As the nation begins to look seriously at ways to gain transparency into the actions of its police and hold them accountable, officer-worn video (or “bodycams”) have become really tempting. After all, we’ve all seen in countless videos on YouTube the raw power of being the fly on the wall. All of us already have video cameras today—on our phones. These cameras can be applied in a truly revolutionary way to help answer lots of questions. And it’s technology we all understand, right?
Or is it?
I’m a police officer, and I personally support the use of bodycams. In 2010 I bought my first body camera, with my own money. It was almost $800. As a technologist (I also run a start-up technology company) I understand that making technology operational is not as easy at scale as it is for an individual or a small group. Americans, and the American president, are now talking about making officer-worn video mandatory for all cops in America.
Since 2008, I have consulted police agencies large and small, written for police magazines and websites, and then worked as an officer and an investigator. I can therefore state unequivocally that barriers to technology adoption by police and courts are about 30% political, 30% cultural, and 30% procedural. I’ll neatly sidestep this 90%, then, because I’m just not good at it. Frankly, it’s above my pay-grade.
So I’ll leave to unions and cities the discussions of whether police departments whose officers are under contracts arrived at by collective bargaining would feel that this is a fundamental change in the terms of those contracts. I’ll defer to academics and researchers regarding the effects these cameras will have on the behavior of officers and the public. And I’ll leave the discussions about when the cameras should be activated and the privacy implications around that in the excellent hands of the communities, their leaders, and advocacy groups like the American Civil Liberties Union, and the Electronic Frontier Foundation (I am a member of both), which have already been thinking about this for a long time.
But when it comes to technology adoption by law enforcement—the hows, the details —well, now I’m qualified to talk. I’ve been in the tech business for almost two decades and my company, StreetCred Software, was funded by one of the technology industry’s greatest stars, accelerated by a national civic coding organization, and it sells technology (not related to video) to law enforcement agencies and courts. As I say, we don’t sell video, so the only skin I have in this game is that I’m a stakeholder, husband and father.
The barriers to adoption of body-cams from a technological standpoint are actually quite fierce. I don’t suggest for a minute that they can’t be overcome. They absolutely can and, in my opinion, must be overcome. I do aver that these details require informed public discussion and public-private cooperation.
These barriers to technology adoption fall into some broad categories: standards and guidance; bandwidth and access control; integrity,storage and retention policies; and funding. This article is long, but not by any means exhaustive. I’m just raising some fundamental challenges to adoption that must be addressed.
Some Raw Numbers
Before I begin, let me just set the stage with some raw numbers. Debates and media coverage often begin and focus on large cities. But it’s deployment to America’s small towns and cities that presents many of the operational challenges. America has about 18,000 police agencies, of which 85% have 24 or fewer officers—and about 50% have 10 officers or fewer. No one likes property taxes, so police budgets are perennially tight.
I’d roughly estimate that, in America outside the major cities, there are on average two cops per 12-hour shift, two shifts a day. Let’s say they only turn on the video during actual interaction with the public. So let’s figure on about 16 hours of video per day, per agency. Roughly 5,800 hours of video per year that will need to be stored, encrypted, and managed so as to be, for example, made available for trial or by public records request.
That’s 89 million hours of video a year, nationally. Bodycam video formats are not cat videos. Like our phones, a lot of it is HD now. But just for calculation purposes, let’s take the YouTube 320p standard for video (about 250MB per hour of video). That conservative estimate means that, in a single year, American law enforcement would create 22.2 billion megabytes (that’s around 20.7 petabytes) of video.
Standards & Guidance
Not only are there currently no national legal standards, or real national guidance, on any aspect of this, but there are also competing commercial video systems out there that lock video into proprietary formats. Just because the Department of Justice says everyone should go buy cameras doesn’t mean those legal or procedural standards will be forthcoming, so this will become essentially an unfunded mandate that every agency in America buy . . . something?
The bodycam market is currently nascent and fragmented, and the vendors each have proprietary formats. For some, agencies have to pay vendors each month for storage, or … well, actually, it’s not at all clear what happens if they don’t keep paying to store the video. Vendor lock-in is a huge concern. Changing from one vendor to another is not something that most agencies even know enough to ask about.
Bandwidth & Access Control
That 20.7 Petabytes of video can’t be stored locally at the agencies. Most U.S. police agencies don’t have full time IT personnel, or if they do they are shared with the city or county, so they are stretched thin. So even if they had the disk space—which they decidedly do not—they couldn’t manage and deal with issues like retention, key management and destruction.
But before you say, “Amazon S3,” or “Dropbox,” note that this video can’t just be parked on the public cloud for several reasons. First, the cops actually can’t get it there: most agencies have very slow Internet connections—think T1 or ADSL—and the “A” in “ADSL” means that the upload speeds are pathetic. Unless they saturated their outbound Internet link 24 hours a day (and sometimes, even if they did), they could never upload the volume of video they’d be generating without fatter pipes. I talk about funding below.
Now, you might think we could reduce the amount of video we upload, but no … Agencies don’t have a way currently to triage “worthless” video, and you never know what will become valuable later. We have to keep it all. Imagine the reaction to an incident in which the police said, “Yeah, they had the video, but, whoops!, they destroyed it accidentally”? Outrage and conspiracy theories would inevitably, and in some jurisdictions justifiably, follow from a skeptical citizenry.
And they don’t just have to keep the video: they also have to index it and make it at least basically searchable. That’s a problem with which the U.S. military continues to struggle mightily, having tried everything from reality TV producers to make highlight reels, to creating metadata for analysis, and they still haven’t solved it*. I’ve tried to locate video in my small department and it takes a very long time to find the right video.
There are, of course, many things private companies can do to help here (that vendor I mentioned that provides the storage? It does some great stuff, and makes a great product), but that data lock-in question, among many others related to data governance and compliance, is raised when the police depend entirely on private industry to provide this public service. Not saying it can’t be done, just saying it needs to be discussed, and in a different context from current contracts.
Which brings us to integrity. Perhaps the toughest nut to crack is going to be around keeping the video secure and ensuring that (a) unauthorized people can’t view it; (b) when authorized people view it the privacy of the innocent is protected**; (c) that the process of distribution (in response to Freedom of Information Laws) is largely automated so as not to burden future generations with full-time video-librarian tasks; and (d) that the public has confidence that the video has not been altered, to the extent that it can be used as evidence in a trial, with full chain-of-custody data and evidence of tampering.
This requires strong encryption, and that requires key management.
Commercial enterprises in the U.S. have been working on key management for decades, and it’s still not solved. Encryption keys are very powerful things, and there’s a festival of questions that need to be answered, starting with things like: What happens when you need to recover data 20 years later, and you can’t unencrypt the video because Ted from IT is gone, and the chief died in 2023, and no one can remember the password? Does the recovery process unlock the entire repository or just the video in question? Who controls it, and what happens to chain of custody?
Etc., etc., etc.
Make no mistake, key management issues are going to be massive, because stored video will have to kept for a long, long time—and here’s where I disagree, respectfully, with the ACLU.
Storage & Retention
To put the “How long should law enforcement agencies store video?” question into the right frame of reference, I’ll rhetorically ask you this: In a murder or rape case, when should we throw away the DNA evidence?
ACLU’s suggestion on retention policy is that, for a number of thoughtful and laudable reasons, police should measure retention period in weeks, not years. They’re doing this to protect privacy of all, including cops, hich is fine in theory but it means that many acts of malfeasance by officers or videos that would demonstrate that an officer was not guilty of behavior of which they are accused would be deleted.
As we have seen over and over again, the CSI effect on juries informs my prediction that, three years from now, if video is unavailable, the police will be disbelieved on principle. This is just as bad as the current situation.
ACLU’s retention suggestion won’t work for this simple reason: with retention periods measured in weeks, but statutes of limitations measured in years, you will have an intractable problem in no time. Retention will be unlimited, because, as we have seen happen again and again when DNA is used to exonerate people, sometimes decades pass before evidence is needed for the absolutely critical purpose of determining, or proving, innocence.
When the president of the United States asks for $75 million to kick off an initiative, that’s a tremendous opportunity—and if I were in the police video business, I’d be salivating. But that amount pales against the hundreds of millions of dollars it will ultimately cost to get this rolling in a sustainable fashion.
I reckon it would cost $8million just for the cameras for NYPD patrol officers, and that leaves out the necessary infrastructure, integration, bandwidth considerations, training, procedures, policy, storage, handling, FOIA and other related costs to the program. That’s NYPD. When you look at the agency where I worked for three-and-a-half years before my current assignment and see that their entire budget is under $2 million annually, you begin to see some challenges at getting the cash together.
How Can I Help?
If I’ve made this all sound hopeless, I sure didn’t mean to. I support anything that brings transparency to policing and serves justice. I support anything that helps rebuild trust of the police and the criminal justice system to the communities they serve. I happen to personally support bodycams on cops.
I just know that it is not as easy as passing legislation and throwing money at it. It is really hard.
The City of Seattle Police Department recently posted notice of a Hackathon, inviting civic-minded experts to help the police department find ways of automating the process of anonymizing subjects in videos (to protect their privacy) while maintaining the format of the original and the clear view of those central to the action. That’s not perfect, but I was encouraged by the outreach to the community.
Ultimately, we are all stakeholders in this, because rebuilding trust between the police and the community, and fostering transparency is what the video programs are all about. Civic coders, investors in government technology, forward-looking cities and civic-minded individuals and groups all have a role to play.
We all need to talk about this hard stuff, and work through the challenges.
* The military is in possession of tens of millions of hours of aerial surveillance video and they have had one hell of a time trying to figure out even basic questions, like, “What are some of the predictive aspects and elements we can look at to determine statistically when something ‘interesting’ is going to happen?” Big data is easy. Big video-data is wicked hard.
**Police video often includes, inadvertently or as an integral part, the most personal information of people including addresses, license numbers, age, weight, date of birth, criminal histories, and other protected data that may not, under laws governing access to criminal justice information, and under state privacy laws, be transmitted over the public Internet in unencrypted form.