五 但她看起来更像那个角色

这是一个关于欺骗性外表的警世故事，以及揭示伪造者和欺诈行为的技术。

One day in the spring of 1983, not long after I turned five, an unfamiliar woman entered our house. She was neither a family member nor friend. Her name was Doris, she was in her late twenties and she was starting as our mother’s help. She came from Glasgow and had a thick Scottish accent. Her o’s sounded like ‘ae’ and she rolled her r’s. It was gently lilting, almost sing-song.

Doris had a mop of mousy brown hair and wore thin steel-rim glasses. She was plump with a ruddy face. She was the type of person you could imagine going for a brisk walk on a cold day and then sinking into a comfy chair, content with a cup of tea and a shortbread biscuit.

She arrived at our house wearing her ‘Salvos’ uniform. It was a navy suit with big silver S’s embroidered on the collars, complete with a bonnet-style hat. Doris said she belonged to the Salvation Army because she enjoyed helping people. She didn’t bring many belongings with her, although I remember the tambourine she kept at the side of her bed.

My mum had found Doris through a magazine called The Lady. A young aristocrat named Thomas Gibson Bowles, who also started Vanity Fair, founded the magazine in 1885. If you watch Downton Abbey, you will have heard of The Lady. It is the place where high society–including the Royal Family–seeks domestic staff, from gardeners to butlers to nannies. You won’t find any celebrity tittle-tattle or sex stories in the magazine whose tagline is ‘for elegant women with elegant minds’. The lead articles from a past edition included ‘Capture the Style that Wooed a King’, followed by ‘Where to Find Bluebells in Bloom’. There was even a recipe to make teatime Bakewell tarts. You get the picture.

My family is not high society, far from it. So I was intrigued as to why on earth my mum had advertised for help in The Lady. ‘I was starting my own business and feeling nervous about hiring someone to look after you,’ she explains, all these years later. ‘I remember thinking if the Royal Family uses The Lady to find help, it must be reliable and the best.’

Doris replied to Mum’s advertisement. In those days, you would send a formal letter expressing interest in the position and a photograph of yourself. An interview would follow. As Doris lived in Scotland, Mum interviewed her over the phone. ‘I remember her strong Scottish accent,’ Mum tells me. ‘She said all the right things. She told me she was a member of the Salvation Army and had worked with kids of a similar age. But, honestly, she had me at “hello”.’ After their chat, Mum called the references Doris had given and was satisfied they were all impeccable.

Doris lived with us for just over ten months. She was for the most part a good nanny–cheerful, reliable and helpful. There was nothing strikingly suspicious about her, except for one thing. After school every Wednesday, she would drive us to a block of council flats in Edmonton. The building was one of those dark grey concrete high-rises. It sat close to the North Circular, a busy ring road in London. An odd man in his fifties, balding, lived in the flat. And so did a young baby. The flat was dingy and things were always strewn everywhere. I still remember the dowdy wallpaper and damp, musty smell. Doris would spend the entire visit holding the baby.

I told my parents I didn’t like going to this strange flat, to see this strange man. Doris insisted that she was visiting the only family she had in London. Her ‘uncle’ made us nice tea and we liked playing with the baby. The weekly after-school trips continued.

On one of these visits, I noticed there were lots of bottles of expensive-looking perfume on the table; they looked just like the ones Mum had in her own bathroom. I mentioned it to my parents. Funnily enough, it was one of the first times I remember my parents not believing me. I was a dreamer. I had imaginary friends and made up elaborate plays. They told me to stop making up stories about Doris. That it wasn’t nice. So nobody suspected anything was amiss. Or not until Doris’s Uncle Charlie died, supposedly.

One night, around nine months into her stay, Doris didn’t come home. When she did return she explained her Uncle Charlie had suddenly died of a heart attack and she had rushed back to Edinburgh for the funeral. Doris’s mum happened to call our house later that afternoon. My parents naturally offered their condolences. ‘Her mum had no idea why,’ Dad tells me now. ‘Doris’s mother said the brother was alive and kicking. In fact, he was sitting in the armchair having tea right next to her in her lounge.’

Dad confronted Doris. She said her mother was in shock and must have forgotten. ‘I told her it was highly unlikely you would forget your brother dying,’ Dad recalls. Doris finally confessed that she had lied because she had really gone to the VJ Day veterans day parade to see Princess Diana. My parents thought it was slightly odd but Doris was obsessed with the Royal Family so it was plausible. She continued living with us.

The series of events that subsequently unravelled sounds totally unbelievable. You’ll have to take my word that it’s true.

We had lovely neighbours at the time called the Luxemburgs. They had kids of a similar age and also an au pair. Doris spent a lot of time with her. Around a month after the Uncle Charlie incident, Mr Luxemburg knocked on our door late one evening. He told my parents that he had just thrown their au pair out. ‘Philip said he found out that she had been involved in running some kind of drugs ring in North London with Doris,’ Dad relates. ‘They had even been in an armed robbery and he believed Doris was the getaway driver.’ The car, it later turned out, was our family’s silver Volvo Estate.

