a friend from his parents’ church worked, by 14. He taught himself basic JavaScript and CSS, and he learned the programming language Python after he showed up at a mathematics lab at the University of Washington and offered to volunteer in exchange for high-school credit. “I just went there,” he explains. “I’m very impulsive, and I don’t like rules.”

Day 1 of his record attempt was a fluke. He was in the vicinity of the Space Needle and bored, so he took the elevator to the top. Something about being able to see the whole city at once captivated him. The next day, Clemans bought a $50 season pass. He learned of the previous record — 60 consecutive days — and vowed to beat it. He came daily by bus and foot, an hourlong commute, often arriving early and staying late into the night.

The Space Needle blooms out of the grounds of the 1962 World’s Fair, a site north of downtown that is now known as Seattle Center. It is 605 feet tall. You enter through a plaza choked with Korean tourists and Andean flute musicians and Marine marching bands, then shuffle up a ramp to the east, above the gift shop and not far from a lineup of idling tour buses. Before handing over your ticket, you are asked to pose for a photo, then board one of three external elevators. When the elevator starts its ascent and the city comes into view, the people inside either fall silent or gasp out loud.

From the observation deck, you can see mountains in four directions, water in two and, just to the south, a forest of skyscrapers. You also see a tangle of cranes; a city increasingly built by the likes of Jeff Bezos and Bill Gates is booming. There are parks, markets, hidden gardens, tiny houses, tiny cars. Bulk carriers glide into port. Seaplanes land and take off. Helicopters float by. The lights of the radio towers on adjacent Queen Anne Hill, a seeming stone’s throw away, blink on and off. Inside, a giant digital wall displays a shifting collage of visitor selfies, along with tourists’ names and cities of origins. A joystick controls a high-definition camera mounted somewhere on the tower, allowing anyone to pan and zoom in on the unsuspecting people below, like a drone operator.

Image
Tim Clemans got into police data as a “hobby.”Credit...Ian Allen for The New York Times

The Space Needle became Clemans’s office and his sanctuary. He brought his laptop, stuffing it into his backpack along with a toothbrush and toothpaste and self-help and business books, and got online with a wireless card. He exercised by walking laps around the outer deck. He offered to take tourists’ photos, posed with them and answered their questions. He wrote code and blog entries. He stared at the city. It made sense from up there. It had patterns. Even its famously chaotic traffic had a predictable ebb and flow.

“The sun is shining!” reads one blog entry. “I can see Rainier. People are happy. It doesn’t get any better than this.” When he claimed the record on Day 61, he appeared on the evening news. He said he would keep going for a full year, then announced on his blog that he would go for even longer — 1,825 days in a row. He began wearing a black jacket emblazoned with the words “Record Holder” in block letters and spending more of his time talking to tourists, meeting people from all over the world. “It was an enjoyable eight-hour visit,” he wrote in another entry. “I finished reading ‘Personal MBA’ and memorized the five parts of a business and the 12 ways to create value.”

Some days he wrote about his attempts to stick to a routine. Others he dreamed up fanciful lists. “When I own the Space Needle in 2034,” he wrote on Day 107, “I will do the following twenty things.” One: “I will sell tickets for $200 to the very top ... for $600 you can also hang off the side.” Two: “Each day the Space Needle will fly a flag. Most days the flag will be the photo of a random lucky guest.” Four: “For $500 you and your spouse can sleep in an elevator.” Sixteen: “I will pay French Spider Man Alain Robert $10 million to climb a leg of the Space Needle once every day for a year.” Eighteen: “I will give the security bag checkers X-ray vision glasses.” Nineteen: “I will personally lead a walking tour for people struggling with suicidal thoughts.”

“This post is for me,” reads the next day’s entry. “I’m struggling to meet my goals like reading and writing computer code four hours a day. However, I’m really good at meeting this one particular goal, going to the Space Needle every day. I’ve been able to do so for 108 days straight for one simple reason: Going to the Space Needle everyday is more important than anything else. I have to do it. So when I’m trying to achieve a goal, I just have to remember it’s more important than anything else.”

