LPRC - CrimeScience Episode 69 – Dr. Mares

Episode Date: October 5, 2021

He is a professor of Criminal Justice at Southern Illinois – Edwardsville. Dr. Mares received his PhD in Criminology and Criminal Justice from the University of Missouri-St. Louis and is currently a full professor at SIUE. His research focuses on spatial and temporal aspects of violent offending and has been published in leading interdisciplinary and criminological journals such as the Journal of Urban Health, Environment and Behavior, Climatic Change and the Journal of Experimental Criminology.  Dr. Mares works with law enforcement agencies as a Research Partner and assists on grant development and program evaluation of innovative policing approaches.  He is also a Subject Matter Expert for the Center of Naval Analyses’ (CNA) Smart Policing Initiative (SPI) where he helps Bureau of Justice Assistance (BJA) grantees develop and apply evidence-based practices in law enforcement. In this episode we discuss the use and assessment of various forms of police technology with a focus on acoustic gunshot detection in the community and the retail space. The post CrimeScience Episode 69 – Dr. Mares appeared first on Loss Prevention Research Council.

Hi, everyone, and welcome to Crime Science. In this podcast, we explore the science of crime and the practical application of this science for loss prevention and asset protection practitioners, as well as other professionals. We would like to thank Bosch for making this episode possible. Take advantage of the advanced video capabilities offered by Bosch to help reduce your shrink risk. Integrate video recordings with point-of-s sale data for visual verification of transactions and exception reporting. Use video analytics for immediate notification of important AP related events and leverage analytics metadata for fast forensic searches for evidence and to

improve merchandising and operations. Learn more about extending your video system beyond simple surveillance in zones one through four of LPRC's zones of influence by visiting Bosch online at boschsecurity.com. Welcome everybody to another episode of Crime Science the podcast and where we try and feature some of those leading criminologists and practitioners out there working hard to make a difference to better safeguard the vulnerable people and places. And so today what we're doing is we're joining Dr. Dennis Mayers. And Dennis is a professor at Southern Illinois University. And make sure, what's the E stand for? I apologize for that.

It's Edwardsville. We're not Carbondale, we're Edwardsville. And we always got the bad rap because we were supposed to be the smaller university, but we've in fact gotten bigger than Carbondale. Love it. Excellent. That's good. All right. So, right. We're kind of in that situation where the University of Florida was always the oldest, where the University of Florida was always the oldest, largest, whatever, best resourced, I guess. But now UCF kind of just, you know, has a massive enrollment compared to us. And by massive, I mean probably another 10,000 students, something like that.

But we have a kind of a conscious effort to not to cap that enrollment here. So excellent. So today what we're going to be talking about with Dennis, with Dr. Mayer, it really is around police technology. And our listeners know, Dennis, that at the LPRC, we work on use cases. Right. And so we try to be science or evidence based in what we do from the strategy that we put together with them to how we affect behavior, how we detect that behavior in the first place. And a lot of that involves technology of all types and including a lot of artificial intelligence these days. And we're blessed that AI is critical to us at the University of Florida, the Safer Places Lab team, as well as at the LPRC. So what I'd like to do is go over to you, Dr. Mayers, and talk a little bit about

police technology. And we do partner pretty tightly, pretty closely with police technology. And we do partner pretty tightly, pretty closely with law enforcement agencies at all the strata, municipal, county, state, and federal, and in a lot of different ways, but really not so much in staying current on the use of technology and their state of technology. So a little general, but what's the current state of police technology in your opinion? And what does that look like? Yeah, I think that's a question that's really kind of dependent on who you work with for the most part. What I see, and that's mostly working with municipal police departments, is that there's a lot of new technology coming in. So you can see that some

agencies are starting to use drones, some agencies are starting to use cell site simulators, facial recognition here and there. But that new technology, I think, is still very much in development and isn't contributing a heck of a lot at this point to the day-to-day bread and butter what police do. And I think where we see the biggest leaps really is in camera networks, license plate readers, and increasingly gunshot technology. I think those are sort of the things that really starting to drive the real-time crime centers and a lot of the municipal police departments. And that's what a lot of the information is. And what you see too is increasingly that departments are using various platforms to integrate that technology. So you have an integrated system that brings together all the data streams from cameras, from license

plate readers, gun detection technology, and other technologies as well. And so I think that increasing platforming of the technology is what I'm seeing at least in municipal departments. That's fantastic. And I really loved how you brought in some of the wide diversity, because when we talk about technology, we tend to think about sensors, we tend to think about action tools, or some that are both. But the idea of drones and license plate recognition and look up and then how pulling how agencies might be pulling this all together, particularly with these real time crime centers. So, good. We've got a little bit to talk about and including also in our kind of a focus today is on that gunshot detection system usage out there. detection system usage out there. So let me ask a little bit about how do agencies put together their overall strategy, including technologies, if you will. Again, if this is too broad, just please let me know. We can narrow it down. But it's a big question, a big part of what we're doing with these major retail companies that operate across, you know, vast areas with hundreds or

thousands of store locations or distribution centers or office environments, you know, even manufacturing plants, so, and their truck transport and so on. So it's now putting together an overall strategy to better protect the enterprise to safeguard everyone and everything to the best you can. And how do police agencies tend to, again, I imagine it's all over the place with whatever 16, 19,000 of those agencies, but how they put together their strategies. And I think, are you asking about the strategies for the integration or just the overall strategies of how to use the technologies and how to deploy it? Well, again, they may be on the cusp like we are,

where the retailers are re-looking at how they better protect their enterprises. What is the strategy? What are we trying to accomplish? And then, like you just mentioned, layering in technologies. What sensors, what action tools do we need technologically at these different points in the enterprise? How do they support us? How do they enable us or hopefully enhance us? And how do we not get lost in all that? Yeah, I think from the point of view, at least of the municipal police departments that I tend to

work with, the key thing is, is how do you solve incidents, right? And I think there's a lot of misperceptions, I think, in general about police technology in the sense that they just think that, oh, you know, it's just like magic. You have these cameras and now you have, you can see everything, right? But that's not how it works, obviously. There's all these things that go into

making technology work and not work. So for instance, there was a homicide in an agency that I worked with just this week, where a piece of technology, a mobile surveillance trailer was sitting right there and the homicide happened right next to it. Now the mobile surveillance trailers got cameras on them, but did they capture it? No, they did not because they were pointed in the wrong direction. And so I think for a lot of agencies, the biggest issue is to figure out is how do you get the most mileage out of that technology? Because police departments, by and large, are not as well funded as people like to think. And so it's always trying to make do with what you can get.

And I think that's the biggest challenge for implementing technology in law enforcement right now is how can we get squeezed the most out of these resources that we have. And I think that becomes a very important issue to like when you're thinking about okay so we have X amount of budgets for, let's say 50 cameras, and now we have to think about where we're, where are we going to put these cameras? Where are we going to get the most traction, the most mileage out of them? And then two, you run into simple logistical problems. So what a lot of folks think is that cameras often are put in high crime neighborhoods, and that's not always the case. In some communities, the infrastructure, so the network infrastructure is simply not there. So you would like to put a camera there, but you can't because you don't have the internet bandwidth to support the camera feeds. So it's all these kind of day-to-day problems that you think that a lot of agencies are facing and trying to sort of muddle through and figure out how can

you best address it. And that's a very diverse field if you look at it, and it varies a lot from place to place and how that gets resolved. That's a really interesting idea that you might have, something's indicated we need better sensors and understanding in a certain area, deterrence maybe even, but the infrastructure is not there to support it, whether it's bandwidth or even just connecting or just won't provide that flow and certainly not in resolution that you might need.

So very fascinating in that way. So let me talk a little bit about, if I could, you mentioned some of the tech. How do you measure effectiveness? What are the outcome measures, the dependent variables, if you will, that law enforcement agencies look at or might use? That's a good one. I think everybody assumes it's crime reductions, right? That's what we always want to see, crime reductions. And that's what a lot of agencies expect. But there are other ways in which effectiveness can be measured in a responsible way, I think. So when you're talking about gunshot detection technology, for instance, one of the good outcome measures could simply be, can we get there quicker?

Because getting there quicker might potentially save lives. You can get the medics there quicker. Getting there quicker may mean that you can make more arrests. So that can be an outcome measure too. And there are others as well. So what about evidence, right? Technology can provide a lot of evidence that we don't always have. So you can get visuals on people with cameras. You can get license plate information with LPRs. And with gunshot detection, you can get shell casings when you get there. Now, that increased evidence recovery that this technology can bring can also by itself be a dependent variable, I think. And then you have a more broader view of what we can get out of this technology as well.

Most agencies don't tend to look at these things as closely as I think academics would like, but as academics, we're very interested in how do offending patterns themselves, how can we get those from the technology. So with gunshot detection technology, for instance, you can quickly find out what the problem properties in a neighborhood are by taking a good chunk of data and overlaying that with satellite data. And you can pick out the exact houses from which a bunch of gunfire comes. And it's kind of a persistent thing, right? So you have one gunshot here, five gunshots the next day, six gunshots two days later, and so forth. And if you build up enough data, you can get a very clear pattern out of it that tells you sort of where these hot spots are. And that by itself, too, can lead to more actionable intervention by police departments and other agencies as well. You know, I really like, Dennis, how you're describing and thinking about it. You know, I really like, Dennis, how you're describing and thinking about it. And it is sometimes a struggle with what we work on and that people want the ultimate metric. And like you say, that's some sort of crime or in our it's just not a strong one-on-one correlation.