At this point, my parents decided to search Doris’s room. They found plastic bags full of credit-card statements and thousands of pounds’ worth of unpaid bills. In a shoebox under her bed, she had stuffed piles of foreign currency, stolen from my parents’ home office. Now on high alert, my dad stood on guard by our front door all night with a baseball bat. He was frightened Doris would come home. Thankfully, she didn’t.

The next morning, Dad went to the police first thing. He drove with them to the flat we had been visiting on Wednesdays, visits my parents knew about, even if they had been misled about the true circumstances. ‘There was a big hole in the front door that somebody had tried to kick in,’ Dad recalls. The weird guy was there, the supposed ‘uncle’ who gave us tea. (Turned out the ‘uncle’ was Doris’s boyfriend and the baby was their child.) He had a big iron bar on the table. Doris never returned to our house.

‘Even as I retell this story I feel sick to my stomach,’ Mum says now. ‘I left you in the care of a serious criminal. And it took us so long to know who she really was.’ My parents never hired anyone through The Lady again. Instead, they asked their friends for referrals.

Looking back, what would they have done differently? ‘I wish we had asked Doris more and better questions,’ Mum says. ‘I wish we had known more about her.’ She now realizes the impeccable referees could just as well have been Doris’s friends, family or even ‘colleagues’ in her drugs ring. And the Salvation Army was a total cover story.

My parents thought they had enough information to make a good decision about Doris, even though in retrospect there was a lot they didn’t know about her. There was a trust gap. And that raises an essential point when it comes to trust: the illusion of information can be more dangerous than ignorance. As the Italian social scientist Diego Gambetta beautifully put it, ‘Trust has two enemies, not just one: bad character and poor information.’1

It would be helpful if the likes of Doris wore labels saying, ‘Be warned, I am a con woman and serial liar.’ But they don’t, and of course it’s in the nature of such a person to be convincing. My parents clearly made a very, very poor decision. Yet they are generally smart, rational people with good judgement. What went wrong?

Baroness Onora O’Neill is a philosopher, a professor at the University of Cambridge and a cross-bench member of the House of Lords. Now in her late seventies, she has written extensively about trust and, crucially, how trust is misplaced. She explores that theme in a TED talk, while also challenging the conventional, simplistic belief that as a society we have lost trust and ought to set about rebuilding it.2 More has to be better, right?

‘Frankly, I think rebuilding trust is a stupid aim. [Instead] I would aim to have more trust in the trustworthy but not in the untrustworthy. In fact, I aim positively to try not to trust the untrustworthy,’3 Baroness O’Neill tells her audience, with understated dry humour.

Her point, however, is deadly serious. Trust is not the same as trustworthiness.4 Encouraging generalized trust simply for the sake of creating a more ‘trusting society’ is not only meaningless, it’s dangerous. For one thing, people are already inclined to want to trust blindly, particularly when greed enters the picture. The Bernie Madoff scandal is a classic case. Think of all the tens of thousands of investors who placed their savings with the aptly named Madoff, who made off with their money in an elaborate $65 billion Ponzi scheme that ran over decades.5 Why did investors trust him about something too good to be true? Mostly because Madoff was charming and moved in the same country club and Jewish social circles as they did, in Long Island and Palm Beach. He was a long con, a person who had built up his reputation over years. Indeed, he was known for being a generous, charitable man (it just turned out to be with other people’s money). And besides, his own family, close friends and showbiz names such as Steven Spielberg and Fred Wilpon, owner of the New York Mets, had invested with him. The guy had to be sound, didn’t he? No, as it turned out.

As O’Neill notes, Madoff is an example of too much trust in the wrong place. Instead, all of us making decisions about trust should be looking at the who, where and why of trustworthiness. Who deserves our trust, and in what respects do we need them to be trustworthy? For instance, if I asked, ‘Do you trust your dentist?’ that in itself is not a helpful question. You might sensibly respond, ‘To do what?’ ‘Intelligently placed and intelligently refused trust is the proper aim [in this life],’ the baroness reiterates. ‘What matters in the first place is not trust but trustworthiness–judging how trustworthy people are in particular respects,’ says O’Neill.6

How well do we carry out that logical goal in practice? It’s not always easy.

My parents’ decision to trust Doris came down largely to their personal judgement and blind faith. They wanted, even needed, to believe that what she was saying was true. Their judgement of Doris was also influenced by trust signals. These are clues or symbols that we knowingly or unknowingly use to decide whether another person is trustworthy or not. The Salvation Army, Scottish accents, The Lady magazine, Doris’s cheery appearance, her references and even her steel-rimmed glasses were all trust signals my parents used to make a decision. Trust signals supposedly give us the ability to ‘read’ each other. They give us reasons to trust someone or ways to demonstrate our own trustworthiness. But it’s still a bet, of sorts. ‘Like all gambles, assessing trustworthiness is an imperfect endeavour; there’s always a chance you’re going to come up short,’ writes David DeSteno in The Truth about Trust.7

Some signals we literally ‘give off’, such as our clothes, our face and our accent. Indeed, studies have shown that the Scottish accent is perceived to be the most trustworthy in the United Kingdom (‘Scouse’ is perceived to be the least).8 Other trust signals are non-verbal but still visible, including our posture or gestures such as a nod, smile, twitch or an averted gaze. Despite the admonition not to judge a book by its cover, these first impressions are insanely influential when it comes to trusting someone.

