Technocreep
Page 3
Regular cameras also played a role in this investigation, as agents pored over masses of amateur cell phone video and surveillance camera footage. The cameras that yielded the best pictures were the ones mounted on the Lord & Taylor store and the Forum Restaurant at 755 Boylston Street in downtown Boston.26
I walked that very stretch of Boylston Street a few months earlier. While those cameras were not hidden, they certainly would never catch your attention. Yet they provided vital evidence. After sifting through all the images, authorities published photos of suspects they dubbed “Black Hat” and “White Hat,” asking for the public’s help.
The “go public” strategy worked, and Dzhokhar Tsarnev was soon apprehended alive. Despite this outcome, some academic researchers called the Boston Marathon bombing case “a missed opportunity for automated facial recognition to assist law enforcement in identifying suspects.”27 Joshua C. Klontz and Anil K. Jain of Michigan State University did an after-the-fact simulation using the Boston suspect photos and a database of one million mugshots released under Florida’s “sunshine” law. They stirred in photos of the Tsarnev brothers taken on various occasions such as after a boxing match in 2009. Using commercial facial recognition software, they had some success in matching them, including a “rank one” matchup between a bombing scene photo released by the FBI and a high school graduation photo of Tsarnev.
The researchers acknowledge that neither of the commercial facial recognition systems they tested is ready for routine deployment in law enforcement applications, largely because of issues with different poses, resolution, and factors as simple as wearing sunglasses. However, you can be sure that work on improving facial recognition for law enforcement is moving full speed ahead.
Soon after the Boston Marathon bombings, Joseph Schuldhaus, vice president of information technology Technology for Triple Five Group, which runs the sprawling West Edmonton Mall and as well as Minnesota’s Mall of America, suggested that video analysis is going to become even more important in fighting crime. “I think we’re going to see the further miniaturization of algorithms at the edge of the device,” he told IT World Canada editor-at-large Shane Schick, “and what I mean by that is when the video comes into the camera these algorithms are going to help law enforcement better process that information, much like when you use Shazam to identity a song.”28
Schuldhaus clearly believes that the public safety and security advantages of surveillance cameras outweigh the risks they pose to privacy. Others are not so sure; but this has not stopped cameras, both public and private, from proliferating around the world.
According to a report in Forbes, “In the United States, it is estimated that there are 30 million surveillance cameras, which create more than four billion hours of footage every week.”29 They are also sprouting a lot of intelligence and new functionality. Their images are processed, in real time, to highlight suspicious packages at airports, to discover people who go where they are not supposed to be, and, even, as proposed by some Japanese researchers, to catch kids smoking in the schoolyard.30
For many years, a conference called Computers, Freedom, and Privacy featured a post-conference tour of the host city’s surveillance cameras. I vividly remember going on the tour of San Francisco in 2004. We stopped after we found about a hundred cameras peering down at unsuspecting people in Union Square and other public venues.
Dedicated volunteers from the New York Civil Liberties Union walked around Manhattan in 1998 noting camera locations, producing what they called “a comprehensive map of all 2,397 surveillance cameras in Manhattan.” When they re-did the same study in 2005, they “found 4,176 cameras below Fourteenth Street, more than five times the 769 cameras counted in that area in 1998.”31
They are fighting an uphill and ultimately hopeless battle. Modern surveillance cameras can be tiny, totally wireless, solar powered, and cost a few dollars. Good luck spotting one of those pointing out of a window or hiding in the pore of a ceiling tile.
It is hard to deny that the presence of video cameras in public places has deterred some criminals and solved or prevented certain crimes. In one case, a bandit robbed a local wholesale club, and was caught on the surveillance camera. A simple scan through the membership records turned up a match for his photo, yielding his name and home address.
The heavy camera coverage of New York’s Times Square is credited with helping police thwart the plot to detonate a bomb there in 2010. However, just as in the Boston Marathon bombings, a tip from a citizen also played a major role. Once, while watching a camera pointed at Times Square, I saw an entire drug deal transpire in plain sight.32 I managed to capture several screen shots and use it as an example of people who obviously did not realize they were being watched. Or who did not care.
Do cameras really earn their keep as crime fighters? The best data on this comes from the United Kingdom, which has had extensive camera coverage for over a decade. The results are not as encouraging as camera advocates would like us to believe. A 2009 Scotland Yard report estimated that only one crime was solved per year per thousand cameras.33 The resulting bad press for cameras was met with claims in 2010 by the London Metropolitan Police that they were actually able to solve six crimes a day with camera evidence.34 Commentators scoffed that most of them were probably jaywalking.
A recent scientific review of various crime prevention techniques looked at thirty-six U.K. studies, ten in the U.S., and one in the Netherlands. These researchers found that “there is little evidence that the following reduce fear of crime: street lighting improvements, closed-circuit television (CCTV), multi-component environmental crime prevention programs, or regeneration programs.”35 Many securities cameras are, as Bruce Schneier famously puts it, “security theatre.”36
A number of motor vehicle registration operations now run an applicant’s photo through facial recognition to see if a driver’s license has already been issued to the owner of that face. In British Columbia, Canada, the government-run monopoly auto insurer, Insurance Corporation of British Columbia (ICBC), uses facial features including the distance between the eyes as well as cheekbone geometry to root out fraud. Their photo database is of great interest to law enforcement, but there are some thorny privacy issues there.