After almost 160 days, Clemans had an altercation with a Space Needle employee. Before he could get on the elevator, she asked him to pose for a photo in the booth at the top of the ramp, just like the tourists. He had been asked to do it a hundred times. It was senseless, and it enraged him. “I blew up at her,” he says. Just like that, the world-record holder was banned from the Space Needle for life.

Adrift, he disappeared from public view. He worked on code, dabbled in robotics and considered a career as a paramedic. He was pulled over for reckless driving and hired a lawyer known for his successful dashcam requests and, he claims, the video didn’t entirely match the police report, so the court reduced the charges. He met a woman on OKCupid, getting her attention by suggesting that they go build sand castles together.

Image
Credit...Screen shots of bodycam and dashcam videos via YouTube.

One day in September 2014, Clemans was with his girlfriend when he read on his laptop that KOMO had prevailed over Seattle in the state’s Supreme Court in June: Police video had to be released on demand. The Justice Department had just started its investigation of Ferguson, but Clemans’s concerns were closer to home. KOMO posted only snippets. “I was just mad that they were getting all this video but not making it all available,” he recalls. He turned to his girlfriend and told her, “I think I’m going to do police data as a hobby.”

“It was like a D.D.O.S. attack,” Wagers says, comparing Clemans’s first email to a distributed denial of service attack, in which hackers send so much traffic to a website that it crashes. “It was going to seize up the system.” The request was simple, but it was just what Perry had warned about: Clemans wanted every single video the Seattle Police Department ever recorded, everything not tied up in an investigation. That September, Clemans sent similar messages to almost every police department in the state. Worried the police would retaliate if they could find him, he used an untraceable email address: policevideorequests@mail2tor.com.

Some police departments started sending videos immediately, some proposed installment plans and some announced they were delaying their bodycam programs. In Poulsbo, a town of 9,500 people across the Puget Sound from Seattle, the mayor desperately emailed her local state legislator and asked him to do something about the cameras and the Public Records Act.

Unlike its smaller counterparts, the Seattle department did not give its unknown nemesis any footage. Clemans responded by programming a bot. It scraped the department’s website for new case numbers, then automatically requested the corresponding police reports, firing off emails 10 times a day. The more the authorities denied him, the more his appetite grew. He asked the University of Washington for all its records dating back to “the formation of the Earth 4.54 billion years ago.” He filed requests with another 60 state agencies, demanding every email they had ever sent — 600 million messages in all, according to a state estimate. The Department of Agriculture informed Clemans it would need 132 years to complete the job.

Wagers decided to try a different approach. At Rutgers, he had been schooled in the reformist ideals of community policing. Like the longtime New York City police commissioner William J. Bratton — whose work when he led the Los Angeles Police Department was the subject of Wagers’s Ph.D. dissertation and who, as it happened, was Chief O’Toole’s boss when he led the Massachusetts Metropolitan District Commission Police — Wagers liked to quote Sir Robert Peel, the founder of London’s bobbies, who in the early 19th century established an Anglo-Saxon tradition of policing distinct from the militaristic us-versus-them Continental model. Peel taught that good policing was about building and holding public trust. “It’s about engagement,” Wagers told me. “We respond. We engage. It sounds kind of trite, but it’s a different way of thinking.” His new plan for dealing with Clemans, he says, “was like any other response to the community: We engage.”

On Twitter, Wagers followed @PoliceVideo, an account that then belonged to the still-anonymous requester. The next day, the account tweeted at Wagers, asking him why the department was giving PowerPoint presentations to city leaders about the coming bodycam pilot program but not sharing the slides with the public. “Screw it,” @PoliceVideo tweeted when Wagers was too slow to respond. “Putting in the request now. Wish you guys would just publish this stuff.”

