Only 52 percent of the state’s homicides result in an arrest when the victim is black. Could automating police work help?
Keith Allen Bledsoe, the sixth teenage homicide victim in Lexington, Kentucky, this year, died as the other five had: by gunshot. On June 26, Lexington police found the 17-year-old Bledsoe’s body in the streets of Harris Court, a cul-de-sac near I-64. If confrontation or argument had preceded Bledsoe’s murder, none of the neighbors reported hearing it to police. If they heard gunshots, seemingly no one peered outside to investigate them. Officers reported no suspects or relevant witnesses, only shell casings and the gunshot wound to Bledsoe’s head. The person who called 911 didn’t report a shooting, instead telling operators about a “motionless” man on the ground around 2 a.m.
Gun deaths in Kentucky are spiking, particularly for young black men. From 2010 to 2017, Kentucky police arrested a suspect in only 52 percent of homicide cases where the victim is black, according to The Washington Post. (Seventy percent of the city’s homicide cases result in an arrest when the victim is white.) Many of Kentucky’s black deaths, in short, go the same as Bledsoe’s: Someone is shot, and no one is called to justice.
There are lots of reasons for people to ignore gunfire when they hear it: They may be concerned about falling under police suspicion themselves, and in areas with high rates of gang violence, being seen as helping the police can make you a target. In Oakland, California, another high-crime city, the month of September saw 395 recorded instances of gunfire, but only 208 phone calls to police to report gunfire.
Ralph Clark is an Oakland native and the CEO of ShotSpotter, a gunshot-detection-technology company. He believes that unreported gunshots don’t just act as symptoms of community mistrust of police—they reinforce it. “When communities see police not responding to these [gunshot] events,” he said over the phone, “but at the same time have the resources to respond to low-level arrests and intercepts for marijuana and stop-and-frisk, that’s a pretty cynical situation.”
Many cities grappling with gun violence have turned to technology to break this cycle, usually empowering police with new forms of surveillance. In Kentucky’s capital city of Louisville, 90 minutes from where Bledsoe was shot and killed, local government has turned to a novel combination of surveillance technologies to react to shots when citizens can’t, or won’t. But automating police could mean enabling the cycle of mistrust that itself abets gun violence.
In 2017, Louisville finalized a $1.2 million contract with Clark’s company. ShotSpotters are microphones, usually attached to streetlights and traffic poles in neighborhoods with high rates of gun violence. They are attuned to the specific percussive audio signature of gunfire; when a gunshot is detected, the devices send immediate location data to police, telling them precisely where the shots rang out and, in some cases, the make and model of the gun and whether there are multiple shooters.
But Louisville is also considering using drones to even further augment the ShotSpotter system. Earlier this year, Bloomberg Philanthropies, the charity foundation founded by former New York City Mayor Michael Bloomberg, gave Louisville’s Office of Civic Innovation and Technology $100,000 to tackle they city’s spiking homicide rate. Over a four-year period, the office will test the feasibility of using self-guided drones to investigate shootings: According to the proposal, the drones would be sent the GPS coordinates of a shooting location, then they would take pictures and videos ahead of first responders, complementing location data with visuals.
It’s a neat system with obvious benefits: “With [an unmanned aerial vehicle] equipped with eyes,” Clark said, “you’ll be able to get dispatch to the scene very, very quickly, first and foremost to see if there’s a victim there. And you can alert EMS. That you can potentially observe a potential [suspect or witness] and identify evidence is kind of intriguing, I think.” Clark and ShotSpotter were aware of the city’s plan, but not directly involved in the pitch process, he said.
According to Bard College’s Center for the Study of the Drone, more than 900 fire and police departments across the country have at least one drone. Eleven of Kentucky’s police and fire agencies have drones, according to the center. One of them is the Georgetown Police Department, about 30 minutes from Lexington.
“Usage of a [drone] would be based upon the totality of circumstances,” Lieutenant Philip Halley, who leads the Georgetown police’s drone unit, told me: “the scope of the crime scene, the distance over which the incident occurred, and whether we thought there may still be persons involved in the area of the incident, whether they be suspects or victims.”
Halley noted that drones, manned or unmanned, are only useful to investigations under specific circumstances. First, the crime needs to have occurred outside, rather than indoors. The sound of gunfire inside a building could theoretically trigger drone response, but Halley has doubts about the devices arriving ahead of police, even with the head start afforded by ShotSpotter data. From there, considerations only pile up: Rain, heavy winds, and fog render drones unusable, for example. Drones aren’t allowed within five miles of airports, and they are hamstrung by weight restrictions and Federal Aviation Administration regulations.
Louisville’s proposal has drawn criticism from privacy scholars, particularly those studying police and technology.
“The power of flying cameras is not equivalent to fixed cameras, as it will capture more area than the existing cameras,” said Andrew Ferguson, a David A. Clarke School of Law professor and the author of The Rise of Big Data Policing. Ferguson points out that drones would capture much more data, creating the problem of mission creep: technology brought in for one purpose eventually being used for another. Drone footage could be matched against criminal databases, mined for audiovisual data, or, as has been argued about ShotSpotters themselves, used to justify further encroachment by police into vulnerable neighborhoods.
“Once you have flying cameras available, they will likely be used beyond the pilot project,” Ferguson said. “They will fly more and capture more data. This is a perfect example of how big data surveillance will change the power balance between citizens and police and erode community trust.”
Policies may be written to limit access to footage or prevent it from entering databases, but for those who oppose the option outright, any policy limitation will be insufficient. Further, laws lag far behind the pace of technology, granting police broad powers when legislative ambiguity doesn’t specifically prevent certain uses. And, of course, no one can know what they will do to police-neighborhood relations. If fear of police is part of the reason Kentuckians don’t call 911 in the first place, what will the sight of law-enforcement drones overhead mean?