Jon Freeman is an assistant professor of psychology at New York University and director of its Social Cognitive and Neural Sciences Lab. He studies what he calls ‘split-second social perception’. When you see someone’s face, you make snap judgements, within a tenth of a second, about their traits, including how trustworthy they are.9 Freeman wants to understand why our brains take these kinds of mental shortcuts.

Freeman, in his late twenties, is a rising academic star. On a typical day, you will find him wearing slim-fitting slacks, a navy blue button-down shirt and nerdy-cute tortoiseshell glasses. Looking at him, you might instantly label him an ‘academic’. The scientist behind ‘blink’ stereotyping likes it that way. And, in this case, you’d be right.

A few years ago, Freeman and his colleagues devised an experiment to see if there was such a thing as appearing trustworthy. Participants were shown pictures of distinctly different male faces of different ethnicities. They were asked to rate how trustworthy or untrustworthy they thought the people in the pictures were. The results were clear–your brain thinks it knows a trustworthy face.10 Humans are inherently wired this way. When our ancestors were approached by a stranger, they needed a rapid-response mechanism. Friend or foe? But the same rapid response in our day-to-day reactions can lead us to make biased trust decisions based on stereotypes.

The researchers devised a second experiment in which they digitally altered pictures of the same person, evolving the images gradually from looking slightly happy to slightly angry. The study found that people with upturned eyebrows, pronounced cheekbones, big baby-like eyes and an upward curving mouth–even if they weren’t overtly smiling–are more likely to be perceived as trustworthy. Those featuring sunken cheeks, a downturned mouth and eyebrows–even if they weren’t obviously frowning–are more likely to be perceived as untrustworthy. S0 the trick to appearing trustworthy? Look slightly happy. Just like Doris did. The problem is, however, there is no evidence that people with those ‘trustworthy’ features are in fact more trustworthy.

You can’t consciously control whether you perceive a person to be trustworthy or not; your brain does it for you automatically, but not always accurately. And our early assumptions about people can be difficult to budge. That’s astonishing and a little frightening. Trust can therefore be easily misplaced. Indeed, the art of the con man is to give off the right signals and appear trustworthy.

Of course, trust signals are not just based on looks and first impressions. Clear symbols of status or authority, from a white lab coat to a police officer’s badge, are also trust signals, for some. Uniforms can be powerful shortcuts for enabling trust. For instance, if my doorbell rings and I look through the peephole and see a stranger in a postal uniform, I will open the door. The postal uniform is a recognizable symbol that reassures me in my decision. Well-known brands rely on the same dynamic. When I am overseas in a country with dodgy drinking water, I’ll buy bottles of Evian or some other brand I know. Why? The name and packaging make me believe the water inside is safe. I don’t have to trust the person selling me the water, I trust the brand.

Trust signals can also come from the endorsement of third parties. Doris used her association with the Salvation Army, one of the most trusted charities in the UK, to appear trustworthy. She was deliberately sending false signals to dupe my parents. We may not outright lie like Doris did, but think of the number of times we drop in an association with a trusted brand or institution to establish our credibility. I do it all the time when I meet new people: ‘Oh, I teach at Oxford University’ or ‘You might have seen my work in The Economist.’ I am not bragging (okay, maybe a bit, sometimes) but intentionally drawing on these signals to build trust. Going back to the definition of trust, The Economist and Oxford reduce uncertainty about me for other people. They build confidence in the unknown.

We use institutional trust signals to help us make all kinds of decisions in our lives. For instance, imagine you have to choose between two lawyers to handle a matter. One has a law degree from Whittier Law School in California (ranked as one of the worst law schools in the United States) and the other has a degree from Harvard Law School. Removing price from the equation, most people would choose the Harvard-trained lawyer. Similarly, ‘You can’t get fired for hiring McKinsey’ is an adage that has been muttered among management for years as a reason for hiring the world’s most prestigious consulting firm. In both these examples, we are trusting the elite reputation of the institution, not the trustworthiness of the individual.

Trust signals are changing in the age of distributed trust. Consider occupational licences: in the 1950s, fewer than 5 per cent of American workers needed a licence to do their job. Today, more than 1,000 professions–approximately a third of all occupations–in the United States require a licence. In some states you need a licence to be a tree surgeon, fortune teller, florist, horse masseuse, make-up artist, ferret breeder, falconer and even a hair braider.11 Of course, occupational licences are necessary to enforce standards for high-risk professions that require you to put your life in someone else’s hands. I like knowing that the pilot flying the plane or the doctor at the hospital are regulated and licensed. But professional licensing rules have become excessive and don’t necessarily help us to make good decisions as to whom we can trust. Do I really need my hairdresser to pay $2,500 per year and to have no fewer than five sinks and at least ten workstations before I can trust them to cut my hair? Surely ratings and reviews left by previous customers would be a better, or at least an equal, indicator of their talent?