Image
Chief Kathleen O'Toole of the Seattle Police Department.Credit...Ian Allen for The New York Times

But Wagers offered to “do you one better,” and he tweeted a phone number. “Here’s my cell,” he wrote. “I give it out to everyone. Call me & happy to answer questions.” It was 6:57 p.m. One minute later, Wagers’s phone rang. It was Clemans. He still didn’t give his real name — he had been overwhelmed by media interest in his transparency fight — but readily accepted when Wagers invited him to lunch.

The next day, Clemans met Wagers and Mary Perry at Police Headquarters. They ordered a pizza. Perry tried to explain that it was technically impossible for the department to release all its archived video anytime soon. Clemans countered with an idea he had. He called it “overredaction.” If the system was paralyzed by the need for frame-by-frame redactions, he wondered, why not automatically redact everything? That is, instead of jockeying with requesters, painstakingly reviewing each video, blurring out the protected parts and burning the results onto DVD after DVD, the department could just use software to lightly blur everything, then proactively publish each blurred video online. Push, not pull. Viewers would be able to make out enough to know if footage merited a specific records request and a more precise manual redaction, and they would presumably ask for only the segments they thought they needed; and no longer would the department be buried under its own video. He agreed to drop his mass requests if the department would try it out.

In fact, Clemans said, he had stayed up all night writing a very rough version of the overredaction code and printed it out. He showed them what he had done.

The pilot program, set to be rolled out in December 2014, would garner considerable national attention. After Ferguson, expectations that bodycams could reduce police violence and ensure accountability were high. In polls taken over the next year, 87 percent of Americans and 89 percent of Seattle residents supported their use. When President Obama announced an initiative to distribute $75 million so police departments around the country could buy more of them, he declared that the cameras would “enhance trust between communities and police.” A widely cited study by University of Cambridge researchers, undertaken in Rialto, Calif., in 2012 and 2013, found that when officers wore the cameras, use of force dropped by 59 percent compared to the previous year, complaints against the police by 88 percent. By now Wagers understood more about the hidden costs of bodycams — to individual privacy, to the overworked staff of the public-disclosure unit — but he and O’Toole began avidly pursuing a federal grant nonetheless. “If research continues to show that they reduce the use of force, reduce complaints, produce positive impacts,” he said, “then there’s a moral cost to not using bodycams.”

The cameras that Seattle planned to use in its pilot program — which would be worn on patrol by a dozen officers in its East Precinct, their video uploaded to a server at the end of every shift — were from two local companies: Vievu, which was founded by a former Seattle SWAT officer, Steve Ward, in 2007, and its main rival, Axon, a division of Taser that is responsible for the reliable spike in the parent company’s stock after major police shootings. Together the two companies dominate a market that analysts believe will soon be worth a billion dollars a year, with much of the value coming from software and storage. Both Vievu and Axon offered bundles that paired their cameras with editing programs that ran in the cloud, processing them on computers maintained by Amazon Web Services or Microsoft’s Azure, which are also Seattle-area companies.

But neither bundle came cheap. The millions of dollars that major police departments were beginning to spend on bodycam contracts were millions they weren’t spending on better training or new officers, and neither company had come close to perfecting automated redaction, without which Seattle Police would simply be collecting more footage without any better way to share it with requesters. Seattle needed an end-to-end software fix, Wagers realized, and Seattle was also a city full of programmers. So the day before the pilot program started, he also tried another experiment: With help from Clemans, he held the department’s first-ever hackathon.

The overflow crowd of 80 people was “straight out of central casting,” Wagers says. “Skinny jeans. Hoodies. People who had sued the department. People I know I’d seen on the protest line the previous week.” Clemans demonstrated his overredaction program: the world as if filmed through a beer glass, fuzzy but familiar, in which it was still possible (too possible, some worried) to tell who was tall and who was short and who was male and who was female and who was black and who was white and who was running and who was chasing. It was well received, and several programmers in the room offered to help Clemans improve the algorithm. But Clemans, volunteering his time and his skills, would be the linchpin. The pilot would run for six months, and if it went well, 12 officers would soon become 850.

