Over the past decade, the Chicago Police (and, for that matter, other urban police departments) have come to rely more and more heavily on technology, apps and database management to supplement their police work. Some might fear it’s slowly supplanting command decisions.
Chicago’s police stations, many of them built new during the 80s and 90s, were constructed with robust internet backbones in order to quickly accommodate the rapidly-approaching tech revolution.
Today, it’s here.
Pole-mounted cameras have been around for years, along with laptops in patrol cars and electronic files for offenders and contacts.
But what Michael Wasney wrote about in last week’s South Side Weekly is far more advanced than many Chicagoans might have expected.
It’s called The Shots Heard Round the City.
It has to do with ShotSpotter, just one suite of software the CPD has now deployed against street violence.
“They told me that this was actually the most rapid rollout of ShotSpotter that SST Incorporated, the company that manufacturers it has ever done,” Wasney explains. “So I actually did not get that confirmed before press time by the CEO, but I think that is very reflective of how emphatically they have embraced this technology, despite the fact that A) they’ve had dissatisfactory results with ShotSpotter in the past, and B) many other cities have faced concerns about its effectiveness, possible legal oversteps with regards to constitutional rights and eavesdropping laws.”
So what can this rapidly-deploying technology do, exactly?
“So as it stands, I guess by the end of February the Chicago Police Department plans to have around 130 square miles of the City blanketed by this acoustic gunshot protection software,” Wasney begins.
And for the record that 130 square miles is almost half the the total city footprint.
“It’s a lot,” he continues, “Which makes sense because again, by the end of February they plan to have 12 of the 22 police districts in Chicago blanketed by the software…it’s basic technology that is literally a physical acoustic box basically. It’s about the size of a toaster. It’s put on rooftops, telephone poles. I think the company advertises that they like it to be above 20-feet above the ground for optimum efficiency, and basically I think there are supposed to be about 24 sensors per square mile, although that varies depending on specific features of where it’s placed – topography, the density of buildings, how tall the buildings are, etc., etc. But essentially that’s going to be in 12 of the 22 police districts, and essentially what happens is say a gunshot is fired, three of the sensors will basically pick up the sound of the gunshot and you need three basically to triangulate it. That sound is then effectively sent to the California headquarters, I think it’s in Newark, California, of ShotSpotter, where it’s essentially determined by human experts as to whether it’s actually a gunshot, at which point it is pinged back to the Police Department and also to smartphones.”
The company claims that its Chicago installations are accurate to 80 feet. Standard Chicago bungalows are often on 25X125 lots, so if a gun is fired on the front porch, the Spotter will find it on that lot or maybe one house away.
But here’s the tricky part. These Shot-spotters are, after all, microphones, and they’re always recording. What happens, we ask Wasney, to all that recorded material?
“Even before a gunshot is fired,” he explains, “The microphones are always recording, and all of that data is retained in a server, which is physically based in California. So all of that auditory sound data is being downloaded onto the server, and then not just when a gunshot is fired, when a sound that whatever the algorithm is, when a gunshot is fired and the algorithm kind of detects that as a possible gunshot then the two seconds of audio before the boom and four seconds after is basically sent to the California headquarters, where then that’s the first time that you see a human expert involved in the process, at which point they go through an extra vetting process where they determine is it actually a gunshot, is it a firework, is it like a car. If it is confirmed to be a gunshot at that point it’s pinged back to the CPD or whatever other agency is using this technology. So, yes, all of the sound data is preserved on their server, although at a certain point it is deleted just for reasons of storage and data.
But, we ask, does that raise issues of people being inadvertently recorded, in a way they wouldn’t have wanted, that has nothing to do with a gunshot or a crime?
“It has happened,” Wasney asserts. “Not in Chicago, but it has happened. There have been several cases, and this has raised a lot of alarm and controversy concerning Fourth Amendment rights and, again, eavesdropping statutes. I believe it happened in Oakland, California. I also believe it happened somewhere in Connecticut, where basically someone fired a gun and in the two seconds prior and four seconds after they basically said something that ended up incriminating them and it was used in court to incriminate that person. Which obviously the representatives of the defendant raised concerns that that was overstepping the constitutionally-allowed… It was basically infringing on the constitutional rights of the defendant.”
And one more issue. Chicago doesn’t own this equipment. We’re not installing it. And We don’t own the rights to it. But we buy our own data from the company.
“So,” Wasney explains, “As it stands ShotSpotter is a subscription service and it was not always that way. This technology has actually been around for decades at this point, but I think once it was 2011 or 2012 the company made a transition from a high-priced system that an agency would buy fully to a subscription service, and there are probably a couple of reasons behind that. One, it was really really expensive for an agency to purchase the software, but I think another reason was honestly because they now have the ability to keep the data and the data is proprietary, so they can sell it if they wish.
There’s so much more in this discussion. There’s CPD’s use of HunchLab, predictive policing software that Wasney describes as “essentially a machine-learning software that basically predicts crime, and where crime will happen,” and helps identify “hotspots” in the City.
“A huge worry around a software like this is that it will simply act as a proxy for other biases,” he asserts, “So if you are having historical crime data that is for the most part focused on the south and west sides of Chicago simply because historically there’s been an over-policing of those areas.”
And finally, this from Wasney’s South Side Weekly article:
“The vast majority of alerts received from the SST’s headquarters to CPD, back to the Chicago Police Department, the vast majority resulted in dead ends. There was no evidence found, no reason to begin a case, let alone make an arrest. Of 4,814 unique ShotSpotter-linked events identified by the OEMC’s data just 508 of 4,800, a little over 10%, resulted in the CPD finding enough evidence to open an investigation. That’s roughly analogous, and this is important, with the rate of cases opened from solely human-related gunshots reported to 911.”
Listen to this discussion on SoundCloud.
Read the full transcript of this discussion here:transcript Jan 11 2018 Michael Wasney