Nationwide, police departments are using predictive analytics to strategically deploy officers around their respective cities. A mobile app for citizens in Baltimore called SpotAgent, for instance, predicts whether a given parking spot might be safe from roaming parking enforcement officers at a given hour. And since 2006, Memphis, Tenn, has used the technology to evaluate incident patterns citywide and forecast criminal hot spots.
But security firm Raytheon is taking predictive analytics to the next level. While current technology, such as IBM’s predictive analytics solution, uses statistics and analytical data to predict criminal activity, Raytheon’s Rapid Information Overlay Technology (RIOT) software monitors public information on social media websites to predict what people will do next.
In early February, a promotional video explaining RIOT was made public, frightening most people who watched it – 81 percent of users on Newser.com tagged the news as “scary.” The video, which has since been removed from the Web, shows how “Nick,” a Raytheon employee, could be tracked using his social media history and how his future actions could be estimated based on past behavior. "So if you ever did want to try to get hold of Nick,” the video narrator says, “or maybe get hold of his laptop, you might want to visit the gym at 6 a.m. on a Monday."
Predictive policing is already in effect nationwide, but could adding such software to law enforcement's arsenal of tools enhance officers’ ability to do their jobs? And is such a technology truly acceptable in our society?
Though Raytheon doesn’t yet have a contract with any government entities for the software, the company has shown the software to some public-sector agencies – and according to the Electronic Privacy Information Center (EPIC), government’s interest in the software is typical.
However, EPIC Open Government Program Director Ginger McCall says the software likely isn't that useful. “It doesn't seem to me that there's that much useful application to most of this data,” she said. “It's one more step along this path we've been going down since Sept. 11, 2001, where the trend is to sweep in large amounts of information, to retain that information, even though most of the information is completely irrelevant.”
In rare circumstances, she said, the software might be useful if the government was using it to track a specific person. When Chris Dorner, a former officer for the Los Angeles Police Department and ex-U.S. Navy reservist, carried out a series of shooting attacks on police officers and their families from Feb. 3 through Feb. 12, 2013 – which left four people dead and three police officers wounded – he was the focus of one of the largest manhunts in LAPD history, spanning two U.S. states and Mexico, according to The Daily Mail.
It took nine days to find Dorner, who ultimately died during a stand-off with police at cabin in California’s San Bernardino Mountains. But what if RIOT software was used to track Dorner’s social media activity? According to a report on beforeitsnews.com, he posted to Facebook during the manhunt, but could his previous activity have given law enforcement some idea as to where he might hide?
Not necessarily. There's currently no evidence that the government is interested in doing anything with social media other than storing massive amounts of data for later analysis, McCall said, and using a few data points to find people who don't want to be found isn't very likely anyway. “It is extremely unlikely that an al-Qaida terrorist is going to be checking in on Foursquare,” she said. And Dorner didn’t check into his locations on Facebook, either. The idea, however, is to look for patterns of past behavior to predict future behavior.
The thing the public should be concerned about, McCall said, is the general loss of control people have over their data and associated privacy. Even privacy-conscious social media users or non-users can have their information shared by their friends, and once information is out, it's almost impossible to prevent other people from using it. One thing the government isn't allowed to do, McCall said, is collect, maintain or even monitor “protected speech,” something the Department of Homeland Security has testified before Congress saying it wouldn't do. So while information like when “Nick” checks in to his gym might be available to the government (if they're interested), what Nick writes on his Facebook wall about his new boat is supposed to be off-limits.
There are several important factors when it comes to data privacy. What kind of information is being collected? Who is the information being shared with? What is it being used for and how long is it being retained? These are all questions that create uncertainty anda gray area when it comes to data privacy.
But the gray area only exists because of the way government chooses to form laws around technology, said Andrew Walls, an analyst for Gartner specializing in information security. When it comes to the public's expectations regarding privacy, there is a consensus, he said: Everyone agrees that personal data belongs to the person it relates to, especially if it's password protected or the person is maintaining some other form of access control over it. Despite this, Walls said, there continues to be an imbalance between cultural expectations and what the government deems as suitable in the course of law enforcement.
In the book Privacy on the Line, by American cryptographer Whitfield Diffie, the point is made that whenever a surveillance capability is made available to the U.S. government, it is used to excess. Whether it's body scanners at the airport, thermal imaging scanners in police cars or predictive social media analytics, the government tends to betray the trust of the public in pursuit of security.
And most people even agree that security is needed, Walls said, but finding a way to make sure technology isn't abused is a problem that rarely gets solved. “Every law enforcement group I've dealt with, and I've dealt with a lot of them, make significant efforts in this regard,” Walls said. “They don't want policemen looking up the details of what all their neighbors are doing.” Because every time there's something in the news about a 90-year-old woman being patted down at the airport, or police storming an innocent family's home with guns drawn because they thought it was a marijuana farm, it creates cynicism in the public. “Trust is very easy to lose and very hard to build,” Walls said.
Lack of trust and worry about the permanence of data is why apps like Snapchat exist. “Why do people want to send pictures to each other that then automatically destroy themselves?” Walls asked. Because people are fighting back against surveillance and security. People are aware that the data they create can be monitored, collected and maintained, so solutions to make the data ephemeral are filling that need.
And “need” is the operative word, Walls said, because social media is no longer a luxury. Someone applying for a job will have a hard time getting hired, he said, if they have no online presence – it looks suspicious or unusual to employers. Because social media is becoming a requirement for many people, there is a big concern about how the government is allowed to use software like Raytheon's RIOT. “It's good for law enforcement,” Walls said, “but the potential for abuse is also enormous.”
Social media expert and Gartner analyst Brian Blau agreed that the potential impact of this type of software is big. “If you can tap into enough signals from individual use of technology, you can really understand what people are doing,” Blau said. “I talk to clients about … how we're all being tracked today and the privacy issues. It's not only the fact that you're checking in somewhere or that there's GPS, but you combine that with contextual data such as schedule or events or even the words that you say.” There are a lot of signals that people put online that, if combined, he said can create a fairly thorough profile.
Many elements of a person's life, such as personal beliefs and relationships, can be determined through signals generated from a Facebook account, while another set of data can be determined from a person's LinkedIn profile, and even more data could be brought in from a niche or hobby website or through a person's use of apps. People tend to try and keep their various social circles separate, but Blau said that putting everything together can in some cases create a rather accurate picture of who a person is, what they’re like, what they do, where they go and what they think.
While the potential for abuse exists, Blau noted that there are also constructive uses of predictive analytics. “Algorithms can be very sophisticated and you can make all kinds of inferences,” he said. “One of the cool things about social these days is that social analytics and mobile app analytics are giving off lots of data through the use of these technologies.”
And people are using these for good, Blau added. “They're using them to help people get better value out of social networks and social apps, and the government could potentially do something with our data. They can go out and buy that data like any other company can. And maybe that's what Raytheon is doing.”
Photo: The street in front of the west Los Angeles police station has yellow tape prohibiting the parking of cars Friday, Feb. 8, 2013, in response to threats by former LAPD officer Chris Dorner. (AP Photo/Reed Saxon)