The next day, on Dec. 20, as TV stations aired reports on the department’s innovative approach to transparency, the bodycam pilot began. Clemans and the department soon developed a rhythm: Some of the videos coming in from the East Precinct were saved to U.S.B. sticks, and Clemans, unpaid and on food stamps at the time, came downtown by bus to pick them up. He processed them on his bedroom computer, tweaking the overredaction program as he went, trying to code in the right balance of concealment and transparency.

Even as he helped run Seattle’s experiment in open government, Clemans continued to receive hundreds of hours of dashcam and bodycam video from other police departments around the state, much of it unredacted. And, just as he had promised to do in his records requests, he was continuing to upload it — unredacted — to YouTube. The first videos he posted show, with perfect clarity, drunken-driving stops from the towns Renton and Tukwila, south of Seattle, and include disturbing footage of a young white man being shocked with a Taser in the dark beneath the tree where he had just crashed his car, the scene illuminated by flashing lights. In another video, two officers rush to help a man suffering from a heart attack on a freeway overpass, saving his life. Others depict foot chases or friendly interactions with homeless people, and they gave Clemans more respect for the police and the difficulties they face every day. “Policing is just really hard,” he says. But he didn’t watch all the footage himself, because he didn’t have time and new projects beckoned.

I watched more than a dozen hours of the video. There were no shootings. Most of it was routine police work, which may be why it was so disturbing to see online. In one set of videos that Clemans posted, a well-dressed white mother of two is pulled over late on a Friday night. She is composed until she fails a sobriety test. Then she pleads for a chance to call the babysitter — according to Facebook, where I found her because the video also revealed her name, she has a young daughter and younger son — before breaking down in loud screams and sobs in the back of a patrol car.

Elsewhere in the state, also on YouTube, a body camera captures the inside of a woman’s home as she explains how her accused stalker has been ignoring a restraining order. A father pleads with two officers to check in on his adult daughter, who he says has intellectual disabilities and has become involved with a man he believes is dangerous. Two college roommates accuse each other of theft and intimidation in their living room. A white woman apparently overdosing on meth, who cries out that she’s pregnant, is restrained and given medical attention. A middle-aged black man opens his apartment door for a group of officers who inform him that he has been screaming all night. Slurring his words, he admits he’s suffering from PTSD and that he has a history of heroin use. He refuses to let them take him to the hospital. He refuses even when an officer reminds him that he has been screaming like this off and on for a month. But now you can hear the fear in his voice. “Am I screaming right now?” he wonders.

One of the first bodycam videos that Clemans uploaded shows a young black woman sitting on a bed in a hotel room. The officer wearing the camera is a disembodied voice but for a fleeting glimpse of his face and torso in a mirror. His manner is professional and sympathetic, and he says he doesn’t want to arrest her. “Will I be charged?” she asks. “Let’s get to that later,” he says. He asks her if she would be willing to talk to a counselor. “So you can get some other kind of job so you don’t have to do this anymore, O.K.?” he says. “Our ultimate goal is that there is no prostitution, O.K.?”

Image
Mary Perry, Seattle Police Department’s Director of Transparency and Privacy.Credit...Ian Allen for The New York Times

He begins filling out a form — Escort Face Sheet — on a clipboard. She answers every question, sharing the intimate details of her life. She tells him about her relationship with her boyfriend, her clashes with her strict father, her time as a runaway, her drift from strip clubs to Backpage.com escort ads, her few regular johns. She claims she’s new to this work. She explains that she charges per hour or half-hour. She has a dog, she says. She can rely on her parents in times of need, she says. She gives him their street address. She gives him her full name. She shares her private email address and phone number. She shares all this with the camera too.

When the interview is done, she asks about it. Is the image clear? “It’s pretty clear, yeah,” the officer says, but “if the press wanted it, we can redact faces — can blur out the faces and whatnot.” Body cameras are new to his city, and he doesn’t know that what he just told her isn’t entirely true. A lawyer will decide that the video is in no part exempt from the Public Records Act, and the officer will later be shocked to see it on YouTube. He will try and fail to have it taken down. The woman in the video is easy to find in her other internet life. She’s on Facebook, Twitter and Pinterest, where she chats with friends and posts images of dresses and animals and nail polish. You can visit her parents’ house on Google Street View.