In some instances, new trust signals will be used alongside institutional ones. Think about lawyers. Law degrees or bar memberships are certainly important trust mechanisms, but here’s the thing: once earned, these qualifications do not change based on a lawyer’s actual performance (unless, of course, he or she is removed from the bar for unethical behaviour). Who judges the judgement and quality of skills of professionals? Indeed, I once made a very bad decision in hiring a lawyer who had graduated from an Ivy League law school and belonged to a top-tier law firm. A friend had recommended him to handle a matter. He turned out to be unresponsive and, frankly, incompetent. If I was hiring a lawyer today, I might go to UpCounsel, an online marketplace that matches clients with high-quality attorneys, based on particular skills, pricing and availability. There I could find a suitable one at the push of a button, just as easily as I could hire an Uber ride.

On every lawyer’s profile, it lists qualifications, years of experience and areas of specialization. Clients can also see in real time how much their attorney is billing and what for. Plus, after the transaction, lawyers are rated by clients with one to five stars, just like on eBay, and given detailed reviews. ‘Reviews are one of the most important things in my online profile,’ says Seth Weiner, a popular attorney on UpCounsel, who graduated from Columbia Law School and left life in the big city firms to work as a solo practitioner.12 ‘If 250 people have been happy with me, it stands to reason that the next person will be as well.’ UpCounsel is using online reputation as a means of solving problems of trust. The fascinating aspect of these real-time ratings is how they help hold people accountable for performance once they are in the profession. They allow us to see beyond the more obvious hallmarks of ‘goodness’ (such as Harvard degrees and fancy offices) that could be totally at odds with the quality of work they actually perform.

So technology might hold people more accountable, but to what degree can technology enhance our assessment of who is trustworthy and who isn’t? Would my parents not have mistakenly trusted Doris if they had been choosing a child carer in this digital age? I knew the right person to ask.

Lynn Perkins is the forty-three-year-old co-founder of UrbanSitter, an online marketplace that connects families with babysitters on the internet. She lives in San Francisco and she talks fast, really fast. ‘I love to help people find the perfect restaurant, the right vacation for a friend’s honeymoon or a new job,’ she says. ‘I’ve matched friends needing apartments with roommates and three couples who are now married.’ In short, Perkins is a connector.

In 2008, after the arrival of twin boys, Perkins decided to take time off from her high-flying career in investment banking and real-estate development. Long hours in the office were replaced with lots of time with other mums. The conversations were somewhat predictable. How do you get your child to sleep more? Why are they so picky about what they eat? Why don’t they listen to me? And so on. Perkins also noticed, however, the inordinate amount of time mums spent venting about the shortage of reliable childcare. ‘They’d say they would rather skip going out on a Friday night with their husband if they couldn’t find a sitter their friend had already used,’ Perkins tells me. ‘It was incredibly revealing about what they were looking for and currently couldn’t find.’

Perkins herself had experienced the nuisance of one too many sitters cancelling at the last minute. ‘It was around the time when companies like OpenTable, for restaurant reservations, and Airbnb were starting to crop up,’ she says. ‘But there wasn’t any kind of on-demand marketplace for babysitters and nannies.’ Why, she wondered, could you book a table in thirty seconds but not a sitter? She suspected it was down to trust.

In the early days of UrbanSitter, Perkins did something simple but smart–she borrowed trust from organizations parents already trusted. She went to local music classes, clubs such as Big City Moms, Little Stars Soccer teams, elementary school groups, you name it, and found all the babysitters the parents in the groups were currently using. Perkins convinced the best sitters she met through these organizations to put their profiles on UrbanSitter. But it wasn’t enough. When Perkins first started UrbanSitter in 2011, her closest friends and the investors she pitched to were convinced it would never work. ‘Are you crazy?’ was a common response.

‘They just could never imagine using a service where people would find their care provider online,’ says Perkins. The doubters wondered how finding a sitter online could be better and safer than asking for personal recommendations from friends. ‘One thing everyone wanted to know was whether my team had met and interviewed every single one of the individual sitters. It was something they thought you just had to do to build trust.’

Even then, people simply couldn’t imagine that social networks could not only deliver all that essential information but also make a better job of it. As is now the case with many online services, you have to join UrbanSitter via Facebook or LinkedIn. If a person has fewer than five ‘friends’, it’s a red flag that it’s a fraudulent account. But the real power of the Facebook login is that it can unlock the value of established personal connections. It reveals how we are connected to others. Whom do you know that I know, be it direct Facebook friends, friends of friends, people we went to school with or who worked at the same company. Trust established in one group and context can travel and spread to another.

On UrbanSitter, when you go to book, you can see how many ‘friends’ have previously booked or are in some way connected to that sitter. These connections make us feel more comfortable and confident about our decisions. They reduce the unknown. The collective wisdom of the crowd is enhanced by the wisdom of ‘friends’. It’s social proof on steroids.