There's a lot of intervening and so forth variables in there. So I like that. us all think about issues and think about answers or solutions that we're trying to put in there and how we might dose them and not thinking that something fails either. You hear in the corporate world that we deal with so much fail fast and fail forward. And so, in other words, failure is a big part of innovation and of improvement in everything, right? And so, to me, you're describing that, that look, there are these other interim measures and metrics and things that we have to get right,

and it's going to take a little while to dial that in to ultimately get where we want, and that is enhanced safety and security for everyone. So maybe going over on effectiveness, you know, we've been involved in just over 30 randomized controlled trials. We don't necessarily have to talk about that, but are RCTs becoming more common or not even just RCTs, other strong quasi-experimental designs? I mean, is that starting to come more and more into play in what you all are looking at or the agencies are and their decision makers? Yeah, I think so. I mean, and that's partially driven by federal funding agencies like BJA, for instance, that really the programs they're putting out more and more require strong designs for evaluation. Otherwise, you just don't get the funding. And I think that's a very good development. You know, for a very long time in law enforcement, there was very little

oversight on the grant monies that were given to these departments. Here's some money, go buy some technology. And we will assume it works sort of was the sort of working hypothesis if you want, right. But in the last decade or so, there's been quite a bit of change in that that I've seen at least, where that accountability of how do you implement this correctly? How do you measure the outcomes correctly? What's the effectiveness of these things that we're doing, right? And I think that's a good thing for policing overall, because it leads to this sort of creation of a more systematic investigation into these police technologies. And so I think that that's a real

good thing. And I think one of the key programs in that is the Smart Policing Initiative that's run through BJA. And I think that's one of the best programs that exemplifies of how we should move forward with police technology and other types of interventions as well. So, yeah, we are seeing more RCTs. In some cases, that's quite feasible. I'm doing currently an RCT with St. Louis Police Department on mobile surveillance trailers. We really thought that out well where we had, we're essentially deploying mobile surveillance trailers. We got three of them. And then we also have a dummy surveillance trailer. So this is essentially our placebo where we put it out and it doesn't have any technology on it.

And we're going to see if that placebo produces sort of the same crime reductions we're hoping to see in the real thing. Obviously, we have controls as well, regular controls as well, with nothing happening in there. But the use of placebo is something you don't see very often in our field. And I think that was one of the very cool additions to that particular project. So yeah, I think that's overall a very good thing. Doesn't mean you always get to do the RCT, of course. Sometimes that is just very difficult. In a lot of cases, we get called in afterwards and people want to know, that sometimes that is just very difficult. In a lot of cases, we get called in afterwards and people want to know, hey, is this effective or not? So, and that's the case oftentimes for the acoustic gunshot detection systems. We often get called in afterwards and say, hey, can you see if this is effective or not? And then, of course,

it becomes much harder to get a really rigid research design together. And then you're often stuck with the quasi-experimental designs, which can be quite effective if you do them well, but, you know, that's just a little bit harder for the most part. Yeah, that's fantastic. And, you know, it's interesting you bring up mobile towers. We might want to confer a little bit, and I'm sure that McKinsey, a research scientist who's on here right now with us, but is probably thinking the same thing, but we've got some of those trailers from a company, LiveView, that we have three at our outdoor labs here loaded with sensors. And

then we work with Walmart, Kroger Company and others with their trailers. And we've been doing day-night, getting intercepts from active offenders, but also intercepts with customers and employees to get an idea of optimal dosing of these technologies day and night, right, to deter, disrupt documents, so forth. So that's exciting stuff, and I couldn't agree more. And sometimes we'll have that same thing that somebody will come to us and say, look, we've already got data. Can you help us make sense of it? In other words, analyze it. RCTs, but they seem helpful. And I think the more we talk about just higher level, more valuable, more rigorous methods, the more we are going to get the practitioners to help us gather the data and then start to insist. And we're seeing that in our area more and more now. They're starting to use these RCT terminologies and the triad or the triangle and so forth, but they're starting

to insist. In fact, to the point now where sometimes it's tough to talk to someone because they question everything. We're like, well, let's have a strike a balance here so we can communicate. And, but we'll always have some sites here for each other. So let's maybe talk a little bit, if we could, acoustic gunshot detection, you know, maybe you could describe it, you know, from your perspective, maybe technologically as well as deployment and maybe some use cases, things like that. That would be helpful, I think, for us and our listeners. Sure. So when we're talking about acoustic gunshot technology, there's a host of different ways that can be done. So the military has it and that's where it originates from the way we use

it in law enforcement. So originally it was designed to detect sniper fire, and that works effectively by when a bullet comes out of a barrel and it sort of goes faster than the speed of sound, it creates this sort of blast wave. Now that can be detected by all kinds of sensors, and that can sort of give you a direction of where the gunfire is heading. And so that was the sort of the original way that it was deployed. In the law enforcement setting, though, the way it typically works, at least most systems, they go by the muzzle blast. So the initial sort of energy that comes out of the barrel. sort of energy that comes out of the barrel. And that loud blast obviously gets detected by multiple sensors, microphones if you want, they're fancy microphones. And because sound travels at

its very specific rate, I think it's what 343 meters per second or something very close to it. And so it reaches these microphone locations at different times. And then you can, of course, very quickly calculate the location from where the gunshot was actually fired. And that's obviously what law enforcement want. They want a location from where it was fired, not necessarily where the bullet is traveling to. So there are all kinds of different detection systems out there. Some are mobile, like we're using on mobile surveillance trailers. Some are easily set up. But I would say the more typical gunshot detection systems that municipalities use in the United States are types like ShotSpotter, where you have a company that installs the sensors.