Body cameras can be knocked loose when the police rush a suspect, as the world learned in July when Alton Sterling was killed in Baton Rouge, La. Officers can likewise forget or claim to forget to turn the cameras on in time to record an incident in full, as protesters saw after they forced the police to release the inconclusive, partial bodycam footage of Keith Lamont Scott’s shooting in Charlotte, N.C. But for slow-moving scenes like this one, in the hotel room with the young woman who seems to trust the officer with her privacy, the technology works almost too well. Bryce Newell, an information-science researcher now at Tilburg University in the Netherlands who did his fieldwork in Washington, interviewing Clemans and Wagers and riding along with officers as they tested their new bodycams, gave a clever name to the problem they posed in a society demanding transparency: “collateral visibility.”

One of the first people revealed to the public by Clemans’s transparency quest was Clemans himself. His mass records requests had drawn the interest of local reporters, who started filing requests of their own to the Police Department, seeking his identity. After they got his phone number, Clemans pre-emptively outed himself, embracing the role of tech seer. He published a letter in January 2015 explaining his mission on the website of the Seattle Privacy Coalition. “I pushed the envelope,” he wrote, “so we as a society can once and for all address accurately recording of the truth, who should have access to the truth and what we are to do with the truth.” He told the local news site Crosscut that the Public Records Act did need to be amended. “It’s not going to change until it becomes a massive problem,” he said. He told Seattle Weekly that his experience with police videos had convinced him that certain things shouldn’t be made public. He left the footage on YouTube precisely to make that point. “I don’t think people are going to deal with this until they have an emotional reaction,” he said.

The designated voice of the people in Seattle’s police-reform process is the Community Police Commission, a board of local leaders — black, white, Latino, Native American — that was created as part of the city’s settlement with the Justice Department. It represents the people the consent decree, and body cameras, were most meant to protect. On a cloudy Saturday in January 2015, with the pilot program underway, the commission invited Wagers to a community meeting on the city’s south side to take part in a public discussion of how bodycams would be used and whether, a flier read, “they will, in fact, increase police accountability.”

Wagers stood before the crowd, about a hundred people in all, and gave a brief technical overview of the pilot program. Then an officer from East Precinct demonstrated how the bodycam worked, briefly filming the crowd as he did so. After that, one commissioner, David Keenan, a local lawyer, asked a group of waiting panelists to state, one after the other, when, exactly, the cameras should be on, and to answer a simple question about the evidence they create: Why do we want it? That is, whom does it serve? “Should it only be used if there’s an accusation against a police officer?” he asked. “Should it be used by criminal investigators and by defendants and by prosecutors?”

Image
The Seattle Police Department avoided the need to edit every video it had taken for privacy exemptions by blurring everything.Credit...Seattle Police Department screen shots via YouTube

The growing national welter of contradictory state laws and department policies signals that there is no settled answer. Some allow officers to view it before writing their reports. (That was the central issue earlier this month in Portland, Ore., that caused protesters to storm its City Hall, where police officers met them with bursts of pepper spray.) Some do not. Some treat all bodycam videos as public records; some limit their release. Some mandate a quick deletion of police video; some have no requirements to delete it at all. Some restrict the use of facial recognition software; most allow it. A technology meant for one purpose, once unleashed, is soon used for many others.

One commissioner, Jay Hollingsworth of the Mohegan tribe, the chairman of the John T. Williams Organizing Committee, named for a hard-of-hearing Native American whose 2010 shooting was partly captured on dashcam, answered Keenan’s “when” question simply: Bodycams should always be on. To his right, another commissioner, Jennifer Shaw, the deputy director of the A.C.L.U. of Washington, said that she agreed in theory but that her organization supported the cameras only if they were used exclusively for police accountability. Next, Ron Smith, a detective who then headed the Seattle Police Officers’ Guild, said that he was also in favor of bodycams because he believed they would exonerate officers from complaints. But the cameras went into homes and on lunch and bathroom breaks, and officers absolutely had to be able to switch them on and off.