The late John Keith Murnighan, when he was a professor in social sciences at the Kellogg School of Management, set out to explain what causes us to trust people we do not really know. Specifically, he was interested in the role that ‘friends’ play in stimulating feelings of trust for a stranger.13 He conducted a series of experiments based on the famous ‘trust game’ that was originally designed by behavioural economists in 1995. In the game, there are two participants, a sender and a receiver, who are anonymous strangers. Both are given a certain amount of money, let’s say $100. The first player can send any portion of the $100, or none of it, to the second player whom they will never meet. The first player is told that whatever amount they send will be tripled, by the experimenter, for the receiver. The receiver then has to make a similar decision: how much of the tripled amount should they return to the sender? Player A can therefore either potentially turn a profit or lose everything. The point of the game is that sending a large amount of money indicates a high degree of trust.

Before volunteers participated in the trust game, Murnighan and his team asked them to provide the names of people they trusted and people they distrusted and the reasons for their feelings. The researchers quickly flashed these names for mere milliseconds to subliminally prime the study participants. It was too fast for anyone to be able to recognize the names that had been flashed. After that, the participants played the classic trust game. The results were stunningly clear.

Participants who had subliminally seen the names of people they trusted, sent on average nearly 50 per cent more to the anonymous receiver than the participants who saw the names of people they did not trust. ‘We found we could stimulate feelings of trust for a stranger without people even realizing,’ wrote Murnighan, an outcome he found ‘both exciting and scary’. ‘Imagine a fanatic fan of Elvis Presley. If I know someone is a huge fan of Elvis, I might casually drop Elvis’s name to activate more trust in me. There is clearly a risk of manipulation.’14

The results of the game help explain how the likes of Madoff managed to deceive so many people. His client list included the rich and famous and his own friends and family–his brother, his sons and their wives. When new investors spotted close relatives on the investment list, it was a powerful trust signal that Madoff himself was trustworthy. Whether Madoff intentionally or inadvertently used the friends and family connections does not matter. The point is, these relational cues can provoke automatic trust that can be very dangerous, especially in cases such as Madoff’s where we might lack the time or expertise to make a deliberate and proper evaluation.

But Murnighan’s experiment also sheds light on the power of online social connections–the wisdom of ‘friends’ can automatically enhance our ability to trust people we do not know.

Interestingly, Lynn Perkins wrongly assumed the most influential social connection would be between parents. Instead, it was between sitters. ‘Over time, parents value the connections UrbanSitter can surface between sitters; they want to book the friends of the sitter they really liked,’ she explains. When you think about it, this is how trusted referrals typically work. Say an entrepreneur asks me for the name of the designer I trust. I introduce them to Amy Globus, whom I have worked with for years, but she is too busy to take on their work. The entrepreneur is then likely to ask Amy to recommend another designer rather than coming back to me for another suggestion. In other words, trust really lies within the group with the expertise (the babysitters) rather than the group with a similar need (the parents).

The social graph* is information manna when it comes to trust. We are now using digital tools to approximate old-fashioned ways of finding people we could trust, through referrals and close connections, but in ways and on a scale never possible before. ‘It’s a digital recreation of the neighbourly interactions that predefined industrial society,’ writes Jason Tanz in an excellent piece on digital trust in WIRED. ‘Except now our neighbour is anybody with a Facebook account.’15

The strange and beautiful truth about the social graph is that it shortens the distance between any two people in the world. Think of it as the string tying together the arbitrary connections between humanity. In 1929, the Hungarian author Frigyes Karinthy first wrote about the theory of the ‘Six Degrees of Separation’, later popularized in a play by John Guare, claiming that everybody on this planet is separated by only six other people. This is where the ‘Kevin Bacon game’ comes from–it would take the average person six or fewer acquaintances to be connected to the actor. Today, that number has shrunk significantly, at least for the 1.8 billion active users on Facebook. In 2016, Facebook crunched their social graph and determined that the degrees of separation number is 3.57.16 In other words, each person on Facebook is connected to every other person by an average of three and a half other people. It means that, despite Facebook’s vastness, it feels intimate. It gives us the sense of connection and trust we used to get from real-life communities and neighbourhoods. Its size, however, is critical to its usefulness.

When my parents decided to trust Doris, they based their decision on faith–they believed what she and her references claimed. In the past, we had to make a lot of decisions based on blind faith or personal experience, but today we can base them on collective experience–the experiences other people have shared through reviews and social networks.

I don’t need to have personal experience with a nanny I hire through UrbanSitter to assess if she or he is trustworthy; I can benefit from other people’s experiences. It’s a dynamic often referred to as indirect reciprocity– and it can speed up the process of trust. Today, on UrbanSitter, the average sitter responds in less than three minutes, down from twenty-three hours five years ago. The average time a parent takes from either posting a job or doing a search for a sitter to accepting the person they trust to look after their kids is less than ten minutes, down from twenty-three hours when UrbanSitter first started. It shows that, online, questions of trust can be settled very fast. And this process is only going to get faster, from minutes to seconds.17

There is just the small matter of things going wrong now and again in this speedy and efficient brave new world.

