A successful thief sets his own rules and the best ones live by them. These were some of Jimmy McNulty’s: target luxury stores only, dress as smartly as the wealthiest customer, engage and charm the salespeople if approached. Never rush, never panic, and always trust in your powers of sleight of hand.
Here he is at 12.59pm on 28 September 2013, ringing the bell of the Leica Store in Mayfair, where cameras sell for thousands of pounds. He is 40 years old, with short, dark hair and of athletic build – McNulty is the name given to him by a Metropolitan Police detective who saw a resemblance to Dominic West’s character in the television series The Wire. He wears a pink dress shirt, a dark cardigan and jacket, smart shoes. Under his arm is a wad of papers. McNulty picks up a camera, and then a pair of binoculars, carefully appraising them. Two store assistants stand a few metres away. When they turn their backs, he slips the camera inside his jacket. He asks to be let out and casually strolls away.
This is McNulty again at 4.39pm on 18 October 2014, sitting at a table in Buy Fine Diamonds, a retailer in Hatton Garden, London’s jewellery district. A salesman lays out a selection of bracelets. McNulty uses a jeweller’s loupe to examine several items. When the salesman stands up to retrieve a piece from a window display, McNulty strikes, pincering a bracelet and dropping it into his pocket. The salesman returns and from under his nose McNulty lifts another bracelet into his palm and closes his fist.
Now here he is at 6.27pm on 13 May 2015 at the Hackett clothing store on Regent Street, Piccadilly. After slipping a few leather accessories into his sleeveless jacket, McNulty picks up a pair of shoes from a display, walks to an empty sales counter and stuffs them down the front of his trousers.
That was his favourite place to hide stolen goods: merino wool jumpers, cashmere scarves, fancy shirts and wallets disappeared below his belt. He hit Salvatore Ferragamo on Sloane Street, Smythson on New Bond Street, Aquascutum on the Brompton Road, Alfred Dunhill on Jermyn Street, Ede & Ravenscroft – the oldest tailor in London – on Chancery Lane. Boutiques in Islington, galleries on the Portobello Road, the Space NK and Jo Malone cosmetics shops. At Linda Farrow, an eyewear shop in Mayfair, he slipped a pair of sunglasses into his jacket and then, as if it were a game, asked the assistant for a business card.
McNulty had a rule for closed-circuit television cameras, too: ignore them. More than 400,000 CCTV cameras watch over London and most upmarket shops have them. But McNulty knew they were used mostly as deterrents. Even if the footage was sent to the police, at best he’d be fingered for a single crime, he thought. Unless, of course, they recognised him as a serial offender and found out his real name. What were the chances of that?
Since the 19th century, doctors have known that some patients who suffer brain trauma lose the ability to recognise faces, a condition known as acquired prosopagnosia (from the Greek prosopon, “face”, and agnosia, “not knowing”). In the 1970s scientists discovered that a congenital form of the disorder affects a much wider segment of the population – ordinary functioning people who have never experienced head injuries and have perfect vision.
Studies suggest that two out of every 100 people have developmental prosopagnosia, meaning they have great difficulty recognising faces, sometimes even their own in the mirror. To identify someone familiar, a face-blind person relies on clues such as voice, gait, posture or unusual facial characteristics.
Among the best-known prosopagnosics was the late doctor and author Oliver Sacks, who became aware of his bewildering predicament as a schoolboy in London. He learned to pick out his best friends, Eric Korn and Jonathan Miller, by their specific features. “Eric had heavy eyebrows and thick spectacles, and Jonathan was tall and gangly, with a mop of red hair,” Sacks wrote in the New Yorker. When he looked at old photographs a decade after leaving school, Sacks could not identify a single classmate. Stephen Fry and Jane Goodall are other well-known sufferers of the disorder, which is associated with lesions in a part of the brain known as the fusiform gyrus.
