The Algorithm That Could Catch Serial Killers

Srdjan Randjelovic/Shutterstock

Around 216,000 people have gotten away with murder in the US since 1980. A nonprofit group wants to put a stop to that.

The Murder Accountability Project contains “the most complete data on US homicides available anywhere,” according to their site. It’s an open-source website that uses vast banks of data from state, local, and FBI sources to provide insight into murders, both unsolved and solved.

Anybody can access the database through the Search Cases bar (below), which can be broken down into geographic area, year of incident, the weapon used, and the victim’s sex, ethnicity, age, etc.

The project was created by Thomas Hargrove, a retired investigative reporter, data journalist, and former White House correspondent. Keeping with that journalistic spirit, the website was created in the interest of public transparency and accountability. As they explain, it hopes “to educate Americans on the importance of accurately accounting for unsolved homicides within the United States.”

Bloomberg reports that Hargrove used the raw data to find a pattern of murders in Gary, Indiana, around 2010. His analysis showed 14 unsolved murders  all strangled women between the ages of 20 and 50. Hargrove pointed this out to police, but they reportedly paid no attention. Lo and behold, four years later, authorities arrested Darren Deon Vann, suspected of killing at least five women. Although it isn't yet clear if Hargrove was spot on, it certainly showed he was onto something.

“I asked myself, 'Do you suppose it’s possible to teach a computer how to spot serial killers?'” Hargrove recently told Bloomberg.

Instead of being a passive bank of information, a computer system could be "trained" to find patterns within the data. They attempted to develop an algorithm that could look for common features and links between previously unassociated murders. For example, the system could notice a pattern of all young, white female victims that were strangled within one hour driving distance, each murder one month apart.

-

Hargrove decided to test out his algorithm on a case that was already solved. He experimented with the system by seeing if it could identify the Green River Killer, aka Gary Ridgway, who killed at least 71 women in Washington State during the 1980s and 1990s. By feeding it all the necessary information and fine-tuning the algorithm, the computer system managed to successfully link the victims of the Green River Killer and thereby identify him.

Further work needs to be done. The algorithm also linked over 100 unsolved murders of women in Phoenix and Los Angeles, many of which were attributed to several different people.  

Nevertheless, this in itself serves as proof of concept that it may be possible for computer wizardry and data to be used to hunt down serial killers. The riches of data made available on the Internet is regularly used to predict behavior, namely by social media networks and advertisers. But who knows, perhaps it could also save lives.

Comments

If you liked this story, you'll love these

This website uses cookies

This website uses cookies to improve user experience. By continuing to use our website you consent to all cookies in accordance with our cookie policy.