Human Flesh Search Engines
by Scott Smith , 10/20/11
How would you like to see your face on a 20 foot LCD screen mounted on a truck, asking “have you seen this person?” If you had been unwise enough to throw your hand in with rioters in Birmingham, England during the UK’s outbreak of unrest in early August, you would stand a chance of having just that moment of infamy. Not content just to print faces of unknown offenders in the local paper, the West Midlands constabulary took things one technological step further and hired a van meant for digital advertising to flash pictures of its most wanted, captured on CCTV during the previous days’ rioting. Granted, the van (and its giant screens) had been offered to the police by its owners, but the jump from posters and print amplifies the real-time effect. Mobile screens that can travel a city shopping for witnesses are an interesting use of surveillance footage to track down unknown offenders. The riots themselves represented a hallmark in the application of CCTV and other kinds of imagery being used for surveillance, including liberal use of open social media by authorities to spot and identify alleged culprits in the unrest. Britain’s technology-enabled public surveillance culture had been nearly 20 years in the making, kicking off in earnest in the mid-1990s following the IRA’s bombing campaigns in London, initiating the so-called Ring of Steel strategy to protect London’s inner perimeter in subsequent years. More sophisticated image recognition technology for car number plates and human faces got a boost when London’s congestion-charging system was put in place in 2003, a key part of which is thousands of cameras used to check vehicles entering the city against rolls of those who have paid the requisite fees. According to some analysts, that system studies and tracks faces as they move through the area as well.
Estimates now put the number of CCTV cameras in the UK alone at between 1.85 and 2.25 million, making it one of the most densely video-monitored countries on earth. One estimate suggests the average Briton is viewed on a CCTV camera some 70 times per day. In a 2009 count, London’s Wandsworth borough, which saw a great deal of action during the recent riots, itself contained more CCTV cameras than Dublin, San Francisco, Johannesburg and Boston combined. Some analysts believe the recession has taken a bite out of private companies and local councils’ ability to pay for “Surveillance Britain,” but the 2012 Olympic Games are expected to give video surveillance and facial recognition technology a fillip. Although still a year off, it is driving deployment even now in strategic areas of London and ports of entry.
Identifying alleged troublemakers is no longer just the job of the faceless men and women in dark operations rooms. The riots also made facial recognition more of a peer-to-peer activity, with online groups formed to weed through thousands of images to put names to allegedly offending faces. Members of one group even discussed collaboratively tapping an existing online service to facilitate use of Facebook photos to find rioters. This so-called crowdsourcing of facial recognition wasn’t new to the riots or the UK. Chinese citizens, for example, have taken it upon themselves to use the Internet, through China’s eerily named “Human Flesh Search Engine,” to highlight, locate, shame and even intimidate those deemed to have offended civic sensibilities. For Canadians upset at hockey riots, social groups have taken identification and occasionally retribution into their own hands. But with the riots happening in and among such a well-wired and social mediaconnected milieu, it wasn't just the government that was interested in harnessing social transparency to law enforcement ends.
Like many other technologies, high-powered facial recognition has trickled down to the average person's fingertips in recent years. Facebook’s apologies for the rollout of automatic facial tagging of photos was just a temporary setback for the company, a tactical stutter-step in advance of a longer-term push into this controversial space.
Likewise, some believe Google’s strong stance on its “real names” policy requiring users of its Facebook-rival social network to use their legal names instead of nicknames or pseudonyms - is motivated by its desire to put not just data but faces with these names. Apple, which has kept a lower profile over facial recognition (enabled in its iPhoto application), has now integrated recently acquired technology to allow developers on its platform to apply facial recognition even more effectively in its upcoming release of iOS.
Google may have more to gain than its rivals, as its business is already so heavily built on analysis and monetization of data it has already captured. Despite recent words from chairman Eric Schmidt cautioning governments against using it too widely, the search giant also recently got its hands on facial recognition software from a Carnegie Mellon spin-off which correlates faces with public databases of information - including any publicly disclosed data such as addresses, phone numbers, or fragments of social security numbers. With a US$12 billion business in advertising sales alone, based substantially on its data-crunching capabilities, the Google is almost compelled to push ahead in what amounts to an analytics arms race, of which consumer profile assembly is a big part.
Around the edges of these intelligent face-spotting networks, analog countermeasures such as hoodies, hats and burkhas have been deployed for as long as people have been concerned about their privacy. More inventive bottom-up responses are now in development. Jan Chipchase of frog design noted in his recent survey of the issue that this is already happening with online reputation “cleaning”. Soon, there will surely will be a brisk business in scrubbing facial trails from the web. If current market dynamics are any gauge, there will even be visual SEO experts making sure the face of the hot celebrity, local estate agent or besuited business guru is sufficiently robot readable.
Many more will want to stay out of the gaze of technology, and will take steps accordingly. Enterprising designers have already created several versions of antipaparazzi sunglasses, which, while a novelty a few years ago when first shown, may have more real world uses for non-celebrities today. One designer recently showed off makeup schemes that confuse facial recognition technology’s reliance on facial geometry. Measures are already percolating on the fringes, from plastic surgery to evolved “black bloc” tactics such as clothing that suppresses heat signature. Sooner than we think, the increased scale of events will escalate both the watching and the desire to be unwatched. With governments, private companies and peers alike wielding the means to capture and identify facial likenesses, we will increasingly be pushed to choose whether we want to optimize or obfuscate, in the same way we now opt in or out of marketing databases. The public relationship with the eye in the sky, the hand-held camera and the remote database will drive more and more of our social, political, economic and technological decisions as individuals.