How Police Body Cameras Work

Police body cameras are often "point-of-view cameras." They mount to a hat, helmet or pair of sunglasses, recording what an officer actually sees as he goes about his job. Others mount to the shoulder or collar of a uniform and record from a slightly lower perspective. Most are roughly the size of a Scotch tape dispenser and weigh anywhere from about 2 ounces (55 grams) to 5 ounces (140 grams) [source: NLECTC]. That's somewhere between a large strawberry and a lemon.

Beyond that, features vary by model. There are more than a dozen to choose from, made by companies like Taser, Vievu, Wolfcom and Panasonic. Extras include high-definition video, wireless operation, anti-tampering safeguards, one-touch activation, ultra-wide-angle views, auto-adjusting low-light modes and night vision [source: NLECTC]. (Night vision isn't all that useful in the accountability context, though, since what matters is what the officer could see, not what was there [source: Wolfcom].)

Most stamp footage with time and date, and some with badge number and GPS coordinates [sources: Wolfcom, NLECTC]. Some take stills. One streams live video to remote devices [source: Wolfcom]. Few models have on-device playback screens, but several stream footage to smartphones for immediate viewing [source: NLECTC].

One of the most impressive features, though, is battery life. Recording times can reach 12 hours or beyond. Twelve hours is a police shift [sources: Crisp, Wolfcom].

Most officers don't record their entire shifts, though. To save power and avoid wasting storage space on lunch breaks and driving, they typically start recording when they're about to interact with civilians. Lest this result in missing something significant, most models have buffering systems, or "pre-record." As long as the camera is turned on, it's recording, but only in 30-second (or 60-second) intervals and usually without audio [source: NLECTC]. After 30 seconds, it deletes that footage and records another 30 seconds, and so on, until the officer hits "record." At that point, audio activates and everything gets saved, including the previous 30 seconds of buffered footage. Pre-record footage can add valuable contextual information, or it could prove critical if an officer hits the "record" button late due to the urgency of the situation [source: Wells].

All of this functionality comes at a price. The cameras run from about $130 to $900 apiece, which may seem unimpressive considering the price of a civilian-issue smartphone. But if you are, say, the Oakland Police Department, and you're deploying 600 of them (and watching your use-of-force complaints drop by 75 percent over five years), it adds up [source: Brown].

Still, the cameras aren't the expensive part.

Oakland's 600 body cameras generate 7 terabytes of video footage every month [sources: Brown, Elinson and Frosch]. Storing that much sensitive data is a challenge, to put it mildly. Oakland has to keep the footage for at least two years, and any video involved in an investigation is stored for longer [source: Elinson and Frosch]. In Duluth, Minnesota, it's at least 30 days [source: Bakst and Foley]. Laurel, Maryland, has to store it for at least six months [source: Weiner].

The sheer amount of data is staggering, but data-security standards for law enforcement further complicate the situation. All footage is considered potential evidence in a criminal or disciplinary case, so the ability to view, delete or modify it in any way has to be strictly controlled, and all interactions with the video meticulously documented to maintain chain of custody [source: Ferrell].

Some departments set up their own secure server systems, which means big upfront costs in hardware and software. It also means training, transferring all footage manually and maintaining the database in-house [source: GCN].

The cloud is another option. Most commercial systems don't cut it security-wise, though, so several camera manufacturers offer cloud services that meet the specific needs of law enforcement [source: GCN]. In Taser's system, all data is encrypted and transferred wirelessly to the cloud when officers dock their cameras after their shifts [source: GCN]. The system automatically records all interactions with the data, and district attorneys can access the footage remotely. If someone who has deleting privileges initiates that process, the system automatically notifies all others with deleting privileges before discarding the data [source: Wells].

However you store and manage the footage — an estimated 1 terabyte per year per camera — it's costly [source: Emergency Management Law]. The Los Angeles Police Department expects to pay about $870,000 per year to maintain the footage from the 860 cameras it purchased in 2014 — and $7 million per year when it puts cameras on each of its 7,000 officers [sources: Police Magazine, Stoltze]. In 2014, Baltimore vetoed a body-camera program that came with projected storage costs of $2.6 million per year [source: Bakst and Foley].