In 2009 a trio of researchers led by Richard Russell published the results of their study, which aimed to determine if there was a third group of people when it came to face recognition, whose problem (or rather talent) was that they struggled to forget a face. Russell, a psychologist who was then based at Harvard, tested four people claiming to have superior face recognition abilities, including a 26-year-old female student who told him: “It doesn’t matter how many years pass. If I’ve seen your face before, I will be able to recall it.” Russell set his subjects and a larger control group two tasks, involving famous faces and unfamiliar faces. In both, the test group performed “far above average”, leading Russell to coin the term “super-recognisers”. “In both face recognition and face perception, the super-recognisers are about as good as many developmental prosopagnosics are bad,” he and his colleagues wrote.
Around the same time, Detective Chief Inspector Mick Neville of the London Metropolitan Police was reaching his own conclusions about people with an exceptional ability to recognise faces. In 2007, Neville had set up a unit to collate and circulate images of unidentified criminals captured on CCTV. Officers were asked to check the Met’s “Caught on Camera” notices to see if they knew any of the suspects. “It became apparent that some officers were much better than others,” Neville told me. “For example, if I received 100 names, some officers would have submitted ten or 15, while in the main they were one-off identifications.”
At first, Neville assumed that the prolific officers simply knew more criminals than the rest. Then he realised that it had more to do with their ability to remember faces: the best identifiers could spot a suspect they had never met merely after viewing a photograph of them.
In early 2011, he discussed his findings at a conference attended by Josh Davis, a psychologist at the University of Greenwich. For his PhD, Davis had studied the use of CCTV identification in court proceedings. “Most of my research had shown that people were not very good at face-matching,” Davis told me one recent morning when we met at a cafeteria on campus. “So I was suspicious of the police claims.”
He agreed to test the facial recognition skills of 20 officers who excelled at Caught on Camera identifications. To Davis’s surprise, most of them scored much better than the norm, and a few were exceptional.
That August, the London riots broke out. Met officers trawled through tens of thousands of hours of CCTV footage, identifying 609 suspects responsible for looting, arson and other criminal acts. One officer, PC Gary Collins, made 180 identifications, including that of one of the most high-profile suspects, who had thrown petrol bombs at police and set cars on fire. During the riots, the man covered his mouth and nose with a bandana and pulled a beanie low over his forehead. Collins recognised him as a criminal whom he had last seen several years earlier. The man was convicted and sentenced to six years in prison.
Now convinced of the super-recogniser theory, Neville assembled a standby team of 150 officers who excelled at identification. Over the next few years, as McNulty was slipping jewels into his pockets and stuffing luxury shoes into his trousers, the officers were deployed in high-profile investigations, including the Hillsborough inquest and the Alice Gross murder.
Gross, a 14-year-old girl from west London, went missing as she walked along a canal towpath in August 2014. The Met operation to find her was the biggest since the London 7 July 2005 bombings. But it was only after ten of Neville’s super-recognisers were brought in to the investigation that her body was discovered in the River Brent.
The team zeroed in on one of the suspects, a Latvian construction worker called Arnis Zalkalns, whose wife had reported him missing a few days after Gross disappeared. A CCTV clip showed Zalkalns cycling along the canal 12 minutes behind Gross. In footage from an off-licence later in the day, the officers recognised Zalkalns, who was buying a few Carlsbergs, and council cameras captured him cycling back to a particular spot on the River Brent at dusk. At his next sighting in a shop, later in the evening, he was wearing fresh clothes. The super-recognisers suspected that Zalkalns had changed because he had been back to the crime scene. They informed the officer in charge, who ordered a fresh search of the stretch of riverbank where they had seen Zalkalns – and Gross’s carefully concealed body was found.
On May 2015, the Super-Recogniser Unit was established at New Scotland Yard, the first – and still the only – dedicated team of its kind in the world. Initially it comprised four officers whose skills had been tested by Josh Davis and who were seconded from elsewhere in the force. (The unit now has six men and one woman.)
