IMPD recommended city contract with ShotSpotter after 2022 pilot
INDIANAPOLIS (MIRROR INDY) — Roughly 18 months after police officials labeled gunshot detection technology as not “fiscally responsible,” an Indianapolis Metropolitan Police Department working group reversed course and recommended purchasing ShotSpotter services, according to a 2023 report obtained by Mirror Indy.
In 2021, against a backdrop of protesters calling for the city to defund the police, IMPD Deputy Chief Kendale Adams voiced his opposition to a proposal that would have allocated $730,000 to spend on a gunshot detection system.
“One of the things this proposal talks about is ShotSpotter. And we’ve looked at that technology. It’s an expensive technology,” Adams said at a July 2021 meeting of the City-County Council. “It’s continuing to be technology we look at, but at this time, it just would not be fiscally responsible.”
But by the following February, police officials said they would test the use of gunshot detection technology on the east side. That pilot, which aimed to test the capabilities of ShotSpotter and two other vendors, concluded at the end of 2022. Police did not release the results of that study.
Mirror Indy obtained the March 2023 report and study results this month through a public records request.
Those records credit ShotSpotter with helping first responders save the life of one man, but also note that just 8.2% of the gunshot detections from two vendors resulted in the collection of any evidence. The results line up with what other cities have experienced, one criminologist told Mirror Indy: While the systems are effective at detecting sounds, they do not lead to an improvement in public safety or solving cases.
Lt. Shane Foley, an IMPD spokesperson, declined to answer questions about the results of the pilot, whether the city was considering a contract with ShotSpotter or how much it would cost. In 2021, officials estimated it would cost $250,000 to deploy the sensors in a 3-square-mile area, plus $200,000 in annual maintenance. It’s unclear how much those costs may have escalated by 2024.
“There will be additional updates provided in the near future,” Foley said in an email. “Until then, we do not have additional information to provide.”
According to a spokesperson for SoundThinking, Indianapolis decided not to move forward with the technology after the 2022 pilot. It’s unclear what conversations the city has had with the company since then. Last month, the company told Mirror Indy it was “participating in a formal proposal process” in Indianapolis and could not comment further.
Last month, leaked company data from a Wired Magazine investigation revealed for the first time the exact locations of ShotSpotter devices on the east side. Additional reporting by Mirror Indy showed that several of the devices are still in place, despite the fact that the city’s pilot ended more than a year ago.
Chris Daley, executive director of ACLU Indiana, said the lack of transparency regarding ShotSpotter in Indianapolis was troubling and that he was disturbed that the sensors are still up.
“Where there are so many civil liberties and public safety issues at hand,” he said, “there should be greater transparency.”
What did the study say?
Information from the 2022 pilot was reviewed by an IUPUI researcher and an IMPD working group that included the researcher and several IMPD officers, among others.
In its report, the group evaluated the technology’s performance in four “verified critical incidents” where at least one victim was shot within or close to the pilot area’s boundaries.
In one case, officers responded to a ShotSpotter notification on the 2700 block of North Michigan Street and were able to administer first aid to a gunshot victim. Police credited the sensors with helping them save the person’s life.
ShotSpotter also alerted police to a person shot on the 3300 block of East 11th Street. Although Indianapolis Fire Department personnel were already at the scene, an IMPD official said had a bystander not flagged down IFD staff, police would have arrived first because they did not receive a 911 call prior to the ShotSpotter alert.
In the two other cases highlighted by IMPD, including a homicide, police responded to 911 calls but later used ShotSpotter data as part of their investigations. The report doesn’t specify how, and Lt. Foley wouldn’t elaborate.
An analysis of data from ShotSpotter by computer science professor George Mohler also measured differences in response time and evidence collected after alerts from two companies’ sensors: ShotSpotter and Flock. Mohler, then at IUPUI, is now a professor at Boston College.
According to his analysis, ShotSpotter’s sensors alerted police to more gunshots than Flock’s and provided officers with more accurate locations.
Sensors from the third vendor, J&M Security Solutions, did not produce enough data to be included in the study, according to the report.
Police also were faster to respond to alerts from the two gunshot detection systems than calls to police, although Mohler noted that the latter included calls from throughout the entire city, not just the pilot area on the east side. Many factors could contribute to the difference, such as where patrol cars were positioned throughout the city, the study said.
One reason IMPD’s working group recommended ShotSpotter is the number of alerts it received. During the study period, ShotSpotter sent 514 alerts to police, only 49 of which could be matched to 911 calls, police said, adding that the system would help them respond to incidents they would otherwise not know about. (Other parts of the report show different numbers of detections matched with calls for service. Mohler did not immediately respond to questions about the differences.)
How many of those detections were false alarms is not clear. Mohler said that false positives were not evaluated as part of the study. While it’s possible that some were false positives, he said it was “a reasonable assumption” that many of the detections were gunshots that either had not been reported through 911 or had been called in with inaccurate information.
He also said the study was largely meant to compare the two vendors, not how the vendors performed compared to 911 calls.
ShotSpotter on its website says it has a false positive rate of less than 0.5%.
Eric Piza, a professor of criminology at Northeastern University, said studies of ShotSpotter in other cities have shown that the technology generally delivers on what the company promises: Police receive alerts to gunfire from ShotSpotter sensors before neighbors call in the shootings, and the notifications provide accurate and precise location information.
Indianapolis ShotSpotter sensor locations
A Wired Magazine analysis of leaked data revealed for the first time the exact locations of ShotSpotter microphone sensors in Indianapolis and across the globe.
“But what we found is that those procedural benefits didn’t translate to any meaningful public safety enhancements,” said Piza, who has studied law enforcement technology for nearly two decades.
The sensors in other cities did not lead to a decrease in the number of shootings or an increase in how many shootings were solved, he said.
After reviewing the city’s report, ACLU’s Daley said he was concerned by the large number of detections that could not be matched with 911 calls or for which no evidence was collected. Alerts from the sensors resulted in evidence collection only 8.2% of the time, at roughly the same rate as 911 calls.
“When ShotSpotter is telling you, ‘we have a very sophisticated algorithm that says gunfire happened,’ that is sending an officer into a situation at a heightened state of awareness that may put people in that community in danger,” he said.