Using data to track down serial killers
A former investigative journalist has developed an algorithm that uses FBI homicide data to track down murderers. Read More
The killings committed by Jack the Ripper are probably the most famous example of a series of unsolved murders. Even today, over 100 years later, new theories about the killer’s identity still occasionally make the rounds. But getting away with murder, even serial murder, is not a phenomenon of the past. The files of hundreds of thousands of unsolved murder cases, what are known as cold cases, fill record rooms around the globe.
In the US, a former investigative reporter is experimenting with a way to use data track both cold and not-so-cold cases. In 2015, Thomas K. Hargrove founded the Murder Accountability Project, a US non-profit to track unsolved homicides across the US. He developed an algorithm that uses FBI homicide data to identify clusters of murders that show an elevated probability of including serial killings. It is estimated that one third of all homicide cases in the US remain unsolved, and Thomas is convinced that this is due in part to a lack of national data. There is not central database to compile and evaluate data from unsolved cases, so no one knows how many of these homicides are part of a series, which would make them easier to solve. Thomas and his team work to gather the necessary data, assess it using the algorithm they developed, then make it available to the public on their online platform. Anyone can search in their database to see what kinds of homicides happened when and where. Other documents provide about a detailed overview of victims, the circumstances of their deaths, and possible offenders. Users can even use the algorithm to search the data and detect the possible activity of a serial killer themselves.
“The project is an outgrowth of a 2010 national reporting project on unsolved serial killings when I was a correspondent for the (now defunct) Scripps Howard News Service,” Thomas told us. He had always been a data guy, and when he got ahold of the FBI's Uniform Crime Report (UCR) for the first time as part of his investigation, a question immediately popped into his mind. “I don’t know where these thoughts come from, but the second I saw that thing, I asked myself, ‘Do you suppose it’s possible to teach a computer how to spot serial killers?’” he told Bloomberg. The algorithm Thomas developed compares data points like the age, race, sex, and ethnicity of victims, as well as the weapon used and the circumstances surrounding the killing. It then identifies clusters. The reoccurrence of a specific killing method, for example, can be an indication of a serial killer.
Using this, he evaluated several years of UCRs and noticed two astonishing things: First, the number of unsolved murders had increased every year since the 1960s. In 2010, he identified more than 185,000 unsolved homicides committed since 1980, and this number has grown to more than 220,000 today. How, he wondered, was that possible given huge advances in technology and forensics, such as DNA testing and a range of helpful gadgets. Secondly, he noticed out that many serial killings went undetected. “Since few Americans realize the depth of the problem, we wanted to make homicide clearance information easily obtainable for the public,” Thomas told us.
Even before founding the Murder Accountability Project and dedicated his life to identifying murderers two years ago, Thomas' algorithms were already proving their effectiveness: In 2010, authorities in Youngstown, Ohio and Gary, Indiana opened new homicide investigations as a result of Hargrove’s findings. The algorithm’s identification of 15 unsolved strangulations in Gary was corroborated in 2014 with the arrest of Darren Deon Vann. Vann confessed to killing women for decades and took police to abandoned properties in Gary where the bodies of six previously unknown strangulation victims were recovered.
Perhaps surprisingly, there are no mandatory reporting laws that require police or criminal institutions to report murder investigations to a centralized, federal institution. “No one knows all the names of these victims because no law enforcement agency in America is assigned to monitor failed homicide investigations by local police departments. Even the official national statistics on murder are actually estimates and projections based upon incomplete reports by police departments that voluntarily choose (or refuse) to participate in federal crime reporting programs,“ the Murder Accountability Project homepage reports.
“The homicide clearance data on our homepage comes from the FBI’s uniform crime reports. The individual case information comes from the bureau’s Supplementary Homicide Report (SHR). Both are voluntary reporting systems. We seek to augment the SHR by approaching states and local police departments that decline to report data to the Justice Department,” Thomas says. Together with a team of retired law enforcement investigators, investigative journalists, and criminologists, he seeks to obtain information from federal, state and local governments about unsolved homicides and calls on these institutions to publish the information.
“Unlike the federal government, the Murder Accountability Project does not regard the reporting of homicide information to be ‘optional’ by police. We believe the American people have the right to know how they are being murdered and whether those killings are being solved by police. If necessary, we are prepared to file lawsuits under local Freedom of Information Acts. To date, we’ve obtained data on more than 23,000 homicides that were not reported to the Justice Department. These efforts will continue.”
His platform is probably the most complete data on U.S. homicides available anywhere. “We know that a significant portion of our users are police. We’ve already presented demonstrations to the International Homicide Investigators Association and the International Criminal Investigative Analysis Fellowship. This year, we will be presenting to the Mid-Atlantic Cold Case Investigators Association and the Southeast Homicide Investigators Association,” Thomas told us.
Communication is not just one-way either with police investigators simply checking the collected data online. If the team spots something suspicious, they get in touch with local police. For example, they believe they have spotted undetected series of homicides in Cleveland and in Atlanta. They've followed up with the police in those communities to explore the possibility that at least some of the killings within highly suspicious clusters were the result of a common offender.
Although the data are primarily there to assist police and the authorities in their investigations, Thomas and his team feel it is important to make them accessible to the public. He also hopes the public can help solve more cases. “It’s impossible to say how many serial killings we have spotted. There are dozens of highly suspicious clusters in which too many homicides of similar victims killed by similar means within specific geographies went unsolved. But we believe police will seriously consider the possibility that suspicious clusters of homicides could be the result of multiple-victim killers. We hope more arm-chair detectives will review our data and make important discoveries that might be useful to police.”
Visit the portal and give it a whirl – maybe you’ll be the one to spot the next Jack the Ripper!