what does a serial killer look like to a computer?

1:16 PM on Friday, June 14th, 2019 (Claremont, CA)

What does a serial killer look like to a computer?

To answer this question, you need to dig a little deeper. Oftentimes, we think of serial killers in the US as middle-aged white men, with a vendetta against prostitutes, or with a history of bed-wetting. The Radford University/FGCU Serial Killer Database will tell you that most serial killers kill for enjoyment; they require the thrill, lust, or power that they derive from such an interaction. This database will also tell you that, of the serial killers that have been documented in this data set, they have an average IQ of 94.5.

So, how can a computer make sense of these linkages? Do we scrape census data for middle-aged white men? Do we put everyone with an IQ between 74.5 and 114.5 on a potential serial killer watch list? Surely, there’s got to be a better and more reliable way.

In 2014, Thomas Hargrove, a retired journalist, figured out how serial killers could be identified using a computer algorithm. Instead of looking at the killer, he focused on the victims.

Humans are creatures of habit. We like routines. We like going the same way to work everyday, we like getting the same cup of coffee every morning, going for the same run everyday, and on and on and on. Routines provide comfort, and safety. They also allow us to zero in on favorites. Despite committing abhorrent acts, serial killers are still human, and as a result they tend to follow patterns and routines when it comes to their victims. By scanning the FBI’s released data set of unsolved murders in the US since 1980, Hargrove was able to identify possible groups of victims.

The data set he relies on gives information about victims such as race, gender, age, location, and the method with which they were killed. This information is then transformed into a numerical identifier. If a large number of similar identifiers appear in a given location, they are flagged as being the potential victims of a serial killer, and Hargrove contacts law enforcement agents in that area with this information. It is then up to them to pursue it or not.

With the advent of serial killer catching algorithms comes the advent of serial killer catching algorithm evading serial killers. One has to wonder if this information could allow a serial killer to effectively beat the algorithm, and law enforcement. A particularly cunning person could kill people of different races, socioeconomic backgrounds, and genders, all in different locations, with varying methods. This would trump Hargrove’s algorithm, and would need an additional factor such as travel analysis to identify a linkage. Theoretically, it would require a surveillance state to catch, which is an even bigger topic up for debate.

Regardless, I found Hargrove’s work incredibly interesting. If you do too, The Atlantic made an interesting YouTube video about it linked below: