The Citizen’s Dilemma

A new app wants to give you real-time information on crime all around you. But what does it ask in return?

San Francisco magazine

December 20, 2017

A pantsless woman has just been arrested in Civic Center. In North Beach, a purse has been snatched and a homeless man is making threats inside a church. At the McDonald’s on Potrero Avenue, a man is throwing unspecified items at employees. At the McDonald’s on Market Street, a man is screaming, and at the McDonald’s on Sutter, there’s been a theft.

I’m watching all of this unfold in real time on Citizen, the 10-month-old app that collects reports about ongoing crimes in San Francisco and sends them to your smartphone. Frankly, I’m not entirely sure what to do with this information, except maybe avoid McDonald’s. But Andrew Frame, Citizen’s CEO, sees in this mess of petty criminality the potential to save lives.

What Snapchat is to goofing off with camera filters, Citizen is to live-streaming crime. Where once the most hypervigilant—or paranoid—among us might have listened to a police scanner or obsessed over the daily blotter, Citizen offers a way to report on, watch, and chat about crime as it’s happening. “Had everyone in Las Vegas had Citizen,” Frame tells me, referring to the October mass shooting there that claimed 59 lives, “they would have had complete situational awareness.”

Citizen works by pushing geo-targeted information to users about suspected criminal activity in their area; nearby users can communicate with one another and add to those reports by uploading video they’ve taken of the crime scene. By stitching together this web of related information, Frame explains, Citizen offers the potential to paint a picture of a fluid, and often hectic, situation almost instantaneously. Where other public safety apps, like Nixle and RAIDS, work as one-way communication channels that allow local officials to broadcast information or post evacuation orders, Citizen—which currently operates in New York and San Francisco—is unaffiliated with local police departments, relying instead on its users to fill in the blanks.

And that, Frame believes, can be the difference between life and death—or, at a minimum, the ticket to avoiding trouble. Take, for instance, the November mass shooting in Rancho Tehama. As the suspect fled from police, he attacked innocent people at seven separate, seemingly random locations. Had those in his path known where the suspect was and where he was headed, they might have been able to steer clear. At least, that’s the theory.

In September, the first week that Citizen was in operation in San Francisco, the program was put to the test: 2,800 people watched via the app as a hostage drama unfolded in a Nob Hill apartment building. From across the street, one user uploaded video from his smartphone while others provided updates and chatted about what was going on. After three hours, the suspect was shot and killed by police. No bystanders were harmed.

The app’s potential has attracted investors. At TechCrunch Disrupt SF, in September, the crime tracker’s parent company, Sp0n, announced that it had raised a $12 million Series A round led by Sequoia Capital, with investors including former NAACP president Benjamin Jealous, Founders Fund, and Kapor Capital. New Age guru Deepak Chopra recorded a video promoting the app when it launched its first version in 2016, asking rhetorically, “Is crime still possible if there’s total transparency?”


There's a fine line between offering, as Frame calls it, “complete situational awareness” and actually fighting crime. Originally, the app was more closely positioned to do the latter. When it first launched in November 2016, Frame called his service Vigilante; it reported suspected crimes to the police and encouraged users to take crime fighting into their own hands by banding together and trading information.

Backlash to the app was swift and severe. “It’s quite hard to tell whether the Vigilante app is a functioning business or a teaser for a new episode of dystopian sci-fi series Black Mirror,” the Guardian wrote. Apple pulled the app from its store after two days, citing guidelines that ban apps that encourage or risk physical harm. Frame now takes pains to distance himself from the earlier iteration. “We’re so far past that,” he says, declining to go into specifics.

In March, Frame made a classic Silicon Valley pivot and renamed the product. Rather than serve as a tool for solving crimes, the new app would be an information clearinghouse. Citizen creates a virtual crime report that other users can add to with video or real-time chat updates. If that all sounds a lot like the plot of Jeremy Piven’s mercifully canceled Silicon Valley crime-fighting procedural Wisdom of the Crowd, well, yeah— but the reality of the app’s machinations is actually a lot more low-tech than that. Citizen says on its website that it pulls data on crimes directly from 911 calls; however, that information is not generally available to the public, and spokespeople from the SFPD say the department isn’t letting the app tap into its dispatch system. “From what we understand, they have someone in a room listening to a scanner and then sending out information on the call for service,” SFPD spokesperson Michael Andraychak says. (Citizen declined to elaborate.)

What’s more, Citizen’s relationship with police is even more tenuous than one might imagine: Andraychak says the app’s team presented Director Susan Merritt and Commander David Lazar with a pitch deck in September. “The meeting was no longer than 30 minutes,” Andraychak says.

That’s not the only way in which Citizen’s model fails to engender trust. There’s also the matter of policing its users, a problem that other platforms dealing with public safety, like Nextdoor, have long grappled with. Put bluntly, people get racist. Matt Cagle, a technology and civil liberties policy attorney with the ACLU of Northern California, describes the problem more diplomatically. “Encouraging people to hastily report on their community members encourages bias,” he says. Citizen’s FAQ states that its “operations team removes any non-pertinent descriptions of suspects, when those details are irrelevant to the incident.” Adds a spokesperson, “We continue to analyze and improve our incident descriptions and criteria, including how we communicate suspect information. We only enter descriptions of confirmed suspects related to any incident.”

Cagle also worries about Citizen’s potential effect at, say, a political protest. Would unverified crime reports broadcast over the app create “complete situational awareness,” or simply bring on a panic? Would having an army of Little Brothers filming every minor act of civil disobedience discourage citizens from exercising their First Amendment rights? When a writer at the Washington Post tried the app out in September, he received an incident report for “Electrical Union Protesters Gathering for Rally in Foley Square.” That’s not a crime; that’s democracy. And if Citizen can’t tell the difference, that’s trouble.

The litany of problematic hypothetical questions surrounding the app doesn’t end there. Would Citizen turn over its users’ data to police under subpoena? “Users deserve to know that this app will not become a source for police fishing expeditions,” Cagle says. It’s easy to multiply the potential problems: Could video uploaded by private citizens be entered into evidence in a criminal case? Could an innocent bystander be confused with a suspect? Or worse, be set up as one?

Practically speaking, though, those issues all pale in comparison to the biggest question facing Citizen: How does it plan to make money? Says Al Gidari, director of privacy at Stanford’s Center for Internet and Society, “I can’t quite figure out what the business model might be.” As always, the answer likely lies in the fine print: The app’s terms say that it retains the right to distribute user information for “any and all” purposes, though it doesn’t clarify what those might be. Perhaps Citizen could sell access to its data to a real estate firm that wanted to subtly redline neighborhoods, for example, or to a McDonald’s that wanted to advertise that it had gone so many days without a crime. The company’s leadership declined to comment further on its business model.

So what to make of this digital, crowdsourced, real-time crime map? Any port in a storm, they say; if you’re in the middle of a mass shooting and just want to know how to get out, those privacy and racial-bias concerns might easily fall away. I pose the question to Stanford’s Gidari: Would you use this app in a crisis? He doesn’t take long to answer in the negative.

“This is a poor substitute for street smarts,” he says. “When the police car is running in one direction, you should run in the other. Do you need to hear more than one gunshot to know you should get the hell out?”