You know, they maintain the sensors. installs the sensors, you know, they maintain the sensors. And when a gunshot goes off, the sensors get triggered. AI technology then classifies it, whether it is actually a gunshot or whether it might be something else like a nail gun or whatnot. But then before it gets dispatched to police, there'll be an additional layer of review. And in ShotSpotter's case, dispatched to police, there'll be an additional layer of review. And in ShotSpotter's case, it's this incident review center that they have where people look over the sound, the acoustic signatures, and review it to make sure that it's actually a gunshot or not. Because obviously, you don't want to send too many false positives to a police department because that's very unhelpful if you want. So that's kind of how most detection

systems work. Most, at least in law enforcement, most of those plug in straight to police department. So effectively, when those gunshots get off, they get reviewed, they automatically go to either dispatch officers, in some cases to have access to apps where they can see on a map where exactly the gunshot was fired from. And so that helps them in their active investigation. So the good things about it is that it does appear to increase the response time or actually decrease the response time, I should say. So they get there quicker. This also leads to more evidence recovery, obviously. So you have a lot more, um, spatial clarity on where you're going to find shell casings, for instance, or where you're going to find your victims. And that helps with, uh, things like knife and link analysis, where you can sort

of figure out the networks in which people use guns. Um, the only thing we're really not quite too sure about it at this stage is whether it leads to crime reductions. So in the study I published in the Journal of Experimental Criminology, it didn't really show very good results with respect to crime reductions in St. Louis. that will be forthcoming in a little while. And there I did find some significant crime reductions. So we saw about 40 or 50 percent reduction in aggravated assaults in the neighborhoods that got shot spotter versus neighborhoods that didn't get shot spotter. Now in both cases, these are quasi-experimental studies, so you have you know, concerned about causality and those things. But I think the results are relatively robust in both cases. And it's an interesting thing that you find crime reductions in one place and not the other, right? And so I start thinking, well, why is that the case, right? And I think the best at this point, at least, and then we certainly need more

confirmation on this, but the key differences between St. Louis and Cincinnati in that respect was that in St. Louis, there was no specific standard operating procedure that guided the officers in how do they respond to these incidents, whereas in Cincinnati, there was a very elaborate one. So in Cincinnati, when officers go out to a shot spotter call, it has to be a two-man response. So there have to be two officers involved in the response. And a sergeant has to sign off after everything is done to make sure that everything was done correctly. log into the ShotSpotter app and then walk around a hundred foot perimeter of that incident on that app location. So that obviously puts officers on the scene for a lot longer than in many other cases. And on top of that, in Cincinnati, they have to knock on neighbors' doors. So they have to talk to neighborhood residents and tell them, this is why we're here. We're responding to this incident. If you have any information, let us know. And I think that that process that Cincinnati uses creates effectively a form of hotspot policing in

those locations. And so you have a lot of police activity in that location. And if you do that over and over and over, you hit the hotspots of gunfire. And I think that's one of the key reasons why we saw that effectiveness in Cincinnati. Now, another reason might be, for instance, that it's simply an issue of volume. So in Cincinnati, I think at the time that I was looking at, they had about 1,500 shot spotter calls per year. In St. Louis, that number is substantially higher. You're talking about 8,000 or so a year. It's probably even more now after several more expansions. But what that means is that you have a lot of officers who are going to be responding to those calls because those calls don't just happen all evenly across the day. They happen very much at the same

time in the same locations. So a few officers are responsible for responding to a bulk of these incidents. And that creates, I think, bottlenecks in the police response to these incidents. And that can be very problematic, I think. So I think the implementation of ShotSpotR creates these sometimes unforeseen consequences where you have these cases bottlenecking and how do you respond to that? respond to that. So yeah, I think there's interesting things going on in this. I think there is potential. Obviously, we need a lot more research from a lot more different sites to get more comfortable with these sort of ideas that I'm throwing out here. But I do think there's something to it that has some level of impact. But I think, too, we have to be very cautious. And a lot of police departments, you know, when they go, which with acoustic gunshot detection systems, they just go all out. And one of the problems with that, and this is something we saw, for instance, in Chicago,

Chicago just expanded so quickly, it's shot spotter coverage, that it becomes almost impossible to find good comparable locations to see if it's actually effective or not. And that is a very common thing that we see in the implementation of this technology, that people, you know, they like it, the officers like it, the police department like it. And so before we even have a good opportunity to do the evaluation, they take away the control areas, if you want, from us. And that makes it that much harder to do proper evaluations on this technology. Again, extremely interesting. And you and I know that journal editors and reviewers, I think they understand these things.

