ShotSpotter Sensors In Worcester Blanket Schools, Hospitals, Businesses: Leaked Data

WORCESTER, MA — The locations of ShotSpotter sensors used by Worcester police to detect gunfire have been made public for the first time, showing that the system blankets the city’s lowest income and often majority-minority neighborhoods, feeding criticism that ShotSpotter leads to a cycle of overpolicing in already distressed areas.

ShotSpotter sensors in Worcester can be found on top of public schools, hospitals and on college campuses, according to previously secret data published last month by Wired magazine. But the sensors are clustered only on the east side in neighborhoods from Burncoat south to Webster Square.

Worcester police leaders say ShotSpotter helps police make arrests, confiscate guns and respond to wounded people quickly. But the data leak comes as the effectiveness of ShotSpotter has been criticized, and as large users like Chicago have decided to get rid of the technology. Cities including Atlanta and Portland, Ore., have also recently decided against using ShotSpotter, citing cost and the human resources it consumes.

A 2021 inquiry conducted by the Chicago Office of the Inspector General found that only about 2.1 percent of the 50,000 ShotSpotter responses over a 1-½ year period resulted in a documented outcome like an arrest, pat down or gun confiscation — a potential drain for officers on the street, the report said.

“OIG concluded from its analysis that [Chicago police] responses to ShotSpotter alerts rarely produce documented evidence of a gun-related crime, investigatory stop, or recovery of a firearm. Additionally, OIG identified evidence that the introduction of ShotSpotter technology in Chicago has changed the way some CPD members perceive and interact with individuals present in areas where ShotSpotter alerts are frequent,” the report said.

Worcester police began using ShotSpotter in 2014, spreading gunshot-detection microphones across six square miles roughly between Bell Hill and Main South (Worcester police do not place sensors, ShotSpotter employees do). The department in 2021 expanded its use of the product to eight square miles and signed up for the company’s new ShotSpotter Connect (now called ResourceRouter) predictive-policing tool. The 2021 expansion added sensors north of Green Hill Park along Lincoln Street and south of Clark Street in Burncoat.

The gunshot sensors work by triangulating sounds that may be gunfire. When a sensor hears a loud sound, an algorithm figures out where it originated from. Meanwhile, a human at ShotSpotter’s Incident Review Center in Silicon Valley listens to the audio to verify if the sound is gunfire or something else like fireworks (or a science experiment, as one Worcester resident reported in 2021). The process happens in about a minute, according to the company.

Worcester police Chief Paul Saucier said there are no plans to change the use of ShotSpotter in Worcester. He described it as a powerful tool local police use to combat gun violence. In 2023 alone, he said, ShotSpotter alerted police to 41 gunfire incidents that weren’t reported through 911. That's about 35 percent of all shooting incidents in 2023.

“You’re getting a pinpoint location, it’s precision policing,” he said last week.

100+ ShotSpotter Sensors

You might walk by a ShotSpotter sensor in Worcester every day and not know it. There might even be one on top of your workplace or the roof of your child’s school. From the street, the sensors look like small plastic boxes, and sometimes blend in with the electrical and cellular equipment jumbled on rooftops and utility poles.

Many of the sensors in Worcester are located on street lights, simply because they are tall and owned by the city. But the sensors are also located on Clark University and Worcester Polytechnic Institute buildings, on almost every public school east of Park Avenue and UMass Memorial medical facilities.

None of the institutions Worcester Patch reached out to would talk about why the sensors are on their buildings. Worcester Public Schools confirmed that the sensors were placed on district buildings in 2014 with permission. A district spokesperson declined to comment further.

SoundThinking, the recently rebranded company that sells ShotSpotter, preemptively sent a statement to Worcester Patch about the data leak that alluded to a volunteer program for hosting sensors. The statement highlighted that the company is pursuing legal action against the people who leaked the information.

“In order to protect community members — including the residential and business owners who have volunteered to have sensors placed on their buildings — SoundThinking does not disclose the exact locations of its technology. If made public, this information can put these individuals at risk of harm and property damage, and the sensors at risk of tampering,” the company said.

SoundThinking declined to answer follow-up questions about whether the sensor placements are governed by lease agreements, if the hosts receive compensation or if local police help find volunteers. The company says it picks locations for sensors based on “historical gunfire and homicide data” from local police. The technology uses factors like how fast sound travels to a sensor and its “angle of attack” to give police location information, which is why sensors are often located high up.

There have been instances where police have used recordings of conversations made by ShotSpotter as evidence in criminal cases. A 2019 review of how the sensors record sound by the Policing Project at the NYU School of Law said the chance of people being surveilled by ShotSpotter sensors is “extremely low.” The report did acknowledge that sounds other than gunshots can trigger the system to start recording audio, but that the microphones are weak, similar to ones found on smartphones.

“Once operational, these sensors are continuously ‘listening’ and a proprietary AI-enhanced algorithm is constantly analyzing incoming audio. The algorithm reviews the audio for loud ‘impulsive’ sounds — that is, loud sounds that start and end suddenly (similar to a gunshot). In addition to actual gunfire, impulsive sounds that trigger the algorithm can include certain construction noises, helicopters, motorcycles, fireworks, and other similar sounds,” the NYU report said.

The density of sensors in Worcester varies by neighborhood, but Great Brook Valley, Bell Hill and Brittan Square have some of the highest amounts per square-mile. There are just over 30 sensors in the neighborhoods north of I-290 and Green Hill Park, and almost one-third are inside the Great Brook Valley Gardens apartment complex.

ShotSpotter Demographics

While the exact locations of ShotSpotter sensors have been secret until now, Worcester has released information showing generally where the sensors are located.

