On 6 August, peaceful protests over the police shooting of a local man in London's Tottenham district exploded into full-blown riots. During four days of assaults, arson, and looting, some of London's thousands of closed-circuit TV cameras captured video of the violence.
In the aftermath of the unrest, police officials began poring over footage in an attempt to identify suspected rioters. They even employed a facial recognition system designed for use during the 2012 Olympic Games, in London. But they found traditional investigative techniques to be much more fruitful than software, in part because many of the rioters had obscured their faces with hoods or bandannas, and because other factors such as poor lighting made it difficult to identify people.
The presence of cameras was clearly not a deterrent to the London rioters. But will technological advances that seem poised to eliminate anonymity prevent civil unrest in the future?
"I don't think we'll ever be able to predict the behavior of crowds, because they're notoriously unpredictable," says James Orwell, of the computing and information systems department at London's Kingston University. Still, Orwell, who has been developing recognition and tracking systems, says that the proliferation of cameras and technologies that help authorities make more efficient use of the increasing volume of video may make an individual think twice.
Researchers working with Orwell and his colleague Sergio Velastin have developed software that allows for the automatic detection of objects or events in video footage, creating continuity between cameras that lets the police track someone's movements. The technology also makes it possible to search video for specific actions, such as people running or cars pulling over to a curb. "It can search content-based metadata such as 'guy in the red tracksuit,' to see if there is a better shot that lets his face be compared with those in online databases," explains Orwell.
Orwell's system can plow through video footage to retrace a person's movements, maybe even to his home. Still, Orwell is quick to issue this caveat: "If they're really intent on evading identification, there's no easy way around that."
The tracking software relies on video footage that's available only to authorities. The system may seem Orwellian, but because of a convergence of technological trends we're near the point where almost anyone can use facial recognition to pick you out of a crowd. And what they can do next is worrisome.
"Your face is a veritable conduit between the off-line and online worlds, and you can't change it," says Alessandro Acquisti, a professor of information technology and public policy at Carnegie Mellon University, in Pittsburgh.
In research presented just prior to the London riots, Acquisti's team used a combination of off-the-shelf face recognition software, cloud computing, and data publicly available from social networks to uncover information about people just from their photographs. His group was able to identify students on campus caught on a webcam and then use algorithms developed earlier to probabilistically infer such information as date and place of birth, as well as Social Security number and credit score.
"As photos become scannable and matchable against online databases, it is creating a form of identification that the person being identified doesn't control," says Marc Rotenberg, executive director of the Electronic Privacy Information Center.
This is not necessarily bad, Rotenberg says, pointing to the potential for identifying missing children or fugitives. But he is quick to turn to the flip side of the coin—a scenario in which people at a public demonstration expressing unpopular views are identified and subsequently harassed.
"I don't think the end of privacy is upon us," Rotenberg says. But he is concerned that as the ability to remain anonymous disappears, the rules governing interpersonal conduct and commerce may not change quickly enough to keep people from being exploited.