At the end of the table were two activists, Marissa Johnson and Dan Bash. Johnson later became nationally known for leaping onstage at a Bernie Sanders rally, claiming the candidate’s microphone and telling him he needed to do more to acknowledge the Black Lives Matter movement. “This conversation about bodycams is a complete and utter farce,” she said. Forget the policy details. She did not consent to being recorded at all. She did not trust the police to do the right thing with the footage. Most of all, she did not trust prosecutors to use it to prosecute officers. “Why do I need a home video of my abuse that’s going to be filmed by my oppressor?” she asked.

An uncomfortable silence filled the room. Then Hollingsworth announced that the other panelists had made him change his mind, leaving the two employees of the Police Department, Wagers and Smith, two white men, the only people onstage who clearly supported bodycams. The mood became tense, and members of the crowd interrupted the next speaker and soon assumed control of the room, taking turns denouncing police brutality and glaring at Smith, who was eventually escorted to his car by another officer.

Three weeks later, the commission, the voice of the people, called for Seattle to delay full deployment of bodycams until the state law could be rewritten. “This is a new technology,” the commission said in a news release, “which may have unintended consequences.”

Much of the moral case for bodycams, that they reduce police violence, rests on a single experiment: the 2012 Rialto study that Wagers referenced. Rialto is a city of almost 100,000 people, most of them Latino but with significant black and Pacific Islander minorities, that sits in the desert and sprawl east of Los Angeles. Its population is much smaller and more diverse than Seattle’s, its police force is much smaller and its murder rate is significantly higher. Of its 115 sworn police officers at the time of the study, 54 regularly conducted patrols, and all 54 were issued body cameras courtesy of Taser. For a full year, half of the front-line officers on a given 12-hour shift were randomly assigned to wear their cameras, while the other half served as the control. The data from nearly 1,000 shifts and 50,000 hours of police-public interactions showed that when officers wore bodycams, they were less likely to use batons, Tasers, firearms and pepper spray or to have confrontations that resulted in police-dog bites, and they were far less likely to receive civilian complaints about their conduct.

Image
Mike Wagers, the former chief operating officer of the Seattle Police Department.Credit...Ian Allen for The New York Times

In an essay published shortly after the White House announced its $75 million in bodycam funding in 2014, two authors of the study, Barak Ariel and Alex Sutherland, hypothesized that it was not cameras alone that drove the positive results; it was the fact that before every interaction with a citizen, officers in the trial were required to announce that they were recording. There may have been a “self-awareness effect”: Both parties were reminded at the moment of contact that they were under surveillance and that they should behave accordingly. One question was whether the effect would hold up if officers did not announce the cameras’ presence. Another was whether it would hold up when the cameras lost their novelty.

Ariel and Sutherland also worried about anyone basing decisions on a single study, which, no matter how rigorous, could well be a fluke, “the statistical equivalent of ‘luck,’ ” they wrote. The Rialto sample was small. The notable drop in citizen complaints, for instance, was from 28 in the yearlong period before the study to 3 while it was underway. At least 40 reasonably scientific studies have followed Rialto, but many, according to a recent survey carried out by George Mason University, have yet to be published.

Rialto suggests a drop in the use of force. A study from Arizona suggests a drop in arrests but a rise in citations. A separate Arizona study, along with one from London, suggests a rise in arrests. When Temple University researchers recently sifted through a Washington Post database of 986 deadly shootings of civilians by police in 2015, they found that when officers wore body cameras, civilians were 3.64 percent more likely to die. The increase was more pronounced, 3.75 percent, in the deaths of African-Americans and Hispanics and a barely measurable 0.67 percent in the deaths of Caucasians and Asians. They hypothesized that a different kind of self-awareness was at work: Officers, aware of their bodycams and more certain their use of deadly force would be seen as justified, were less likely to hesitate.