Costs can potentially be offset, though, by a reduction in civilian complaints, faster investigations of complaints and, since video evidence can clearly exonerate officers of wrongdoing, reduced lawsuit payouts [source: Wells].

The U.S. government, for its part, is issuing grants to local police departments to assist with equipment and maintenance costs — tax dollars well-spent if the cameras really do increase safety, accountability and transparency in policing [source: Flores].

But to many, that outcome is uncertain. And one potential drawback of the technology has some wondering if it's worth the risk.

While intended to record the actions of police, recording the police means recording the civilians they interact with. Few people actually see the vast majority of the footage — police might view it while preparing their reports, and a clip that ends up being pertinent to an investigation might be widely viewed — but it's still there, and both the legalities and the long-term consequences of this new type of surveillance are unclear [source: Weiner].

How do two-party consent rules apply in the body-camera context? How will law-enforcement and government agencies use the countless terabytes of video down the road? Will members of the community decide not to approach officers with tips if they know they'll be recorded? Will domestic-violence victims be less likely to seek help?

Some officers can turn the cameras off in highly sensitive situations — hospital interviews, for instance, or car accidents with injuries, both of which can have medical-confidentiality implications in additional to personal-privacy concerns [sources: Weiner, Wells]. Some departments require that officers offer to stop recording when they enter a private residence [source: Weiner].

The privacy of the officers is at risk, too. Officers can forget to turn off their cameras in the bathroom [source: Wells]. Their private conversations with co-workers might be recorded [source: Stanley].

Or much worse. In 2014, an Arizona officer responding to a domestic-dispute call ended up recording his own murder. The officer's body camera was filming when a man he was questioning pulled out a gun and shot him. Under Arizona law, the police department was forced to release an edited version of the video to media outlets when they requested it [source: Kaste].

The American Civil Liberties Union (ACLU) advocates requiring permission from those filmed before a video can be released. It also sees redaction capabilities as essential, exemplified in the case of an Albuquerque officer who had his body camera running when he responded to a scene and found a child being strangled. Under New Mexico law, the police had to release the video when media outlets requested it [source: Weiner]. (It's actually rare for states' open-records laws to apply to this type of record, though, at least according to the Harvard Law Review.)

In the end, that the ACLU supports body-camera programs even conditionally speaks to the depth of dissatisfaction with the current state of policing. But whether body cameras will affect real change is uncertain.

Theoretically, a body camera can tell us what really happened. Condemn or exonerate an officer based on unbiased, visual data.

But there are several glitches, the first being that officers don't necessarily turn them on. The North Tonawanda police department in New York requires that officers wear body cameras, but police union contracts require that recording with them be optional [source: Vaughters]. During a six-month body-camera trial in Denver, officers captured only 25 percent of their use-of-force interactions [source: Major].

Another problem is that video footage isn't really unbiased. A body-camera recording doesn't show the scene; it shows the officer's view of the scene. And even then it may not show everything. A collar-mounted camera doesn't follow an officer's eyes as they move. Few of the models have wide-angle lenses that match the human visual field [sources: Wolchover, NLECTC].

Finally, video evidence is ultimately open to interpretation. Studies have shown that factors like gender, cultural background and even where one focuses one's eyes can affect how someone interprets video footage [sources: Kahan, Benares, Harvard Law Review]. In 2014, members of a New York grand jury saw cell-phone footage of an officer holding an unarmed, prone man in a chokehold, the man repeatedly gasping "I can't breathe," until he died. They ultimately decided not to indict the officer for a crime. Others very clearly saw a criminal act [sources: New York Times, ABC News].

A long history of similarly disparate interpretations brings us to this: The link between video evidence and police accountability is questionable.

But maybe that's not the point. The Rialto study and a lot of anecdotal evidence suggests that when people are on camera, they behave better. And it's not just the police. When civilians realize they're being recorded, they adjust their behavior, too [sources: Wells, Weiner]. And if it's true that with body cameras running, police are less likely to use force, and civilians are less likely to find fault in police behavior, maybe the best possible outcome of surveilling all police work is not for police to be held more accountable for unjust violence but for there to be less unjust violence to account for. Which actually might be doable.