understand these things and they're always pressing us to, even in the journal articles, describe the real world, the grinding research issues that we all are faced with, like you said, bad data or this seems good, let's go, or it seems so good that we're going at scale here already. And so that precludes some of the more rigorous research and evaluation we'd like, as well as what you mentioned earlier. And that is, look, there are other metrics and measures, too, that we can use, you know, to maybe triangulate our data and make just make a lot more sense of what's going on. And, you know, I really enjoyed listening to your kind of description about what's going on there in the Midwest with some of this. And I think back just a few years ago when Dermot Shea, NYPD's police commissioner now, but at the time we had a kind of a exchange idea type of visit with them up there. And he was deputy commissioner of operations. But one of his

several initiatives he was working on was gunshot detection and placing those spotters in different areas and trying to figure out how to best do that. And there was a few things that I took away. One, they were absolutely stunned at how few gunshots, legitimate gunshots are reported to the police by the public. I mean, they just couldn't believe it. And so, but the other thing is they were doing what you were saying. They were treating every gun detected gunshot event as a crime event that merited a detective response level response. They taped the area, they'd look for the brass, as you said, they were looking for, you know, touch DNA and fingerprints and other markings, you know, forensic markings on the, on the brass or other things that they could find, linking all and putting

all that in so they could do, as you mentioned as well, that link analysis. And all right, where all is this shown up and what all can we learn from this so that we can better put the right people away and capacitate the high rate, high impact offenders, but get better at maybe prevention overall. So exciting stuff and interesting. And, you know, also I had one of these, what are the mechanisms? Well, you, you described that. And then also some of the limitations, which you described as well. And we had a, we had an organization here the other day that has, they are, they are showing that it's another military adapted technologies,

that it's another military adapted technologies, shot detection and locator, but they claim that they have one unit. They don't require multiple to triangulate, but rather theirs because of the way they've developed it for use overseas in combat only requires one. Is that something that sounds familiar or plausible? And they had data that seemed to reinforce that. I suppose it depends on what exactly you're measuring. So I think if you're measuring sort of this pressure wave or this breaking the speed of sound wave, that that can be plausible with one microphone. But I think it's hard to have really only one microphone, one sensor really, and get a super accurate location very far away. So I think it can work for a fairly narrow range, so a couple hundred feet maybe. But if you're trying to cover a very large area, one microphone is just not

going to do it. Okay, interesting. And I do recall talking about the shockwave, the different angles and versus the muzzle blast. And there was a lot of that discussion. We were standing out in typical Florida, you know, 98 degree heat, you know, 70 percent humidity. But when we're talking about some other sensors. So I did want to get your take on that. So I appreciate it. And we're going to be learning more shortly from them. And it's in place on one of the mobile towers, since each one of them are loaded up right now with different capabilities. You know, one thing also is we're looking at another technology that's more for indoor use. I understand there are a lot of limitations and something like that, if not a lot, a lot of limitations. Any thoughts on indoor detection, which is clearly an issue that we deal with, big box stores, distribution or fulfillment centers.

But we also have the small box guys that may or may not, these may not even work in them at all. What are your thoughts on indoor triangulation? Yeah, I mean, with an indoor system, you're not really that concerned about the location. You're more concerned about did a gunfire, did gunfire actually occur or not? And so a lot of these sensors for indoors kind of look like smoke detectors, the little tiny devices that just go on the wall and you kind of forget about them. But they can still tap into outside networks so they can notify police agencies of shots fired, for instance. And so if you have a sensor, then

let's say at a specific wing, you know that in that wing gunshots have happened. And that's good enough, I think, for sort of a retail setting where that shooter is going to be on the move anyway, probably. There's also systems that I was actually learning about that the other day that then can, in response to guns fired in a building, can then perform a lockdown on that building. So they can shutter the doors and lock everything down. I don't know the specific designs of this and whatnot. And I have some concerns about what happens when you lock an active shooter in a building with other people and how that specifically works.

But I think it's interesting, right? And I think the indoor sensors are substantially cheaper than the big, you know, acoustic systems that law enforcement use. I think those sensors, I think, go for about 200 bucks each. So you can quickly cover a building for relatively low cost with those sensors. Now, the question then becomes, is that going to be worth it? You know, I think from a business continuity planning issue, you're going to think, okay, yeah, we should have that. But from a realistic perspective, you know, how often does this actually happen, right? Most of the gunfire that happens across the nation happens in a very small

number of places. And those are typically not, you know, those are typically not commercial establishments. They're typically on the street. And so, you know, I wonder if maybe this is a necessity. I understand from a business point of view that you want to make sure you get all your, you know, all your bases covered in that sense. But whether it's cost effective, I'm not sure. I'll be honest on that. And I think in our case, I would say from a researcher's standpoint, we've got a lot of access. These retailers that we deal with have quite frequently experienced these active killer situations, including, of course, active shooter. And some of them even very recently and some have come in waves.