It is almost impossible to get entrepreneurs, people like Lynn Perkins, to tell you the exact number of bad incidents, minor or serious, that happen on their platforms. I have prodded and cajoled to try to get actual data but instead you tend to get a generic response. ‘They do happen but they are extremely rare,’ insists Perkins. Take Wendy, a mum living in Seattle who hired a sitter to look after her six-month-old daughter one night in May 2016. The day after, she received a call from her bank about a cheque with a dodgy signature. It turned out the sitter had gone through Wendy’s drawers and stolen a cheque that she had written out for $1,300. ‘As soon as the parent let us know what happened, the sitter was blocked from our system,’ Perkins tells me. Bad apples will make the cut, it’s inevitable, but it’s also likely they’ll be exposed more quickly.

‘She seemed so trustworthy’ is something we commonly say after trust has been misplaced. On the flip side, you have probably been on the receiving end of someone saying ‘I trust you’. But what do these statements mean? And, coming back to Baroness O’Neill’s point, what is it that we should be basing our trust upon?

Turns out, there’s a relatively simple formula to trustworthiness that goes beyond ‘but he had kind eyes’ or ‘she looked the part’. It doesn’t matter if you are deciding whether to trust an estate agent, a lawyer or a babysitter, the three traits of trustworthiness are the same: Is this person competent? Is this person reliable? Is this person honest?18

Competence comes down to how capable a person is to do something. Does he or she have the skills, knowledge and experience to do a particular role or task, be it cut my hair, mind my children or fly me to Uzbekistan.

Reliability comes down to a person’s consistency in doing what they said they would do for you. Ultimately, it’s about you knowing, ‘Can I depend on this person?’ Will he or she follow through?

Honesty is about integrity and intentions. ‘What are their interests and motives towards me?’ Basically, it’s whether their intentions are aligned with yours. What do they gain by lying or by telling the truth?19

Political scientist Russell Hardin eloquently argues that trust is really about encapsulated interest, a kind of closed loop of each party’s self-interests. He argues that if I trust you, it’s because I believe that you are going to take my interests seriously–whether it be for friendship, love, money or reputation. Why? You won’t take advantage of me because it benefits you not to do so. ‘You value the continuation of our relationship, and you therefore have your own interests in taking my interest into account,’ Hardin writes in Trust and Trustworthiness.20 For instance, I trusted the estate agent who recently sold my house to get a good price not because she was nice or cared about me but because her commission was directly tied to the sale price. That’s encapsulated interest. Or as economist Adam Smith would put it, the estate agent’s future payoff is a strong enough incentive for her and good enough reason for me to trust her.

We often mentally ask of someone, ‘Do I trust you?’ A better question is ‘Do I trust you to do x?’ We need to think of trust as trusting someone to do something. For instance, you can trust me to write an article but trusting me to, say, drive a lorry would be a grave mistake. You can trust me to teach twenty-something-year-old graduate students, but put me in a class of five-year-olds and I would probably lose my rag in my attempt to teach them to read and write. What we are trusting someone to do fundamentally changes the alchemy and order of importance of those three necessary traits: competence, reliability and honesty. Trust is contextual.

I am not suggesting that we have to go through this kind of assessment with every person we need to trust. Take something simple like getting on a bus or train. We don’t want to be assessing the driver’s skills. If we thought about every decision to trust, we would spend the entire day asking questions and making checklists. We might never even leave the house. But let’s say we’re hiring someone for a job. How do we make good decisions about whether or not to trust what they claim about themselves?

Not so long ago, CVs were the principal tool in job hunting, listing what someone had–allegedly–done and with whom, but they didn’t provide much proof. Indeed, a survey done in the UK by Powerchex, a company that screens CVs on behalf of finance companies, found that, out of 4,735 job applications, 18 per cent contained outright false information.21 The most common lie was to claim a 2:1 university degree when they had been awarded a 2:2. Embellishing the truth, such as exaggerating job titles (from project assistant to project director) and tasks, is very common. CVs were often ‘creative’, at best, but pretty much all the average employer had to go on.

Today, many people send their LinkedIn page instead of a CV and include links to their other online portfolios and social profiles. Think for a minute about the number of profiles out there that contain information on you. I easily have fifteen: my profiles on Amazon, eBay, LinkedIn, Facebook, BlaBlaCar, Uber, YouTube, Twitter, TED, Oxford University Faculty page, my personal website, my literary agent’s website, my speaking agents’ websites and so the list goes on. And that does not even include all my accounts that require a Facebook login such as Spotify, Airtasker and Airbnb. Online profiles are another example of the shift in trust signals. Information that used to be held by institutions or small groups of friends, family or colleagues is now distributed among many people. In this sense, trust signals have become socially fluid.

