ShotSpotter CEO on how gunshot-detection tech works, what he would tell Tacoma skeptics

TACOMA — A test run of gunshot-detection technology is coming to Tacoma with the goal of shortening police response times, improving the investigation of gun crimes and aiding in collecting physical evidence.

The Tacoma Police Department announced the pilot program last month to implement ShotSpotter thanks to an $800,000 grant from the Bureau of Justice Assistance. Police said it will allow them to analyze how effective the technology is before committing city funds to it.

There are a number of unknowns about the project, including where in the city the sensors would be installed, how long the program would last or whether police would share ShotSpotter data with the public. According to a department spokesperson, police are in the process of bringing ShotSpotter to Tacoma at the end of June.

The technology has been criticized for its cost and use of police officers’ time, and it frequently stokes debate over whether it leads to the over-policing of communities of color. In 2021, ShotSpotter audio landed a Chicago man in jail for nearly a year, according to the AP, until his case was dismissed after prosecutors said they had insufficient evidence.

Proponents of ShotSpotter, including police departments, prosecutor’s offices and some city neighborhoods, have described it as an effective tool for a fast police response, investigating shootings and for allocating officers.

Ralph Clark, the CEO of SoundThinking, the company that owns ShotSpotter, sat down with The News Tribune on Wednesday for an interview. It has been edited for length and clarity.

• Can you explain how the ShotSpotter technology works?

Clark: It’s a sensor-based technology. We’ll deploy sensors in these sensor arrays that are approximately 20, anywhere from 20 to 30 sensors per square mile. And these sensors are designed essentially to ignore anything but large impulsive noises: pops, booms and bangs. And when it detects a pop, boom, or bang, it’ll timestamp that, and then the sensors, working together almost like a beehive, will collaborate with one another to figure out the exact location or source of that pop boom or bang event.

Then we’ll go through a couple of steps where we’ll filter out those pops, booms and bangs that the machine is pretty confident aren’t gunshots, so we weed out that kind of noise. And then everything else gets presented into an incident review center, where we have trained acoustic experts that are listening to and looking at these events in real time. And then only publishing those fully vetted alerts that they are highly confident are gunfire.

• How do you ensure that a gunshot is recorded and not a car backfiring or a firework?

Clark: We’ve been doing this for a really long time. Contractually, we’re obligated to have at least a 90 percent detection rate. … we’re at 97 percent. And this has been independently audited by a firm called Edgeworth Analytics. There’s a lot of tools that we have available to do that. You know, so the first thing is the machine has gotten better and better over the years. We’ve detected things that are gunshots and things that are not gunshots, and of course, as everybody’s reading these days the capability of neural networks and machine learning and the like — we’re leveraging all of that to be able to knock down a lot of noise even before it gets presented into our human-led incident review center.

In the incident review center where we have the human-review process. We’ve trained, very carefully trained, our reviewers to be able to distinguish gunshots from other impulsive noises that aren’t gunshots.”

• How are ShotSpotter sensors deployed to give coverage to a neighborhood?

Clark: We’ll design the array, and in this particular example, we’re talking about three square miles. Let’s say we say 25 sensors on average per square mile, so that’s kind of 75 sensors. Then on a Google map we’ll kind of place the ideal sensor array and say, ‘OK, if we could get sensors here, here, here — we like them spaced out in a way where we can feel comfortable that anywhere a gun is fired in our coverage area, we’re looking to get, like six, seven, eight, nine sensors participating. We need three to triangulate. But we want to over-deploy sensors to make absolutely sure that we’ve got really good coverage.

We look for places where we can get relative height. So we like them up high on top of rooftops. We prefer to be higher versus at street level. Our sensors don’t require line of sight, because there’s nothing visual about it. It’s all acoustics. So the higher we are in the middle of a rooftop, we’re perfectly happy around that.

Oftentimes, a customer will have public infrastructure that we’re able to get permission to deploy our sensors on. Sometimes we actually do go on light poles and utility poles. Our preference is typically not to do that.

• How can just detecting gunshots help address crime in a neighborhood?

Clark: It’s important for folks to understand that 80-90 percent of criminal gunfire goes unreported via traditional 911. So we close that gap by reporting on it precisely, within 30 to 45 seconds with a very exact location to send officers 100 percent of the time. So you kind of go from being basically blind to gunfire to seeing it in real time, with actionable alerts where we’re preserving the evidence, the digital evidence of that shot being fired.

And as a result of that, using it as a response tool, which is the first and primary application of our ShotSpotter technology, by dispatching officers and first responders to these events, very quickly, and precisely, to the extent that they’re encountering a victim, they can provide life-saving interventions.

The secondary benefit is sometimes you will be able to have enforcement interventions with the people involved in the shooting. Because, believe it or not, sometimes people do stick around long enough to be arrested. That doesn’t happen frequently. We don’t position this as an arrest machine. That’s not really what it’s about.

