As San Diego nears the end of its $1 million four-year contract with ShotSpotter, city officials are reconsidering yet another piece of surveillance technology that hasn’t lived up to its promise. Over a hundred cities in the United States have partnered with ShotSpotter to integrate the company’s gunfire-detection sensors into their wider surveillance networks.
In November 2016, San Diego became one of them.
Now, the San Diego Police Department has to decide on whether it’s worth the cost. At least one officer who oversees the sergeant in charge of the ShotSpotter technology doesn’t think so. SDPD plans, in the meantime, to review all relevant crime statistics and work with elected officials to see if ShotSpotter is still in the community’s best interest.
The sensors were exclusively installed on streetlights and electric poles in four neighborhoods in southeastern San Diego — Valencia Park, Lincoln Park, Skyline and O’Farrell — deemed by SDPD at the time to have among the highest rates of gun crime in the county.
The sensors were pitched as an effective tool for law enforcement, but a Voice of San Diego review of available data on the accuracy of the sensors suggests that the technology is not living up to the company’s own selling points.
On its website, ShotSpotter claims that only 20 percent of gunfire incidents on average are called into 911 across the country. The company says its technology picks up 90 percent of all gunshots within one minute and can decipher the location of those shots within 25 meters. The stated goal of ShotSpotter technology is to allow police officers to respond to a gunfire incident without waiting for a call from a member of the community, thereby increasing the speed and efficiency of their response times.
ShotSpotter’s senior vice president of marketing Sam Klepper praised the company’s self-proclaimed ability to detect gunfire even further.
“We have a 97 percent accuracy rate, with a very small false-positive rate of 0.5 percent,” Klepper said, citing the company’s 2019 review of data from every city it works with. The data could not be independently verified. The 0.5 percent false positive rate, Klepper said, refers to any time there was an acoustic event picked up by a sensor that didn’t turn out to be a gunshot.
But that statistic doesn’t square with police data obtained by Voice of San Diego from a California Public Records Act request. The data covers all 584 times that ShotSpotter devices were “activated” — meaning an audio sensor picked up what it determined was a gunshot — since it was installed in late 2016, up until the date that the request was filed in February 2020.
In four years, police have made two arrests while responding to a ShotSpotter activation. Lt. Shawn Takeuchi, an SDPD spokesman, could only confirm that one of those arrests was directly related to the activation, but declined to give more information on both.
Seventy-two of the 584 ShotSpotter activations were considered “unfounded” by SDPD in its records over the past four years, meaning officers were either falsely alerted to a gunshot or they couldn’t find any evidence of it. Takeuchi said loud noises from a jackhammer, a nail gun or a standard hammer could mirror gunfire to a ShotSpotter sensor. The unfounded rate thus stands at 12.3 percent between the four neighborhoods, a whopping 25 times higher than the 0.5 percent false positive rate put forth by the company.
“Officers get there and they find out that there’s jackhammering going on or construction going on at that location,” Takeuchi said. “Then more than likely, it was an unfounded call — it wasn’t a proper activation.”
When shown SDPD’s data, Klepper said ShotSpotter’s number is based on a national average and the company relies on individual police departments to report any instances of false positives.
ShotSpotter also markets its technology as a way for officers to gather physical evidence, like bullets or shell casings, at potential crime scenes that they otherwise wouldn’t have collected. That way, SDPD’s crime lab can try to pull a DNA profile from the evidence and add it to its firearms database. How often this happens, however, is unclear. Both SDPD and ShotSpotter declined to release any specific information about evidence collection.
The data itself suggests that police often don’t find anything worth logging. Nearly 60 percent of the time, officers who were dispatched to a location from a ShotSpotter alert produced no reports, because there was no evidence to collect. Takeuchi said this could happen at sites where gunfire was accurately detected.
“Just because the calls are no report required,” Takeuchi said, “that doesn’t mean there was no gunfire. That just means that there was no information that the officer obtained, or evidence that they collected.”
Voice of San Diego asked University of Pennsylvania professor Oscar Gandy Jr., who specializes in the information sciences of policing and technology, to also review the data. He raised doubts about the community benefits of the sensors. He called into question the use of police resources on ShotSpotter, considering how often officers are dispatched to a potential crime scene and produce or collect nothing.
Cornelius Bowser, a gun violence prevention activist, said the gang members in southeastern San Diego whom he works with aren’t phased by the presence of the audio sensors.
“They’re not worried about no ShotSpotters,” Bowser Sr. said. “It don’t deter them, that’s for sure.” Community members are more concerned with surveillance cameras and getting their pictures taken, Bowser said, than they are with audio sensors.
So what effect does ShotSpotter have?
Jennifer Doleac, founder of the Justice Tech Lab and an assistant professor of Public Policy and Economics at the University of Virginia who has written about ShotSpotter, said it’s still unclear after many years of police departments using this technology.
“ShotSpotter has resisted attempts (by me and others) to do a rigorous evaluation of its impacts,” she wrote in an email. “They’ve clearly found that they can get cities to sign their contracts without such evidence.” She encouraged cities to begin demanding evidence of the technology’s accuracy and usefulness before signing or renewing contracts.
ShotSpotter’s Klepper said the company is happy to work with researchers and has been doing so. In March, the Union-Tribune reported that San Diego had seen a 17 percent reduction in gunfire incidents from 2018 to 2019. That data, however, was pulled from a study the company conducted internally and can’t be independently verified.
Despite the conflicting numbers and ambiguous outcomes, Gandy said it’s possible that ShotSpotter could provide some value to the community — if, for instance, the technology itself limits the interactions between officers and members of the public, unless it’s absolutely necessary. But recently, San Diego has had a reckoning with SDPD’s ongoing surveillance practices, with communities maintaining that the private collection of their data erodes public trust in law enforcement.
Takeuchi said the Shotspotter contract ends Nov. 21. Not everybody in SDPD, though, believes that the cost of renewal is worth it. In a heavily redacted email chain from December 2019, Lt. Jeffrey Peterson, who works in SDPD’s operational support unit and oversees equipment, said he would defer to one of the chiefs but offered his own opinion on ShotSpotter.
“I don’t support keeping it based on the cost,” he wrote.
Almost a year later, Peterson still feels the same way. He told VOSD nothing had changed his mind since he wrote that email.