In online services such as UrbanSitter, people create specific online profiles listing detailed information. The amount and type of details people will voluntarily disclose is extraordinary. ‘One parent included a long description about their pot-bellied pigs,’ Perkins explained. ‘They wanted to make sure the sitter was comfortable with the family keeping pigs as pets. It sounds weird but is actually great expectation management.’22 Both sides knew what they were in for.

The information a sitter provides in their profile is verified through different online checks. Do you really have cardiopulmonary resuscitation (CPR) certification? Did you really get a Level 5 childcare diploma from South Thames College? Do you really have a clean driving record? Or in the case of Doris, do you really belong to the Salvation Army? In fact, only 25 per cent of the sitters that start the registration process make it on to the platform. Put differently, 75 per cent are rejected.23

UrbanSitter uses forty different criteria to find the six sitters most compatible to the parent. The algorithm takes into account the age of the children, the parents’ social connections to the sitter, where they live, when the sitter is needed, specific preferences and so on. The marketplace also uses a reputation system just like eBay–parents leave feedback and rate sitters after each transaction.

‘People rate everything now. Your Uber driver, the guy who delivers your food, and not all things are equal. There are only so many things you can do wrong or right when you’re delivering my food,’ says Andrea Barrett, UrbanSitter co-founder and vice president of product.24 ‘But when you’re watching my child, there are a lot of things that can go awry. Your ability to deal with the behaviour of my child, how much you engage with my child, how much you cleaned up, were you friendly, did I like you? There’s a lot going on.’

Reviews significantly influence a sitter’s ability to get a booking–a sitter without reviews is two times less likely to get a booking than a sitter that has at least one. A valuable badge experienced sitters and parents carry is called ‘Repeat Families’. Basically, it means you were invited back.

Reliability is also easier to demonstrate online. If you have booked a place on Airbnb, you may have noticed that hosts are categorized by how quickly they respond. My response rate is 100 per cent but my response time is, um, twenty-four hours. This means I respond to all new messages but by Airbnb standards I am slow to message back. Similarly, UrbanSitter categorizes sitters based on how quickly they reply. ‘If you are looking for someone last minute, it’s good to know upfront how long it will take them to respond,’ says Perkins. ‘But on a deeper level, I think it’s an indicator of reliability in some weird way. If this person is really slow to respond, will they show up on time? Are they really interested?’ Perkins is right; time is often used as an indicator of reliability.

The hardest trait, without a doubt, both to prove and predict online, is honesty. How do you get a real read of someone’s intentions and integrity? Perkins knew this was a big problem she had to address. In 2014, she wondered if asking the sitters to make short videos about themselves and their interest in kids would make a difference.

Most videos start with, ‘Hi, my name is so and so and I would like to give you a bit of information about…’ Then they get more personal, explaining why they want to look after your kids. One wondrous offshoot of the videos is seeing the rooms people are in. Some sit casually in their bedrooms on unmade beds, others choose the sofa in tidy living rooms.

I tried an informal experiment. I read the written profiles of twenty sitters and wrote down characteristics that summed up my first impressions. I found I kept using the same vague words such as ‘nice’ or ‘seems friendly’. I then did the same exercise using the videos. The descriptors were clearly more specific, such as ‘considerate’, ‘warm’ or ‘nervous’. Hearing sitters talk and seeing a small glimpse of their environment felt like seeing inside their lives. ‘It allows them to be human,’ Perkins says. Today, it is mandatory for all sitters to leave a thirty-to-ninety-second short clip video about themselves. There is this, though: Doris would have made a great video.

So would UrbanSitter have caught that she was a fraud?

On 5 July 1993, a ‘dog’ cartoon by Peter Steiner appeared in the New Yorker. The pen-and-ink artwork features two dogs, one sitting on the floor and the other in a chair in front of a computer. Underneath is the prescient caption, ‘On the internet, nobody knows you’re a dog.’25 In 2013, to mark the twentieth anniversary of that iconic cartoon, the web comic ‘Joy of Tech’ created an updated version. It featured two agents with sunglasses standing in the National Security Agency (NSA) surrounded by computer screens. ‘Our Metadata analysis indicates that he is definitely a brown lab. He lives with a white-and-black-spotted beagle mix and I suspect they are humping,’ reads the caption.26

Sitting in the meeting room of Trooly*, an ‘Instant Trust Rating’ venture launched in July 2014 based in Los Altos, California, Steiner’s cartoon came to mind. Before I arrived to meet Savi Baveja, the co-founder and CEO, his team had come up with an idea. ‘We thought it would be helpful to run you, Rachel Botsman, through the Trooly system,’ Baveja says. ‘We’ll do it on the projected screen so you can see what is happening in real time.’ Great. Five other people were gathered round the conference table. I felt a tad uneasy about what they might find. Baveja, in his late forties, a former senior partner at Bain & Company, has a calm and thoughtful demeanour. He picked up on my silent pause about the plan. ‘Don’t worry,’ he reassures me. ‘We did a test run before you got here and you are all good.’

