...each member has taken a battery of tests, administered by scientists, to establish this uncanny credential. Glancing at a pixelated face in a low-resolution screen grab, super-recognizers can identify a crook with whom they had a chance encounter years earlier, or whom they recognize from a mug shot.Read more here.
...Being a super-recognizer can be draining: there is no off switch for this mysterious capability. It is not uncommon for a super-recognizer, out on the town with friends, to bolt off after spotting someone with an outstanding warrant. Before joining the unit, James Rabbett, a young detective with a hipster beard, won an award for making two hundred arrests in a year. Rabbett, who has a cocksure manner, displays powers of recognition that are exceptional even by the unit’s standards. He told me that, since joining the team full time, six months ago, he has made nearly six hundred identifications. Rabbett sometimes makes as many as five arrests a week while off-duty. This is fantastic for racking up stats, if less than conducive to a fulfilling social life. “It’s become a bit of a burden,” he allowed. Once, Rabbett was off-duty in Finsbury Park when he recognized a jewel thief and chased him down. A year earlier, he had glanced at the man’s image on a wanted poster. “You’re nicked on suspicion of stealing a bag a year ago!” Rabbett said. The thief, we may fairly assume, was surprised. (He subsequently pleaded guilty.)
...Faces are special,” brain scientists like to say. Days after birth, an infant can distinguish its mother’s face from those of other women. Babies are more reliably engaged by a sketch of a face than they are by other images. Though human faces are quite similar in their basic composition, most of us can differentiate effortlessly among them. A face is a codex of social information: it can often tell us, at a glance, someone’s age, gender, racial background, mood. Using f.M.R.I. scans, researchers have discovered that certain areas of the brain are hardwired for processing faces.
...The journalist Clive Thompson, in his 2013 book, “Smarter Than You Think: How Technology Is Changing Our Minds for the Better,” suggests that for some tasks the most formidable tool is neither a human brain nor a computer but both working in tandem—a “centaur” strategy. The super-recognizers follow this approach. To navigate the Met’s database, they rely on a proprietary software program to guide them. Each time an image is entered into the system, a human tags it with metadata; if the super-recognizers are searching for a white male between the ages of forty and fifty who is known for nonresidential burglaries in East London, they can select for those criteria.
...Computers may not be great at spotting faces, but logo-recognition algorithms are very effective. And it turns out that many criminals not only commit the same crimes again and again; they do so wearing the same outfits. (The super-recognizers find this hilarious, and joke about “crime clothes.”) Cameras had captured a burglar committing several break-ins wearing the same Everlast sweatshirt, which happened to be a perfect match for the one that he had worn in a mug shot taken after a previous arrest. Another thief was devoted to a newsboy cap. As I scrolled through the database, even my untrained eye picked out multiple shots of the thief in his trademark hat. There are other tells. When Mick Neville gives talks, he shows a slide featuring “the cross-eyed Peckham arsonist,” and gently suggests that the next time the man feels like setting a fire he might want to consider wearing shades.
Four out of five people identified by the super-recognizers are “habitual” repeat offenders. A member of the unit, Andy Eyles, explained, “If you do something and get away with it, you’re not going to branch out.” One wall of the office features a list of “Prolific Unknowns,” with thumbnail images of individuals who have been recorded repeatedly committing the same offense but have not yet been identified by name. If you can find a “linked series” of images, Neville explained, and then identify the perpetrator, you’ve cracked a dozen cases at once.
Recently, a handsome, olive-skinned man of about forty showed up in multiple clips. He would walk into a jewelry store or a boutique and engage a saleswoman in conversation, posing as a wealthy customer in search of a gift for his girlfriend. When the saleswoman turned her back, the man used sleight of hand to pocket a bracelet or shove a cashmere sweater down his pants.
As the super-recognizers scoured the database for images of clean-cut white male shoplifters, they were startled by how many pictures of the man they had accumulated. “It just grew and grew,” Porritt said. Eventually, they had images of nearly forty thefts: the man had stolen clothes, antiques, designer sunglasses, luxury cosmetics. He’d made off with more than a hundred thousand dollars’ worth of merchandise. After circulating his image in the press, the team received a tip that his first name was Austin. When they consulted the database of booking photos, they found their man: one Austin Caballero, who had been arrested for shoplifting in 1993.
