StreetCred Software Uses Big Data to Slap Cuffs on Inefficiency
While tech-savvy police departments may be common staples of television shows such as "Law & Order," the reality of police work can sometimes be quite different. A case in point is the tracking of Class C misdemeanor warrants in the state of Texas, an activity that for many officers includes the arduous task of verifying, organizing and tracking down information about thousands of offenders.
Serving such warrants—and collecting the fines owed by the targets—not only makes police agencies money, but also saves the money it costs taxpayers to incarcerate those unable to pay the fines. Enter StreetCred Software, a new company that founders Nick Selby and David Henderson said is working to put the cuffs on an issue they say is costing taxpayers significant amounts of money.
"What we're doing is prioritizing the warrants based on locate-ability and collectability," said Selby, a former analyst with The 451 Group and today a sworn police officer in Texas. "And to do that, we tap into a bunch of law-enforcement computer systems, we tap into a bunch of public records databases, we tap into a bunch of open-source information and open-source intelligence. … [We] aggregate and then we correlate."
By correlating information, the software is able to determine a ranking for each warrant based on the accuracy of the information the police have about the person and the likelihood the person will be able to pay if caught. The software also tracks the amounts owed by each person and notifies officers if the person has a history of violence. If an officer checks out the home or work address of the target and determines that it is wrong, he or she can mark it down in the program and change the ranking for the warrant.
The information the company factors in comes from a variety of places, ranging from government to law enforcement to commercial sources, to determine the score of a given warrant. Once that information has been pulled together, the software begins to query it.
"We articulate the questions that we ask through a series of first parallel and then sequential algorithms," said Selby. "Right now, we're asking about 120 sets of questions against the corpus."
The algorithms do not take into account race or gender. However, they do consider age.
"We know that like an 18-year-old college student is going to move a lot more frequently than say a 60-year-old person is," said Henderson, who has spent more than 15 years in law enforcement. "That's important for us, because when we deteriorate their score, one of the things that deteriorates it faster is if you are in an [age] bracket that is going to be more likely to try to move as opposed to a bracket [in which] somebody's more likely to stay."
The purpose of every question, Selby said, is to throw data away.
"Our process is, we're not looking for good warrants, we're looking for unservable ones," he said. "It does us no good if we know exactly where somebody is and they don't have any money. It does us no good if somebody's wealthy, and we can't find them. So we want to throw those warrants away. … The question we answer as a company is, 'If I've got eight hours to spend in a police car, who should I want to go after today to maximize my chance of bringing in the most money for our agency?'"
In Texas, Midland County Sheriff Gary Painter told eWEEK that his department believes the program will help keep officers up-to-date with warrant information and make the process of collecting fines more efficient.
"It costs me $68.31 a day to house a prisoner," he said. "And if I can keep them out of jail, I'm saving taxpayers' money. If I can get them to pay, then we're making money and keeping them out of jail. That's where we need to be."