Thomas Hargrove, a 61-year-old retired news reporter from Virginia, was always the numbers guy at his paper:
In 2004, Hargrove’s editors asked him to look into statistics surrounding prostitution. The only way to study that was to get a copy of the nation’s most comprehensive repository of criminal statistics: the FBI’s Uniform Crime Report, or UCR. When Hargrove called up a copy of the report from the database library at the University of Missouri, attached to it was something he didn’t expect: the Supplementary Homicide Report. “I opened it up, and it was a record I’d never seen before,” he says. “Line by line, every murder that was reported to the FBI.”
Every year he downloaded and crunched the most recent data set. What really shocked him was the number of murder cases that had never been cleared. (In law enforcement, a case is cleared when a suspect is arrested, whatever the eventual outcome.) Hargrove counted 211,487, more than a third of the homicides recorded from 1980 to 2010. Why, he wondered, wasn’t the public up in arms about such a large number of unsolved murders?
To make matters worse, Hargrove saw that despite a generation’s worth of innovation in the science of crime fighting, including DNA analysis, the rate of cleared cases wasn’t increasing but decreasing — plummeting, even. The average homicide clearance rate in the 1960s was close to 90 percent; by 2010 it was solidly in the mid-’60s. It has fallen further since.
His innovation was to teach a computer to spot trends in unsolved murders, using publicly available information that no one, including anyone in law enforcement, had used before. This makes him, in a manner of speaking, the Billy Beane of murder. His work shines light on a question that’s gone unanswered for too long: Why, exactly, aren’t the police getting any better at solving murder? And how can we even dream of reversing any upticks in the homicide rate while so many killers remain out on the streets?
It took a few years for Hargrove’s editors at Scripps to agree to give him enough time to lose himself in the FBI’s homicide data. With help from a University of Missouri grad student, Hargrove first dumped the homicide report into statistics software in 2008. He spent months trying to develop an algorithm that would identify unsolved cases with enough commonalities to suggest the same murderer. Eventually, he decided to reverse-engineer the algorithm by testing his ideas against one well-known case, that of Gary Ridgway, the so-called Green River Killer, who confessed to killing 48 women over two decades in the Seattle area. Hargrove thought that if he could devise an algorithm that turned up the Green River Killer’s victims, he’d know he was on the right track.
“We found a hundred things that didn’t work,” he recalls. Finally, he settled on four characteristics for what’s called a cluster analysis: geography, sex, age group, and method of killing. For gender, he stuck with women, since they make up the vast majority of multiple-murder victims who aren’t connected to gang-related activity. When he used women between the ages of 20 and 50 — the cohort most commonly targeted by serial killers — the algorithm lit up like a slot machine. “It became clear that this thing was working,” he says. “In fact, it was working too well.”
The Green River Killer came up right away in this algorithm. That was good news. Hargrove’s algorithm also pulled up 77 unsolved murders in Los Angeles, which he learned were attributed to several different killers the police were pursuing (including the so-called Southside Slayer and, most recently, the Grim Sleeper), and 64 unsolved murders of women in Phoenix.
Then there was a second group of possible serial killers, those unrecognized by local police. “The whole point of the algorithm was to find the low-hanging fruit, the obvious clusters,” Hargrove says. “But there were dozens and dozens of them all over the country.”
In 2015, Scripps spun off the last of its newspapers, and Hargrove and the other print reporters lost their jobs. “The only guy who left with a skip was me,” he says. Hargrove, who was 59 at the time and had worked at the company for 37 years, qualified for a large severance and a nice pension, leaving him well-covered. Now he had enough time to go all in on his data project. He founded the Murder Accountability Project, or MAP, a tiny nonprofit seeking to make FBI murder data more widely and easily available.
Using Freedom of Information Act (FOIA) requests, MAP has tried to chase down data from the many municipalities and counties that weren’t supplying their murder data to the FBI, out of bureaucratic laziness, a lack of manpower, or perhaps just rank incompetence. MAP has already assembled case details on 638,454 homicides from 1980 through 2014, including 23,219 cases that hadn’t been reported to the FBI. This is the most complete list of case-level details of U.S. murders available anywhere, and the group’s website has open-sourced all of it. Anyone with statistical analysis software, available for free online, can start looking, across jurisdictions, for serial killers. Anyone can compare convicted killers’ timelines against the timing of unsolved murders to determine if a connection is plausible. “You can call up your hometown and look and see if you see anything suspicious,” Hargrove says. “If you’re the father of a murdered daughter, you can call up her record, and you can see if there might be other records that match. We wanted to be able to crowdsource murder.”