But then there’s a third benefit there too by getting to the scene precisely. That’s recovery of physical forensic evidence in the form of shell casings. And why this is so important is that when you can recover shell casings and run them through a system called NIBIN. It’s the National Integrated Ballistic Information Network, NIBIN, and it’s essentially, let’s call it a fingerprint for guns.

So when a gun is fired that isn’t a revolver, it ejects a casing, and that casing will have a distinct marking on it that identifies the fingerprint, if you will, of the specific gun that was used. And what very, I would say, forward-leaning agencies are doing is they’re investigating all gun crimes, even if it doesn’t result in a victim, a blood victim. Because what they know is if you’re investigating a shooting today, it might prevent a homicide or shooting tomorrow.

• What do you think having police in a neighborhood more often can mean for the relationship between officers and the community?

Clark: Without a doubt it improves it significantly. You know, let’s use Seattle as an example. Like if someone were to fire a gun in Madison Park, there probably will be some calls to 911. There will definitely be a response to those calls. Because the presumption is people shouldn’t shoot guns in Madison Park. You go down the road not very far, because that’s where I live, I live in the Central District of Seattle. Guns are fired. Chances are people aren’t going to call 911, which means there’s no response. So my question is, the residents of the Central District deserve, absolutely deserve the same care and response from police as the folks in Madison Park. There should be no difference in response. Our technology is an equalizer to help police be able to provide a level of care and duty to these communities that are suffering from ongoing, chronic gunfire.

In the process of doing so, when police are showing up in that way as guardians, ‘We’re here to protect you. We’re treating gun violence as a priority.’ Versus the narrative that happens before where guns are fired, people don’t call because people have just been resigned to the fact that nothing’s going to happen because they don’t trust the police and then police don’t show. Then the narrative becomes, ‘Well they don’t care.’

• What do you tell people who say this technology doesn’t reduce crime?

Clark: It’s not designed to reduce crime. In fact, there’s no single modality that I’m aware of that can reduce violent crime. And I think what it really does take is a collection of tools and a real commitment and process from not only law enforcement, but, frankly, also communities that have a role to play in overall prevention and reduction of violent crime. There’s no panacea to that. And we’re not trying to position our single-tool technology as being the panacea for that. So that’s my first response is that they’re measuring the wrong thing, and a very unrealistic thing. What they should be measuring and looking at is OK, is this a tool that can help bridge the public safety gap that exists when guns are fired, and there’s no call for service and there’s no dispatch? That’s an untenable situation.

One criticism of ShotSpotter has been that while it generates gunfire reports, officers often aren’t able to connect them to crimes. How do you respond to those who say that shows ShotSpotter is wasting police resources?

Clark: I fundamentally disagree with that, and I don’t challenge the math necessarily. So what people are measuring is, ‘Hey look, 5 percent of the time you make an arrest out of 100 ShotSpotter alerts.’ Then I’d say, wow, 5 percent sounds pretty interesting to me. It might not sound high as a percent, but the nominal number can be pretty, pretty big, right? Or you’re recovering physical evidence of this or saving lives. So I think it’s a mistake to kind of look at those percentages and not look at the bigger picture.

It’s about the nominal number and the impact, the nominal number of lives saved. And in those situations where you’re actually able to have an enforcement intervention, the nominal number of serial shooters you can take off the street. Because again, we know that it doesn’t take a big number of shooters to drive violent crime.

• What can ShotSpotter do that is not addressed by community groups trying to curb violent crime?

Clark: I think a lot of critics — uninformed critics — will try to make it an either-or question. It’s not an either-or question. It’s an and-also question. I mean, we want to see those violence interruption capabilities combined with having a response tool for police all working together. Because the other way, frankly, our data can be used, to kind of take it out of the policing response mode, is you have a really good picture of where gun violence is occurring very precisely. And that data can be used to inform certain social-services interventions as well. Trauma care resources, economic development resources. I mean, very, very precise to let you more efficiently apply those resources to kind of get at more root cause issues with respect to gun violence.”

• What factors go into the pricing of ShotSpotter for a city?

Clark: We price on a subscription basis. So we charge an annual fee per square mile. Right now, our MSRP is at $70,000 per square mile per year. We do some special things for smaller cities, and sometimes we’ll do trials.

• Is incident-level ShotSpotter data public record?

Clark: No. I would say the data that any subscribing agency has, and then how they choose to share that with the community, that’s a call for them to make. We don’t make that data generally available. It would be up to the agency to do that. But I will say we encourage our subscribing agencies to drive to data transparency and be very open with the communities about where they’re detecting gunfire.

Police Chief Avery Moore has said that ShotSpotter won’t target communities of color because it only detects gunfire. ShotSpotter would ideally bring police to gunfire incidents more often. What do you say to people that say that can lead to over-policing?

Clark: No, that’s helping police show up as public servants, just the way they do in Madison Park, by the way. We want that. No one wants to see over-policing. Like, I don’t want to see you stopping and frisking me. I don’t want to see you being super aggressive because my tabs are out of date by two days or whatever. Those are examples of, you know, people may be feeling like over-policing …

But I can tell you this. A community says, ‘Hey, someone fires a gun in my neighborhood, I want police to show up.’