Detective Sergeant Eliot Porritt, who had worked on the Alice Gross murder, was the most senior recruit. A 36-year-old former plain-clothes officer from north London, Porritt had been largely unaware of his superior face recognition skills until a few years ago. “As a boy, I watched The Terminator and Aliens with my father, who worked for Billboard and Hollywood Reporter magazines. I now remember him being amazed when I noticed that an actor – Bill Paxton – was in both films, even though he looked different in each role,” Porritt told me. “But I didn’t think too much of it at the time. I assumed everybody saw what I saw.”
The main function of the super-recogniser officers was to attend large events, such as music concerts and the Notting Hill Carnival, and spot criminals there. In their downtime, they were tasked with trawling through the Met’s forensic image database, which holds more than 100,000 stills of unidentified suspects captured on CCTV camera or on mobile phones in London since 2011. Each picture is linked to an unsolved crime – in essence, a cold case – and is tagged with the date, location and type of offence, along with the suspect’s distinguishing features, such as race and hairstyle.
As they scrolled through the images, the officers first checked whether they recognised anyone from their time on the streets or previous Caught on Camera appeals. The next challenge was to link suspects involved in multiple crimes, using their powers of recall and recognition to match images – a process they called “snapping”.
“Basically we’re saying, ‘This guy and that guy in those two pictures are the same person – snap!’” Porritt told me when I visited the Super-Recogniser Unit one afternoon. “And you’ve got two strands to it: the people we already know and who we try to link with as many crimes as possible; and people who we don’t know but who we still link and then try to identify.”
It was difficult, painstaking work: the images were often grainy, the lighting poor and camera angles awkward. Furthermore, a criminal’s appearance could change over time. But if snapping led to an arrest it would be worth it: a person charged with multiple crimes was likely to be sent to prison rather than receiving a suspended sentence and being left free to reoffend.
In early August 2015 one of Porritt’s junior colleagues (who asked for his name to be withheld) was looking at CCTV images from the borough of Kensington and Chelsea, where he had worked as a beat cop. The officer noticed the same, smartly dressed thief – the man the team later nicknamed McNulty – in two stills taken in upmarket shops. Snap. Then another, and another – snap, snap. As he broadened his search to other affluent boroughs of London, the officer kept seeing the same face. He printed out the images of the serial shoplifter and tacked them to a wall in the office. He told me: “When I had 13 or 14 crimes, I said to Eliot, ‘There’s £35,000 worth of goods stolen by this guy. We need to do something.’”
They downloaded the CCTV clips from where the stills had been taken. McNulty’s hands were so fast that in some cases the officers had to slow the footage down to ascertain exactly when the theft had occurred.
“I hate using the words ‘talented’ or ‘good’ for a criminal, when they could be so many better things, like a street magician or a dextrous watchmaker,” said Porritt. “But when we watched him [McNulty], it was like: ‘That’s good.’”
Porritt sent out a media appeal and the London Evening Standard published a story about the “sleight-of-hand thief”. Through tip-offs, the police learned that the man’s first name was Austin. Porritt’s colleague keyed the name in to the Met’s custody image database, which stores mugshots of everyone arrested in the capital. There were 73 Austins. Number 14 looked familiar. “I think I’ve got him!” the officer shouted.
Jimmy McNulty was Austin Caballero, a born-and-bred Londoner. He was on the arrests database because he had been caught stealing an expensive rug in January 2015, but the police, unaware of his other crimes, had freed him on bail. He skipped his court hearing. There was a bail address but when Porritt and his colleague knocked on the door, there was no sign of the thief. Caballero was on the run.
By then, it was clear that the task of searching the database should be the primary focus of the Super-Recogniser Unit, not just something to do in quiet times. The team had identified, apprehended and charged dozens of suspects, from shoplifters to commercial burglars and perpetrators of assaults. (Three in every four of the unit’s completed cases have resulted in a suspect being charged in court – against fewer than one in five cases in the wider Met force.)
The super-recognisers had also assisted other units struggling to close cases. Police in north London had obtained CCTV images of a man accused of sexually assaulting women on buses but were unable to identify him. Porritt and his colleague Alison Young used Oyster card data to map out the suspect’s travel patterns and noticed that he often began his journey at Camden Road Overground station. One afternoon, they went there to make inquiries. While on the concourse, Young, by chance, spotted the suspect – whom she had only seen in CCTV stills – passing through. “Oh, my God, Porritt, that’s him,” she exclaimed. They ran after the man and slapped on handcuffs. (The sexual offender pleaded guilty in court and was convicted.)