On June 15th, 2011, downtown Vancouver was engulfed in riots after the home team lost the final game in the Stanley Cup hockey series. People were stabbed, police officers were injured, and there was extensive property damage. Digital devices helped to feed the violence, as rioters reacted to the presence of media and personal cameras. However, digital photos also played a key role in tracking down the offenders.37
British Columbia’s Information and Privacy Commissioner, Elizabeth Denham, had to decide if using the ICBC’s motor vehicles registration database to try to identify offenders would violate the province’s privacy laws. She ruled that it was acceptable for the police to provide candidate images to ICBC for possible matching. However, a court order would then be required for the matched-up results to be revealed to the police.38
Police in Vancouver also put out an appeal to the public to assist in the massive post-riot investigation. They set up a special website, riot2011.vpd.ca, with photos of “people who are alleged to have committed criminal offences.” The public was offered a simple “click and identify” system to provide information.
Did it work? In July 2013, two years after the riot, the Vancouver Police Department (VPD) announced that they had recommended charges against 352 alleged rioters for 1,204 offenses. The accused were as young as fourteen years old. As an example, the report describes three high school friends from Victoria, British Columbia, who “were captured on video committing multiple crimes throughout the night, including break-ins to four separate businesses.” VPD Superintendent Dean Robinson says they are not yet finished hunting down suspects, and “those rioters out there that believe they can wait us out and hide with anonymity, we will find you and you will be brought to justice.”39
; There are a number of reasons why the Vancouver Police Department turned to the public for help. One is just good public relations—hooligans trashing the city’s downtown does not sit well with most law-abiding citizens. Also, some of the demonstrators had no criminal record, and may have been too young to be in the motor vehicles system.
While the surveillance camera and cell phone videos used by the VPD were certainly helpful, they are nowhere near the state-of-the-art in surveillance camera technology. A remarkable photo of the crowd on Georgia Street taken a few hours before the riot was posted by a company called Active Computer Services. It is actually a composite image of 216 high-resolution photos stitched together, and it reveals an uncanny level of detail.40 You can zoom in from the massive scene to identify individual faces with ease. Active Computer Services has a particularly telling motto on their home page (“I spy with my little eye … ”) and they tout the “forensic science” applications of their technology.
According to Charlie Savage in the New York Times, scanning for a wanted face in a crowd is still a tough computer science problem.41 However, Savage writes, progress is being made: the U.S. government–backed Biometric Optical Surveillance System (BOSS) works with two cameras, equipped “with infrared and distance sensors. They take pictures of the same subject from slightly different angles. A computer then processes the images into a ‘3-D signature’ built from data like the ratios between various points on someone’s face to be compared against data about faces stored in a watch-list database.”
The Department of Homeland Security ran a test of BOSS in September 2013, using it to scan about six thousand fans attending a hockey game in Kennewick, WA. The faces of twenty volunteers were placed in a database. The challenge was to find them among the hockey fans, at a distance of fifty to one hundred meters, quickly enough so that if any were terrorists they could be located and intercepted. The results have not yet been disclosed.
Several commentators have noted that this type of surveillance system is often launched for crime-fighting or anti-terrorism purposes, but people quickly find other uses for it, including commercial ones. The day is not far away when the kid selling soft drinks at a stadium may pass you a note that says your car’s lights are on, having linked your face in the crowd to your license plate. They might even figure out a way to charge you for that service.
While this BOSS technology is not yet operational, experts say it will be deployed within five years. Privacy advocates suggest that we need to make rules now about how it can be used in the future, or we will simply default to ubiquitous surveillance.
The FBI, the Department of Homeland Security, and the U.S. State Department are very enthusiastic about facial recognition. According to Brian Merchant, writing for Motherboard, “the Department of State currently runs one of the largest facial recognition operations in the world. It uses a database of 75 million photos or so to cross-check visa applications.”42
While we hear a lot about the prevalence of cameras in the U.S. and the U.K., another country is on track to become the world leader in video surveillance. China is already estimated to have one surveillance camera for every forty-three of its citizens.43 I have been taken into a secret monitoring center in a major Chinese city where operators watch a gigantic wall of monitors covering every major intersection. The Chinese have a significant home-grown surveillance camera industry, ironically boosted by the United States, which slapped export restrictions on surveillance technology in 1989. That fueled China’s own research and development in this field.
Private use of facial recognition technology is also growing daily. The contours of this expansion were neatly summarized by Motherboard’s Jordan Keenan, who wrote, “If you use social media, have a driver’s license, shop in stores, and walk in public, chances are good that your faceprint will soon be assigned to your identity, and eventually be used on a daily basis to build a profile of you at a level of detail you hoped would never be possible.”44 Improving facial recognition is also the reason you have to maintain such a stern expression for your passport and visa photos.