But they could not find a current address for Caballero, and had to wait for him to strike again. On New Year’s Day, 2015, Porritt received a call at home. In the predawn hours, a man had taken a taxi to buy cigarettes; he refused to pay the driver and ended up assaulting him. When he was arrested, the man gave the name Jack Donaghy. (Apparently, he was a “30 Rock” fan.) It was chaotic at the station, and the officers were about to let “Jack Donaghy” go when somebody thought to fingerprint him. It was Caballero. Two of Porritt’s colleagues went to the station and presented Caballero with a collage of photographs documenting thirty-six incidents of shoplifting. He pleaded guilty to all of them.
...Talking with the super-recognizers, I found myself wanting someone like Alison Young stationed at the airport. Shouldn’t anyone applying for a job that involves face-matching—from a guard at a federal building to a sentry outside a nuclear plant—take the Cambridge Face Memory Test? Josh Davis and other scholars who study super-recognizers believe that the gift may be roughly as common as face blindness, which would mean that one in fifty people has a brain that is especially well suited to such jobs.
...“In the Met, we solve about two thousand cases a year with fingerprints and another two thousand with DNA,” he told me. “This year, we solved twenty-five hundred crimes using imagery, and it’s about ten times cheaper than those methods.”
...The super-recognizers acknowledge that they are not infallible, which is why they have the peer-review process as a safeguard. According to statistics that they freely share with the press, seventy-three per cent of their identifications have led to criminal charges; many of these suspects, realizing that they have been caught in flagrante, plead guilty. But thirteen per cent of the unit’s identifications have been wrong. Sometimes the super-recognizers have identified someone as the culprit of a crime only to discover that the suspect was in jail when the incident took place. Porritt emphasized that suspects very seldom go to prison solely on the basis of their identifications. “It’s never our word alone that puts someone away,” he said. “What we do, by identifying suspects, is help direct the investigation.”
...if legal authorities accept that there is a spectrum of facial-recognition ability—and that this skill can be evaluated with simple tests—it could be a boon to defense attorneys, and could inspire an overhaul of criminal procedures that involve such identifications. In recent decades, the status of eyewitness testimony has been undermined as scholarship has demonstrated that after stressful situations people are often mistaken about what they have seen. But what if, even on a good day, the people offering testimony have an impaired ability to recognize faces? When a police officer testifies that he saw a defendant fleeing the scene of a crime, should we not ask how that officer has fared on the Cambridge Face Memory Test? In July, a cop in Savannah, Georgia, confronted a young African-American named Patrick Mumford in a driveway and repeatedly Tased him, in the mistaken belief that he was another man, Michael Clay, whom the cop had a warrant to arrest. “They look a lot—a good bit alike,” the officer told a neighbor after realizing his mistake. “It’s not far off.”
In fact, Clay and Mumford don’t look alike. The officer could have made the mistake because he’s a racist, or because he was badly trained, or because, as a white man, he is better at recognizing white faces than brown ones. But wouldn’t it be helpful to know where he falls on the facial-recognition spectrum? If he’s demonstrably inept at recognizing faces, the public might wonder whether he should still be deployed, with a gun, on the street.
The police department of St. Petersburg, Florida, recently announced that it is working with psychologists at Dartmouth and Harvard to test the facial-recognition capabilities of its officers. Other departments are likely to follow.
Neville clearly shares this concern. He feels that he has unlocked a mystery of the human mind which could signal a revolution in policing. He is frustrated when he encounters skepticism, as if he were claiming to have discovered officers with telepathy or E.S.P., and he feels a nagging suspicion that had he written an algorithm instead the brass would be rushing to embrace it. “People don’t want to believe that humans could be better than a machine,” he told me. “And the sad truth in this wicked world we live in is that people don’t want to pay a human. They want to buy a machine.”
h/t Ann Althouse
No comments:
Post a Comment