fb-pixel Skip to main content

Police clicking into crimes using new software

Harvard students’ software seeks to help police dismantle networks

Florian Mayr (left), Matt Polega, and Scott Crouch started Nucleik, which helps investigators like Stephen Gregorczyk (second from right). Matthew Cavanaugh for The Globe

At a recent intelligence briefing in Springfield, State Trooper Stephen Gregorczyk learned that a local drug dealer had been robbed and was looking for revenge. Knowing little about the dealer, Gregorczyk turned to a new software tool that gathers all intelligence and background information about a suspect onto a single platform.

“I walked back to my desk,” he said, “typed in the name, and boom:” Every known location for the dealer, every brush with law enforcement, and every person ever linked to him popped up in one neat view on Gregorczyk’s computer screen.

The intelligence chief of a special unit combatting gang violence in the Western Massachusetts city, Gregorczyk now had a road map to plot how his team could head off the dealer before something bad happened.


“In this situation, we have the ability to stop a shooting before it happens,” Gregorczyk said.

Called Nucleik, the software is being tested by Gregorczyk and his gang unit. Nucleik is the brainchild of three Harvard University engineering students who hatched it as a class project for a professor with friends in law enforcement. The students were struck by how little technology was used by police to organize all the information they gather in their surveillance of gangs.

Using a hypothetical case, the tool shows how data can be analyzed.Nucleik

“We’re seeing these guys fighting crime everyday, putting their lives on the line everyday, but they’re not doing it with the right tools,” said Scott Crouch, cofounder of Nucleik.

So they set about to create a single platform for multiple uses, whether as a mobile app used in the field for street-level info or as a powerful desktop tool that could sift through mountains of data. The Springfield gang unit has been trying out the first version of Nucleik since mid-summer.

“Normally you’d need probably five pieces of software to do all of this and it would take hours. Now with one software, it takes minutes,” said Crouch.


For example, Nucleik could match disparate information to identify only those gang members of a certain age that frequented a particular address; the system can also display a criminal network graphically, allowing police to see the far-flung connections among gang members, drug dealers, financiers, and information spreaders.

“Police officers are wasting time every day using so many different software products, filling out so many reports,” said Matt Polega, another Nucleik co-founder. “And that’s time that they could spend back out in the field protecting our neighborhoods.”

Law enforcement has been collecting and learning from crime data for decades, and early on have used such information for major breakthroughs — determining that most crimes in a given jurisdiction occur at just a small number of locations, for example.

But as investigators have expanded the depth and reach of the information they collect, the tools needed to analyze that growing pile of data have not kept up.

“Unless you’re Columbo, you can’t do that kind of analysis yourself,” said Glenn Pierce, principal research scientist for the School of Criminology and Criminal Justice at Northeastern University.

In Springfield, ranked the 12th most dangerous city in the United States in 2012 by the FBI, Gregorczyk and his fellow troopers joined an innovative program that uses the counterinsurgency model from Army Special Forces to target gangs in the city’s North End, which has had a dramatic uptick in crime and violence during the past decade.


“We’re not a traditional law enforcement unit,” said Gregorczyk, who, like other members of his team, has a military background. “We’re a very small team and have a very specific mission.”

The mission, in military parlance: to deny, disrupt, and ­degrade gang and drug activity.

So when Crouch, Polega, and the third member of the Nucleik team, Florian Mayr, approached Gregorczyk about developing software to help with that mission, the intelligence chief readily welcomed them.

There are three primary tasks law enforcement information and intelligence systems should be able to do, said Kathleen Carley, professor at Carnegie Mellon University’s Institute for Software Research and chief executive of the company that handles the off-the-shelf version of ORA, a widely used social network analysis tool. One task is to gather information quickly, then present it visually and analyze it to identify useful links — such as those among the drug dealer in Springfield and his street associates.

“From the beginning we realized we needed a whole integrated system,” said Crouch. “Because the data collection and the analysis tools are in the same system, we’re allowing [investigators] to leverage all that data instantaneously.”

For the troopers, Nucleik has sped up some slower aspects of the job; an intelligence summary that once took Gregorczyk two hours, for example, now takes less than 10 minutes.

The more powerful component of Nucleik is its analytical tool. Social network analysis is a common counterinsurgency tool that has been used to identify and then dismantle the far-reaching elements of a terrorist network. Its application for crime-fighting, however, is relatively new. Moreover, Carley pointed out that rarely does one analytical tool perform all three necessary tasks well.


It certainly helped that Gregorczyk and his colleagues had used such tools from their own time in the military.

“Our team is unique in the fact that we have the expertise to do social network analysis,” Gregorczyk said. “I’ve taught it at West Point and a lot of our special forces-trained guys have had experience with it.”

People are network designed, explained Gregorczyk. “Cops used to ask who is the boss? And they would draw a line from him to the next guy then the next guy. But that doesn’t tell you who is valuable to that network.”

But social network analysis allows police to make finer distinctions among these relationships. The algorithms used in Nucleik, for example, can help identify who has the most influence in a gang, or who transmits the most information in the fastest amount of time.

“We’re not going after everyone. We’re going after key players. For a real purpose: to surgically dismantle gangs,” said Gregorczyk.

After graduation this May, Crouch, Polega, and Mayr plan to develop the next iteration of the software and launch it with more police agencies. They are now working out of Harvard’s Innovation Lab for young companies. They also hope to raise seed funding this summer.

“Our vision for Nucleik is to scale the use of the product to the point where departments can effortlessly share information,” Crouch said. “One of the major problems with law enforcement is that most departments use distinct and incompatible systems, impeding the ability to share criminal information. We hope to fix that.”