And those executives that have planned for and helped handle and recover from these events are part of them. Some of them are on our board of advisors and that we're working with them even currently. And I think a big part of this is this is real and this is the direction not to run toward. That's simple, right? Yeah. Yes. Yeah. I think as far as that goes, that's, when you're in that location, you hear the gunfire, right? So I think the biggest benefit of these systems is that you can get police notified quick. And I think that's the benefit of it. The police have an idea of, okay, this is, particularly if you have a very large office complex, for instance, good luck finding a shooter in that complex, right? If you have a multi-story, multi-campus kind of site, having a notion of

where the gunfire is coming from would be very helpful in that case. That is something that when people call the police is very difficult to discern from those calls. When people call about gunfire, they often have difficulty indicating where it came from. And even if they do have a good inclination, that may be wrong because, you know, acoustics bounce off buildings and do all kinds of crazy things. So, you know, the detectors, the sensors in that respect can provide a very good location if you're talking about a very large site, of course, like a school site, like a big campus, very large site, of course, like a school site, like a big campus, a big office campus or whatnot. So I think there's some advantages to it. I think the cost is relatively reasonable.

So yeah, I see why a lot of companies are doing that. I think there is the issue too, like you brought up, the mass shooting incidents that have been pretty prolific. And so I can see that big box retailers in particular are concerned about that. So I get it now. I understand why people are interested in it. Yes. And they're describing what you said with the buildings that a lot of the echo, the echoes going on and just general confusion. And, you know, you've got a range of people of different ages, you know, you've got children, you've got, you know, elderly. And so the idea is again, you know, is there a way that we can integrate lighting or other sounds so that they run this way, not that way. But then the opposite for the first responders, if you will run this way, You know, so it's early days, but it's something that's tragically needed to provide, as you

said, a more immediate recognition, you know, detection and recognition for the occupants to help them make better, safer decisions for themselves and others? And then secondly, you know, for the first responders, and then thirdly for prevention in the future, are there things we can do? And so we've had some after actions with some of the retailers where their distribution center or stores or offices have had an active killer shooter event in them. And then what were the lessons learned and so on? So this is really helpful though, to get some of your insights into this issue.

And just to add to that, too, there is, you know, acoustic detection being developed in all sorts of ways, too, right? It's not just about gunshot detection, too. There's companies out there that are developing ways, for instance, to detect rising aggression levels in a crowd. And this is being used in like entertainment districts, for instance, a street with a lot of bars, where the microphones essentially figure out if a loud conversation becomes louder and louder, it picks up on that and then can immediately notify police as well. So there are all kinds of interesting things that are happening in terms of acoustics, but also obviously other aspects of technology. So it's a very new, brave new world with this kind of technology and the possibilities of it, I think.

this kind of technology and the possibilities of it, I think. That's really interesting, escalating volume and maybe at some point with some other AI combined with escalating use of a certain aggressive terminology, you know, by the sender and receiver serve in Bali there. And I know too, with some of the ones we're working with, we're looking at, you know, glass breaking, metal banging, discreaming. But some of the other, yeah, threatening and duress speech and on and on that might be helpful to start to diagnose, you know. And so we're looking at with the AI right now, just what are all these baselines like everybody else? Okay. What's anomalous movement speech or how, you know, closure rates and things between two individuals and so on. That could be helpful in the moment and then later for documentation. So I think moving back out and you've touched on this a little bit, Dr.

Mears, but thinking a little bit about how does this acoustic gunshot detection, how do these systems impact crime and safety, let's say from more of a theoretical standpoint, but you know, what do these mechanisms look like if we're trying to work in a specific neighborhood or block or some, in our case, we're talking about specific building clusters or inside of even a space or a parking lot. What are some of the mechanisms of action of these systems that are useful for the current place users and then those that are trying to protect them? When you're talking about those mechanisms i i assume you mean the mechanisms to which it can reduce sort of the incidence of of the gunshots or is this not what you're saying yeah yeah that's

probably where i would start yeah the first is always yeah it seems how do we reduce these crime attempts these events that are so dangerous uh and then we can move from there. But yes, no, I think that's job one. Yeah, obviously. So my work is mostly focused on street violence, right? And because that's how law enforcement systems are being used to target outdoor street violence, because the gunshot detection systems don't really pick up gunfire if it's inside a vehicle or inside a building. I mean, there might be cases where it does, but in most cases, it's probably not going to pick it up quite as well as if it's open out in the street. And so I think the rationale here is that if we can get there quick, if we can provide a consistent response, it provides that level of deterrence that police often are looking for,

right? Because normally, if you get people shooting a gun, you might get a call for service, you might not get a call for service on that gunfire. And as you mentioned, a lot of people don't call in. I ran the numbers in St. Louis at some point, and I think the actual number of gunshots that were called in amounted to about 17, 18% of the total amount that was picked up by the acoustic system. So a very small fragment. And so being able to respond more consistently to these incidents, I think sends out a signal. It sends out this signal like, hey, we don't tolerate this here. Now that's probably very different from the way acoustic systems will be, you know, utilized in the retail sections where it's really more about an active shooter situation where we want to police here now.

And so response speed for those applications is a lot more important because it's not really about the turns, I think, in that situation as much. because it's not really about deterrence, I think, in that situation as much. I think it's about making sure that we, you know, contain the situation, get to police there, get that response going, and contain the situation with the least amount of casualties. So I think the two systems, the two different applications of the systems, both have this idea of, okay, we need to reduce gunfire. But in the case of the law enforcement segment, it's more about this general deterrence that we're trying to accomplish. Whereas in the retail section or sector, we're really more concerned about making sure that we contain it as quickly as we can. That's a great way to look at it. And we do look to a certain extent too at the other

side of the coin and that so much of the crime that occurs in and on retail environments is spillover, as you know. It's something that came out of that neighborhood. It's a guy shooting at each other in vehicles, and they end up there, or they are there for a certain period of time, or they run over there. We had, for example, a Kroger shooting where a white male ended up shooting black patrons at a supermarket chain that we work with. But really, that individual was actually going to shoot his wife, who happened to be African-American, at a church. They got wind of it. They had a heads up, were able to secure and activate a plan they actually had at that church. church so he just went across the street to that that that supermarket and started to shoot at and shoot people that looked similar to his wife you know so that's just one example you know as you

said where you know we didn't deter him we didn't get gain general deterrence unfortunately uh these things are always so complicated just like the one at the incident i just described but it's another example yet another one of spillover crime. So I think another part is just looking at, I really just kind of sort of come in on a glide path here and talk about, you know, your excellent article. You alluded to it a few minutes ago in the Journal of Experimental Criminology, you know, acoustic gunshot detection systems. In this case, it was your quasi-experimental evaluation. You know, what were you kind of, what was going on? What were your, what was your intent, if you could describe to the listeners with this, with your research, what'd you do? What'd you find?

What's it mean? Do you have another hour or? No, I hear i think you know our intent was to because we had done a study before on this we had done a study before in st louis and it was a relatively simple time series analysis study that we did and if didn't find results at the time this was back in 2012 i think and um after that the the gunshot detection system in St. Louis expanded. And we, after a while, decided, well, let's take a look and see if we have enough data to perform another study that's a little bit more rigorous with more data to see what's going on. And so we got all the data. We figured out where these gunshot detection sensors are, you know, what the coverage area is. And so we looked at the neighborhoods in that coverage area and compared them to a similar number of neighborhoods outside a coverage area. significantly faster. It also reduces significantly the number of citizen calls for service for gunfire, but it didn't reduce gun violence or aggravated assaults, homicides.

And so that's obviously a little bit disappointing when you have these kind of findings, null findings, and we didn't find anything significant in that respect. But that is what it is sometimes, right? And that can be an outcome of a variety of factors, as I mentioned before. So we were just really interested. And one of the key reasons why I was interested in it is because there is still very little work on this, right? There's only a few evaluations that have been done so far about gunshot detection technology. So any little bit that we do helps to and helps us to understand, well, does it in general work or does it in general not work? And if it does work in some situations, but not in others, what is the underlying reasons for this, right? And knowing and figuring out what those reasons are can help

us modify the situation to where we can make this more effective in all places. I think that's sort of the rationale why I do this kind of work and why I'm interested in it, is trying to figure out how can we reduce gun violence to the best of our ability by using new technologies and new ideas. So that's excellent. And I can appreciate too so much about the research. And as you said, you've done some more. We need to do more. And you're getting things published that are helping all of us better understanding, laying down some baseline data, things that we can all use and grow and build incremental research and find the findings that we need to take better, more science informed action and so forth. So, but also the idea that, okay, some of the needles

move, but not all the needles that we'd like to see moved as a result of this, of this treatment, this intervention. Is there any research out there that as a result of this treatment, this intervention. Is there any research out there that you're aware of that's looking more and more at the dosing questions? How many units, where we put the units to have maybe greater effect? It sounds like what's going on now is designed to start to move the needle around. Look, we need quicker response regardless. If it reduces overall crime events, that may or may not follow immediately, but at least we're getting there quicker and maybe incapacitating some of the victimizers and rapidly helping the victims. Yeah, I think one of the questions obviously is like, what can we do to make it most

effective? I think in itself, the technology works quite well. So most of the systems that I see at least appear to capture the majority of the gunfire incidents. And there's ways to figure that out, obviously. But I think the question too,, you know, the idea of the faster response is a question that's kind of an interesting one, right? So even with, you know, acoustic gunshot detection, you might get a faster notification of the gunfire. But does that really matter at the end of the day when police still have to drive three or four minutes to the scene? And I think that's kind of one of the limitations that people don't always think about. You know, so with gunshot detection systems, you might get a quicker initial notification, which would lead to a quicker dispatch. But that doesn't necessarily

mean that you shave off that much time to make a huge difference in finding somebody at the scene, right? At least an offender. If you get there in, let's say, four minutes instead of five minutes, is that really going to increase your shot of arresting somebody? And I think that's kind of a limitation of the technology where it doesn't necessarily do what we think it always does. And I think that's kind of a limitation of the technology where it doesn't necessarily do what we think it always does. And I think that's something we have to think about, too, is that the things we think technology will do for us, it will not always be exactly what we hope for. Or sometimes it's one of those unintended consequences where it seemed like a really great idea, right? consequences where it seemed like a really great idea, right? But the reality is, okay, if it still takes you four minutes to get there, in all likelihood, people are going to be gone. And so

we have to focus on the things we can do with it. So we can indeed find shell casings with it. We can indeed, it helps us figure out if victims and witnesses in these incidents are telling the truth, because oftentimes you have very hesitant victims and witnesses in these circumstances who are not always willing to share the full information, the full story, out of fear of more reprisal and those kind of things about these incidents. And gunshot technology can in some ways help with that, right? If we know exactly where the incident happens, we can start looking for evidence there. And that's one of the benefits, I think, an unintended benefit if you want off the system, things that it's doing for us that we didn't necessarily anticipate. And I think that's kind of the importance of the research in this field,

is trying to figure out what of the elements are the things that matter and which are the elements that perhaps don't matter as much as we thought they would. Excellent. So I guess to sum it up, what else do we need to do, right? Like any good journal article, what further research seems indicated from your perspective right now that's going to better safeguard people, but again, going some steps back, that's going to better enable X, Y, and Z to make that happen. Yeah, I think for now, I think it's an issue in many ways of getting agencies to do these evaluations. A lot of agencies haven't really looked at their gunshot detection systems. They just put it in there and they haven't evaluated it yet. And I think that's what needs to happen, that particularly when we put new systems in, we need to design the evaluation

from the ground up, be involved as research partners from the ground up so that we can help already prevent some of the mistakes that were made, but then focus on what can we do well with the system as it is. And I think that going forward with that is, you know, once we get on board as research partners with these agencies, help them as much as possible in making sure that we can do a solid evaluation, right? And even with gunshot detection systems, an RCT is almost impossible to do because you can't just have a few sensors here and there with most of these systems. They just typically cover a very wide area and that makes an RCT quite difficult. But that being said, you still have to have some kind

of good solid control areas that are very comparable to it to make some good inferences about the effectiveness of it. So I think in general, yeah, more research is what we need in this particular field. And that is kind of up to, you know, funding agencies to make sure that they keep hammering on these evaluations being part of the funding requirements. And that's something that the federal agencies are doing really well, but municipal agencies and state agencies are typically not requiring as much of these evaluations. I think that's something that should be pushed a little bit harder, to be frank. So yeah, I think there's a lot of potential there. I think there's a lot of ways forward, but a lot of that depends on the opportunities for us as researchers to be involved in that.

I think many researchers would love to be involved in this kind of work, researchers to be involved in that. I think many researchers would love to be involved in this kind of work. But it's just, at the moment, there's just not that much demand from the agencies in that respect. Fantastic, again. And I really appreciate your time. It's really interesting. And I guess one quick aside in listening to you two, and we have this issue when we're studying even armed robberies. But is there enough activity going on to properly power a good RCT, right? These are such relatively rare events, too frequent for safety reasons, but are they frequent enough for us to properly power a good rigorous test? Or are there ways that you guys have found to bootstrap and things like that? I would say with gunshot detection systems, I think there are certainly enough observations,

right? I think if you're thinking about gunfire itself, that is plenty abundant in many municipalities. If you're talking about what is often used as a dependent variable, gun assaults or homicides, given the coverage areas of much of those gunshot detection systems. So, for instance, some of the leading companies will only install a minimum of three square miles typically. Well, within three square miles, you're going to get enough observations of gunfire and gun assaults to make some good conclusions on that. You know, unless, of course, you have an agency that has not a really substantive gun problem and is still implementing it, then it might become an issue. But so far, at least, most of the agencies that have been implementing it have serious gun violence problems.

And I think that makes sense that we put those things there first, right? So from that point of view, from the numerical point of view, evaluating gunshot detection systems is really not that big of a deal. But obviously, you know, the not being able to do an RCT, that is kind of a limitation of that kind of work. Okay, excellent. So I appreciate it, all your time, your expertise, your experience. and practical issues and talk in plain language so much to help the practitioner and really all of us better understand and think about for improvement purposes rather than anybody just trying to, I guess, impress somebody else. But rather, look, let's better safeguard and let's

dig in here and let's learn together and improve. And it really, really seemed to us that your research, what your topic, but how you approach the topic and then further how you articulate that to us, how you explain that to the rest of the world and how you're integrating with real world, you know, guardians out there in the form of law enforcement in those communities. So, you know, kudos to you, and we appreciate your time today. Appreciate that, Reid. We'll be back in touch. And for all of our listeners, again, we appreciate you tuning in to Crime Science, the podcast. You know, and please let us know any questions, comments, or suggestions you might have at operations at lpresearch.org in this case. Everybody stay safe. at lpresearch.org. The content provided in the Crime Science Podcast is for informational

purposes only and is not a substitute for legal, financial, or other advice. Views expressed by guests of the Crime Science Podcast are those of the authors and do not reflect the opinions or positions of the Loss Prevention Research Council.