By the end of 2015, the team had much to celebrate. Yet there was frustration, too: the unit’s most high-profile target was still at large. The tally of offences linked to Caballero, rising by the week, stood at 42.
At 2am on New Year’s Day, Caballero ordered a taxi. He was out of cigarettes and did not feel like walking to the shop. On the way home, he tried to jump out of the cab to avoid paying the £5 fare. When the driver locked the doors, Caballero hit him with his shoe. The driver called the police, who arrested Caballero and took him to Holborn Police Station. “New Year’s Day is the worst shift you can possibly work,” Porritt said. “Holborn was absolutely manic, with prisoners all over the place, chaos.”
Caballero gave his name as Jack Donaghy and claimed never to have been arrested before. He was charged for the assault and bailed. One of London’s most wanted men in terms of numbers of crimes – involving stolen goods worth more than £100,000 – was about to walk free.
Just before he could do so, the custody sergeant noticed the red hand that appears on the police computer screen when the person booked has not been fingerprinted. Caballero tried one last ruse, saying that he had to rush home to be with his children. It failed. When his prints were scanned, his real identity was revealed.
At 10am that morning the super-recogniser team was notified. Porritt drove straight to New Scotland Yard to write up the case summary for the police lawyer. “It was longer than my dissertation at university!” he told me. Meanwhile, in the interview room at Holborn Police Station, his super-recogniser colleague James Rabbett showed the suspect the poster that the unit had made using CCTV images of his crimes. “Caballero was gobsmacked,” Rabbett told me. “He then got a bit arsey, saying, ‘I’m not speaking to you until I’ve had a cigarette, I’m getting out of here, you can’t do this to me.’ And then it was: ‘Yeah, it’s gonna be a full hands-up. Let’s get it done.’”
When formal questioning started, Caballero gave his name and date of birth, then smiled, put his head on the desk, and refused to talk. Based on Porritt’s evidence, the police lawyer agreed that Caballero should be charged with 42 crimes.
Word of the Super-Recogniser Unit’s success was spreading. In January, Porritt flew to Cologne to advise German police investigating the mass sexual assaults that had occurred in the city on New Year’s Eve. Enforcement agencies as far afield as India, Australia and the United States – as well as other parts of the UK – visited the unit or requested information on its methods.
One of the most common questions asked of the team is whether computers will put the super-recognisers out of a job. After all, some countries, including the UK, already use facial recognition technology at passport control. Porritt’s unit has its own software but this has been responsible for only one of the 2,010 identifications made since May 2015. DCI Mick Neville reckons that it will be ten to 20 years before software is advanced enough to be a useful tool, and even then super-recognisers will still be needed to analyse the results and identify the suspects.
Josh Davis, the University of Greenwich lecturer, agrees. “Algorithms will get better and we will be able to build 3D representations of faces. But people change appearance and we as humans are primed to see through those changes.”
Meanwhile, studies into the science behind super-recognition continue. Anna Bobak, a research fellow of the Centre for Face Processing Disorders at Bournemouth University, said that the exceptional ability to identify faces has a strong genetic component and that efforts to train people to be better recognisers had yielded mixed results.
In her experiments, Bobak found that whereas most people concentrate on a person’s eye region when looking at them, super-recognisers often focus on the centre of the face, around the nose. “That’s not to say that the nose is important, but more that people can perceive the whole face better,” Bobak told me.
On 1 April, Austin Caballero appeared at Blackfriars Crown Court and pleaded guilty to 42 charges. He was convicted and sentenced to three years and nine months in prison. Weeks later, his name still featured in the whiteboard list of “top ten serial offenders” in room 901 at Scotland Yard – and the number of his crimes was rising. For even though Caballero was now behind bars, his old thefts, caught on camera, were still being solved.