As reporter Thomas Hargrove was researching a story on prostitution, he came across some unrelated federal data on murders, including the age, sex, and race of every victim, the weapon used in the crime and more.
This unexpected windfall of raw numbers piqued his curiosity. “I wondered if it would be possible to teach a computer to spot victims of serial murder,” he tells A&E True Crime. After bugging his bosses for six years, he finally got the go-ahead to develop a mathematical program to detect murder patterns around the country.
After retiring from the now-closed Scripps Howard News Service, Hargrove, 62, founded the nonprofit Murder Accountability Project in 2015 to draw attention to and improve the data on unsolved homicides in the United States. The site now contains the most complete database of homicides and unsolved homicides in the country, he says.
A&E True Crime spoke with Hargrove about the 2,000 or so serial killers he believes are prowling the streets right now, how his algorithm helps tease out data and what Hollywood gets wrong about these murderers.
You developed an algorithm in 2010 to detect possible serial killers who haven’t been caught. What are some of the data points you analyze?
What works in identifying some types of serial killers is to take all the records of individual killings and cluster them into groups. The clustering variables are the geography, whether it’s a county or metro area, the gender of the victim and the method of killing. Was she shot, was she strangled, was she poisoned, etc. Then we calculate the rate at which homicides were cleared through arrest within every cluster and then we look for clusters that had abysmal clearance rates.
What’s considered abysmal?
Nationally, in the case of women, we tend to clear about 80 percent of female murders. We look for clusters of female murders that had nowhere near that rate. We set it at about a 30 percent clearance rate.
[Watch Cold Case Files in the A&E app.]
How do you define ‘serial killer’? The FBI says it’s someone who kills two or more people in separate events but I’ve also seen three victims as the threshold.
We prefer the Justice Department’s definition of two or more in separate incidents.
Serial killers are much more likely to be men. What other characteristics do they tend to share, according to your algorithm?
The algorithm is only good at detecting uncaught serial killers, so that means the algorithm tells us nothing about the killer. It identifies individual victims who have an increased probability of being linked by a common killer.
Do victims have any common characteristics?
The FBI tells us that most serial killer victims are female. I haven’t gone back and looked at age or race.
How many serial killers are operating in the country right now?
There are probably more than 2,000 unrecognized serial killers.
Do you always inform local police departments or the FBI of your findings?
The short answer is, ‘not yet.’ We don’t regard [all the findings] to be actionable intelligence. All it is, is a list of murders that our mathematical system says has an increased chance of being linked by a common killer or killers. Our standard is, we will contact police when we have reviewed cases and have put together names and narratives within a cluster, and after that review determine that a serial killer is likely. These are unproved serial killings, even by us. But we don’t want to take a chance of not putting out that information.
How often do they take action?
I’m not at liberty to discuss that. It’s a confidential communication. But our experience is that law enforcement has been very professional. But they’re overwhelmed. There are not enough homicide detectives.
You’ve said that police can be reluctant to publicly announce if a serial killer is at work. Why?
In general, there is a reluctance to want to cause panic. ‘Serial killer,’ is a dreadful phrase that people are horrified and fascinated by.
What crimes have been solved or are currently being investigated due to your database?
There are active investigations around the country pursuing leads that the algorithm has prompted. But I can cite zero arrests because of those investigations. Catching a serial killer is not easy, by definition.
You believe serial killers are targeting victims in Chicago. Tell us what you’ve found.
We put names and narratives and locations to the asphyxiation and strangulation deaths that the algorithm is identifying in Chicago. I’m absolutely convinced that there is more than one killer active.
Based on your data, where else are serial killers likely lurking?
You’re going to find there are potential serial victims almost everywhere. We are convinced to our very bones that there are active serial killers in Chicago and Cleveland, and we have contacted police. I’ll say almost every major American city has multiple serial killers and multiple uncaught serial killers. Good luck sleeping tonight.
Serial killing is a much more common phenomenon than we know. There are 220,000 unsolved murders in the U.S. since 1980. There are hundreds if not thousands of serial killers tucked into those cases.
There are killers everywhere. I just think we should catch these guys.
Serial killers have been fictionalized over the years in such films as ‘Silence of the Lambs and ‘Psycho.’ Does Hollywood ever get it right?
No. The most common kind of serial killers are drug-organization enforcers and gang enforcers, not the Hollywood presentation of the serial killer. A great many serial killers are just crazy, and dysfunctionally so. A not-insignificant number are homeless. That’s not good Hollywood. That’s bad Hollywood. A lot of these people [may look] emaciated because he doesn’t eat regularly, he doesn’t bathe or wash his clothes. Who wants to see a movie like that? I have a lot of complaints about Hollywood glamorizing a mental illness and a societal failure.