On August 8, 2000, a woman wearing a toque and dark sunglasses entered a Safeway store near downtown Calgary, Alberta, Canada, pushing a toddler in her shopping cart. She wrote a note addressed “To Whoever finds my son,” wheeled him into the cookie aisle, and simply left the store. According to media reports, the two-year-old kept saying “where is my Mommy?” but she was nowhere to be found.45
Police were called, and appealed for the public’s assistance on the television news. When the mother did not come forward, Alberta’s Minister of Children’s Services ordered publication of this photo, shown here at reduced quality to preserve privacy.
Figure 5. Woman in Safeway with baby. Government of Alberta.
The baby’s mother was soon tracked down in the State of Washington. But how was Safeway able to supply that picture? I visited the store and found the inconspicuous camera posted over the entrance. Sure enough, everyone who entered was being captured on video.
Back in 2000, this was a shocking discovery for me. It seemed unnecessary for a grocery store to capture the arrival and departure of every customer on video. What else were they watching? Now, it is hard to imagine an urban space that is not within the reach of a surveillance camera. It’s not just that they’re capturing your image: it’s what might they might do with it, now and in the future.
Just what are all those surveillance cameras doing when they are not taking pictures of suspected terrorists, shoplifters, or mothers who abandon their kids? They are watching us, creating a potentially eternal archive of everything we do. The same technology that allows law enforcement to zoom in on bad guys can impinge on the privacy of law-abiding citizens in some very creepy ways.
The one-way nature of surveillance cameras is one of their most unsettling features. Aside from the occasional blinking light, they tell us nothing. We tell them everything. One way to level this playing field is to wear our own cameras. University of Toronto Professor Steve Mann coined the term “sous-veillance” to describe the countermeasure of wearing cameras to record our own version of how things happen.
One of the first famous uses of this approach was the 1991 videotaping of the beating of African-American construction worker Rodney King by the Los Angeles police. The police officers were acquitted, despite compelling video evidence against them, sparking the 1992 riots in that city.
Now, dashboard cameras are commonplace, at least in the United States, and definitely in Russia, where they are almost mandatory to survive the country’s traffic and scam artists who stage fake accidents.46 Video evidence gives you an edge in many situations, and some people are already logging their lives as an offbeat kind of hobby. I spoke to one of these lifeloggers, and he estimates the cost of recording his every moment in audio and video at about one dollar a day for storage media. His cost in terms of relaxed social interaction, however, might be much greater. The possibility of recording everything you see, hear, smell, and touch was also the subject of a research project called Lifelog, funded by the Defense Advanced Research Projects Agency in 2003 but abruptly canceled in 2004 after privacy groups voiced objections.47 According to many experts, the program has continued, at least in spirit, both inside and outside the U.S. government.48
The movie Déjà Vu (2006) envisioned a world in which satellites look down at people and peer inside their homes with laser imaging, using computer reconstruction to replay a terrorist attack and then travel back in time to avert it. Camera technology is definitely moving in the creepy direction suggested by the movie. Scientists at MIT have announced the ability to see through solid walls to an accuracy of ten centimeters. PhD student Fadil Adib, speaking on a Network World video, says “we’re doing localization through a wall, without requiring you to hold any transmitter or receiver, simply by using reflections off the human body.”49
The researchers gave their project a benign name, “Kinect of the Future,” suggesting it might sim
ply be the next evolution of Microsoft’s popular gaming device. However, a system that can peer through walls will have applications far beyond video gaming. It probably would have been greeted much differently if they had called it the “Anne Frank Finder.” That may indeed be much closer to how this technology will really be used.
Even before the Boston Marathon bombings, the U.S. Air Force contracted with the 3D biometric imaging firm Photon-X for a new kind of surveillance camera. By using a combination of infrared and visible light, and by indexing muscle movements that are unique to each individual, the company claims it can produce a unique “bio-signature” for a person and then silently track them.50
The company is also promoting something they call the Spatial Phase Imaging Technique (yielding an unfortunate acronym, SPIT), which purports to read your fingerprints at a distance of up to ten feet, with “longer distances being developed.” They also claim they can “passively capture 3D geometry for skin, hair, eyes, teeth, clothing, and anything else that is in frame, with no special preparation of the subject.”51
While surveillance cameras do not yet follow us everywhere, we do a pretty good job of filling in the gaps with our own cameras. We snap billions of photos and many of them end up on Facebook and photo-sharing websites. By putting our real name next to photos, we provide the fodder for all kinds of nefarious data mining.
TV studio cameras have large red “tally lights” to show the anchorperson where to look, but far too many unwitting TV presenters have been embarrassed by their “off camera” comments that made it to air, so they don’t really trust the lights.
While their lenses may be almost invisible, laptop computers and smartphones are equally risky. Unless you douse it in a glass of water, as a friend of mine did when he learned his smartphones was infected with some nasty malware, there is a decent chance that your camera can be hijacked by a hacker.52