After Rialto, Ariel and Sutherland set out to replicate their influential study on a much larger scale, collaborating with eight large and small police departments covering two million citizens in the United States and Britain. Their resulting analysis of 2.2 million officer hours, published this May, seems to validate their concerns that Rialto offered an incomplete picture. It found that police use of force actually went up by an astonishing 71 percent when officers could turn their cameras on and off at will and went down (by 37 percent) only when they recorded nearly every interaction with the public from start to finish. A second analysis of the same data, published late last month, supports Rialto’s finding that citizen complaints drop significantly, almost to zero, when bodycams are present, while underscoring the authors’ warning that more research is needed to understand how and why and under what conditions. “It may be that in some places it’s a bad idea to use body-worn cameras,” Ariel said in announcing the May results, “and the only way you can find that out is to keep doing these tests in different kinds of places.”

The Seattle Police Department introduced its new YouTube channel, “S.P.D. BodyWornVideo,” on Feb. 23, 2015, less than two weeks after the Community Police Commission warned about unintended consequences in its news release. Featuring footage from the pilot program, with the images automatically blurred and the audio muted, it was a national first, and tens of thousands of people tuned in. The earliest videos were of Seattle’s explosive Martin Luther King Day demonstrations the previous month, and they appeared on the channel as a kind of fever dream. Protesters, rendered in fog and grayscale and digitally stripped of their voices, were mostly unrecognizable as they marched onto a freeway demanding justice for Michael Brown and were arrested and shackled on the ground and then trundled into police cruisers.

The videos didn’t shake the local opposition to bodycams, and they didn’t meet the standards of the state Public Records Act — if asked, Seattle would have to provide more precisely redacted versions — but they met a national need for positive news about the police. The YouTube channel was featured on the “Today Show,” and the department’s innovation was the subject of coverage, some skeptical and some not, by The Guardian, Vice News, “Marketplace,” “All Things Considered” and The New York Times. The tech press couldn’t get enough. One well-reported article in the online publication Backchannel called Clemans “the body-cam hacker who schooled the police.”

Wagers began receiving calls and email from departments around the country. The future of policing, he saw, was now more than ever reliant on what technology the police could afford to buy from companies like those that built modern Seattle. He pledged that Clemans’s code, once completed, would be open-source and freely shared so that smaller agencies, which didn’t have the staff to handle video-records requests, let alone the money for bundled bodycam contracts, could stay on the cutting edge. In one overredaction test, the department processed 2,400 videos in three hours on rented Amazon Web Services computers in the cloud, a task that would have gummed up the public-disclosure unit of the Seattle Police for weeks had the files been redacted manually. The computer rental cost the department $1.20 — 40 cents an hour — Wagers marveled. To meet the demands of the Public Records Act, all Clemans needed to do was get his algorithm to redact only faces and other private details, rather than the full frame.

Clemans was now an employee: The department had offered him a six-month contract at $22.60 an hour. But when the pilot program ended in July, and the question of whether to expand it loomed, he still had not perfected his code. He was trying to solve a problem, mostly by himself, that teams of professional engineers at Taser and its rivals have yet to fully crack. Distracted, he noticed other inefficiencies at Police Headquarters and he felt increasingly compelled to fix them. After visiting the understaffed 911 center, then in the news for its slow response times, he came up with a new idea: a program that automatically highlighted the most pressing calls, allowing dispatchers to be more efficient. But the captain in charge of the 911 center didn’t welcome his intrusion, and he belittled the idea at a tense late-August meeting brokered by their superiors. Clemans, by his own admission, “got really mad.” He swore at the captain. He yelled until he had to be escorted from the building. And just like that, in an echo of his Space Needle blowup, he was exiled from headquarters. He made a threat on his way out: “I’m going to P.D.R.” — public-disclosure request — “the [expletive] out of you,” he shouted.

In September 2015, when Seattle won a $600,000 Justice Department grant to expand the bodycam program to the rest of the police force, which the mayor pledged to more than match, Clemans wasn’t there to help celebrate. He was working remotely, back in his bedroom or hunched over a laptop at the Starbucks he now frequented on the 40th floor of Columbia Center, the tallest building in town, where the views were stunning and free.