In 2022, former city manager Ed Augustus Jr. sent city councilors a report documenting five years of ShotSpotter data, including a map with a gray line around neighborhoods like Main South, Vernon Hill, Bell Hill, Great Brook Valley and upper Lincoln Street. The maps showed about 1,500 red dots inside the gray line marking the locations of gunshot alerts between January 2017 and April 2022. That map matches the one published by Wired showing the precise location of the sensors in Worcester.

Comparing sensor locations to Census data from the 2022 American Community Survey shows that more than half are located in Census tracts with poverty rates above the Worcester average of 19.3 percent. More than 66 percent of sensors are in tracts where the median household income is below Worcester’s 2022 median of $63,011.

About 35 percent of the sensors are located in Census tracts with a majority Hispanic population. The rest are in majority white areas, but ones that are more diverse than other parts of the city. One tract along lower Lincoln Street with eight sensors is 47 percent white, 35 percent Hispanic and 18 percent Black. By comparison, the Tatnuck area is 68 percent white, 7 percent Hispanic and 17 percent Black, according to Census data — and Tatnuck is about as far away from ShotSpotter sensors as you can get in Worcester.

Except for the expansion up Lincoln Street, the sensors have been in the same place since 2014, Saucier said. The department has not, and does not plan to, ask ShotSpotter to move the sensors around. Shootings in Worcester tend to happen in densely populated areas, he said, not the less populous single-family neighborhoods on the west side of the city.

“We’re not going to put [ShotSpotter] on the Holden line,” Saucier said. “We have a data meeting every week, we see where the shootings are. Right now, it’s pretty consistent.”

SoundThinking says its system isn’t racially biased because sensor placement follows “objective criminal gunfire” data provided by local police. Critics of the technology say crime data can itself be biased because it comes from years of police focusing enforcement on low-income, majority-minority neighborhoods.

ShotSpotter may also work to keep those patterns in place. An analysis by the MacArthur Justice Center found that police in Chicago were using ShotSpotter false alarms to make patrol decisions, leading to a higher police presence in areas where gunfire wasn’t necessarily elevated.

Saucier said no community in Worcester wants gun violence, and so the sensors should be a welcome sight. He wants people to see officers responding when gunshots ring out. Ditto for when the sound isn’t a gunshot, because then residents know police are taking every alert seriously.

ShotSpotter Here To Stay

After 10 years in Worcester, ShotSpotter hasn’t ended gun violence. A 17-year-old was shot and killed in the Union Hill neighborhood in February on a street between several ShotSpotter sensors, the city’s first homicide in 2024. On Monday, an 18-year-old was shot and wounded along Main Street, again close to gunshot sensors, and within sight of police headquarters. Chasity Nuñez and her daughter Zella, 11, were shot and killed Tuesday along Lisbon Street, near a gunshot sensor, but outside the immediate ShotSpotter zone.

But SoundThinking makes clear that its products aren't a cure gun violence, more like a prescription that manages a disease over a lifetime. SoundThinking emphasizes three main benefits of gunshot detection that only help police manage and react to gunfire: discovering unreported gunfire, reaching gunshot victims sooner, easier evidence collection.

“Gunshot detection by itself is not a panacea for gun violence, but if used as part of a comprehensive gun crime response strategy, it can contribute to a reduction in response times, help to save the lives of [gunshot victims], and improve evidence collection rates,” the company says on its website in a FAQ under the question, "How effective is ShotSpotter?"

ShotSpotter is just one in an array of SoundThinking products that offer police departments assistance with tasks like data analysis, investigation management and resource deployment (that's the ShotSpotter Connect tool that Worcester signed up for in 2021). The publicly traded company touted record profits in 2023 of $92.7 million, 14 percent higher than 2022. That success partially relied on expansions in existing cities, and adding new cities to its roster, according to financial filings.

According to estimates in 2014, the city paid about $860,000 for the first three years of the system, with about half of that paid for by CSX as part of a compensation agreement when the railroad company expanded its Franklin Street shipping terminal. Worcester took over full payments beginning in July 2016.

"With the end of CSX funding, this budget funds the city’s ShotSpotter gunshot detection program, which has made our police department more efficient and our neighborhoods safer," the fiscal year 2017 city budget book said. There was no line item for the total cost of the system that year, or in any year since.

When Worcester added new territory in 2021, the cost increased by about $192,950 per year, which includes the predictive policing tool ResourceRouter, then known as Connect.

Saucier has been a proponent of ShotSpotter since day one, and still is. He says that the three benefits SoundThinking touts — quicker response times, easier evidence collection and finding unreported shootings — are real, and help Worcester police manage gun violence. Worcester has relatively few gun homicides (five out of a total of six in 2023, but the number fluctuates each year) compared to other big New England cities like Springfield, Hartford and Boston — all cities that use ShotSpotter.

It's possible ShotSpotter could help police find and arrest people who are responsible for gun violence. Saucier said a small group of ten to 15 people are responsible for a majority of shootings in Worcester. The group expands and contracts with arrests and other events, but will essentially always exist in some form, he said.

"It’s a tool, it’s not going to come in and eliminate all gun violence," he said.

Each city has to find its own solutions to gun violence, Saucier said. Chicago may be getting rid of ShotSpotter because it’s not the right tool for a large department in a city with 2,500 shootings per year. Worcester police investigate each verified shooting like a homicide, a commitment of resources many departments can’t make.

“I don’t care where the sensors are as long as they go off when they have to. Why would you not want to respond to gunfire?” he said.

Correction: A previous version of this story said sensors were installed on school buildings under former superintendent Maureen Binienda, but a spokesperson confirmed Wednesday the sensors were installed in 2014 when Melinda Boone was superintendent.