Days after the grant announcement, Attorney General Loretta Lynch came to Seattle to declare it a national model for police reform. The federal monitor overseeing the consent decree, who had previously lauded the overredaction experiment when calling for rapid deployment of bodycams, now praised the department’s systematic approach to investigating any use of force — a significant change from the cursory reporting of the past. Wagers, meanwhile, was helping Chief O’Toole to select his replacement. He had by then been in Seattle and away from his wife and children for more than a year.

On Oct. 29, 2015, a week after Wagers announced he was leaving the department, Clemans texted in his own resignation. He no longer believed the department could ever truly be transparent on its own, he told me. His time inside the department, far from co-opting him, had weaponized him after his blowup with the 911 captain: He knew even better which records to request. He wrote a new bot capable of sending the department up to 10,000 requests in 24 hours. He asked for 911 records, “tables & columns of all databases,” all front-facing dashcam video from 2014 and “all computer codes written by Tim Clemans.” Over the next few days, he piled on, requesting “all roll-call notes,” text records from the department’s license-plate scanners, the source codes of key department software and “all Outlook calendars of supervisors, managers & command staff.” He also started building a website where he would publish it all, which he called the People’s NSA. He expanded his demands to include every public record in every format — paper, tape or digital — the City of Seattle had. He soon asked 39 neighboring cities for the same. By the end of 2015, according to a study on public-records requests he obtained via a public-records request, Clemans had filed 2,272 requests to the Seattle Police Department alone, making him again its No. 1 antagonist. One day at her office, Mary Perry showed me a printout of the latest ones, five pages long. “We’re being crippled,” she said. Wagers was on his way home, Clemans was making more demands than ever and she was back where she started.

This summer, I joined Clemans at a Black Lives Matter march in Seattle. He was there to watch the police as protesters demanded justice for Philando Castile and Alton Sterling, whose deaths were captured not by bodycams but by citizens’ phones. His overredaction code was about to be revived. Wagers had taken a job at the Virginia offices of Amazon Web Services, the hidden architecture behind America’s police-video revolution, where he was contacted by a company that wanted to merge Clemans’s software with the website crimereports.com. Now, a prototype was in the works, with support and partial funding from Amazon.

Searching for blurred video of a crime down your block may someday be as easy as searching for a restaurant on Google Maps. But a new state law making footage harder to request in bulk — the Tim Clemans Act, his friends called it — had just gone into effect thanks to the efforts of Poulsbo, the small town that had contacted its state representative. Clemans had also been studying YouTube’s privacy policy, and he would soon decide to remove the videos he posted there from public view. For now, though, he wore his own D.I.Y. bodycam, a cellphone strapped to a chest harness, and was streaming live to the internet as he followed the protesters and the police on a borrowed bicycle.

The protesters gathered outside the department’s East Precinct. An old man, homeless, busking below a mural, quickly collected his things before any violence could break out. But none did. According to the judge who would preside over the latest consent-decree hearing a few weeks later, notably saying the words “black lives matter” from the bench, the department was making steady strides toward an exit from federal oversight: Its de-escalation tactics were a national model, he said, and its use of force was substantially down. He never mentioned bodycams.

As the light drained from the sky, I watched a bicycle officer help Clemans pump up his tires and fix his brakes. Young black women and men took turns at the bullhorn, their voices loud, their demand — stop killing us — clear, and the officers hung back. The only people filming anything were the other activists. Fifteen months after its pilot program ended, Seattle has yet to agree on its answer to the simple question about bodycams: Whom do they serve?

McKenzie Funk is a founding member of the journalism cooperative Deca. He last wrote for the magazine about the wreck of a Shell oil rig in the Arctic.

Sign up for our newsletter to get the best of The New York Times Magazine delivered to your inbox every week.

A version of this article appears in print on  , Page 31 of the Sunday Magazine with the headline: Open City. Order Reprints | Today’s Paper | Subscribe

Advertisement

SKIP ADVERTISEMENT