When you first connect with people, total strangers living on the other side of the world, how do you know if they pose a serious risk? ‘People talk a lot about reviews and ratings. Well, by definition, they’re backward-looking,’ says Baveja.27 Indeed, the goal of Trooly is to fill the trust gap caused by the sheer speed of online commerce.

My first and last names are entered into the Instant Trust software, plus my email address. That’s it. No phone number, age, date of birth, occupation or address. Anish Das Sarma, co-founder and chief technology officer, tells me the machine learning software is searching whatever public and permissible data it can find associated with my identity. ‘Watch lists, the National Sex Offender Registry, social media, et cetera. But it’s also scanning the deep web, the parts of the web not indexed by search engines, so you wouldn’t surface them in a search. We have indexed deeper than Google,’ Das Sarma adds.28 I think he is referring to the weird websites I might hang out on, getting under the hood of what I really do on the internet. He asks if he can add my middle name to send a ‘stronger signal’. Sure.

A long thirty seconds passes. Then the results appear. ‘Look, you are a one,’ says Baveja, pointing at the screen. ‘Only approximately 15 per cent of the population are a one; they are our “super goods”.’

‘So how many are “super bad”?’ I ask the team.

‘About 1.5 to 2 per cent of the population end up between five and four. There are some exceptions in some populations we’ve done because some populations attract super creeps,’ Baveja explains. ‘The vast majority are twos. And for around 10 to 15 per cent of people we screen in the United States we can’t generate a confident score because there is either not enough of a digital footprint out there or enough accurate inputs.’

Honestly, the level of information the search had pulled was staggering. I didn’t mention my maiden name, so I assumed that nothing linked to ‘Simmons’ would appear. I was wrong. Links to clubs I had joined at Oxford and Harvard were in the search list (yes, I was a member of the darts society and, yes, I spent almost nine months in the Territorial Army Officers’ Training Corps). The data was consolidated into five different categories. The most basic was verifying my identity. Was I who I said I was? Check. Next was criminal record and possible unlawful activity. Clean. Phew, they missed the car accident case I was involved in when I was twenty because it was dismissed in court. And then there is a category called ‘anti-social’. ‘A lot of people won’t have done anything explicitly good or bad,’ explains Baveja. ‘So we have spent a lot of time figuring out the granularity of how to get at “what kind of person is this?” We want our customers to look at more than just a superficial score before making decisions.’

Context for the anti-social category is king. Clients tell Trooly the types of traits they want to screen against and bad behaviours they need to weed out. For example, UrbanSitter may want to screen potential sitters against hate language. They might also want to know if you have ever had a drug or alcohol problem or if you have ever been involved with pornography. A home-sharing site might care a lot if you are a sex party organizer. A ride-sharing platform will want to know if you are a terrible driver. I also get a one in the ‘anti-social category’, meaning I am ‘pro social’. I feel relief, even a tinge of pride that I have been given an A* trust rating.

Baveja is a warm, considerate and clearly intelligent man. His sentences tend to contain lots of thoughtful questions, as if he is constantly seeking better answers. He went to Stanford University for his undergraduate degree in electrical engineering. A few years later, he completed his MBA at Harvard Business School where he graduated as a Baker Scholar, meaning he was one of the top students in his class. He then became a management consultant, rising to the highest echelons of Bain & Company, first becoming a partner and then joining the company board. ‘While at Bain, I started to look into traditional background and credit checks, and realized how dangerously flawed they are,’ he says. ‘A background check is just that–it is retrospective and does not foresee the future. But does this have to be the case? I thought there must be something better.’

Our behaviours have changed but the trust mechanisms we use in society have stayed pretty much the same. For one thing, in the US, UK and much of Europe, the current background-check system is still slow and manual, often relying on low-paid and overworked court runners to rummage through records and so on. No wonder all kinds of mistakes happen, especially if your last name happens to be Jones, Smith or Harris, or another common moniker. Ron Peterson, who lives in California, knows this problem all too well. ‘In Florida, I’m a female prostitute (named Ronnie); in Texas, I’m currently incarcerated for manslaughter,’ says Peterson. ‘In New Mexico, I’m a dealer of stolen goods. Oregon has me as a witness tamperer. And in Nevada–this is my favorite–I’m a registered sex offender.’29 If you are falsely mismatched with someone else’s felony, it’s known as a false positive. It’s an alarmingly common problem.

Out of the people the traditional checks label ‘bad’, how many are actually criminals? Worryingly, a study conducted in 2016 by Simone Ispa-Landa, an assistant professor at Northwestern University, and Charles Loeffler, a professor at the University of Pennsylvania, found that one out of three American adults has been involved with the criminal justice system and has a record, even if they were not found guilty of a crime.30 The United States Attorney General’s office recently found something similar. Half of all case files in the background system contain no information about how the cases turned out–whether the person was found guilty or not, or even prosecuted. Bad data fouls the system and the most common groups falsely labelled are blacks and Latinos. ‘What we’ve ended up doing is taking all the biases and pre-existing preconceptions of the criminal justice system,’ Baveja says, ‘and ossifying them, institutionalizing them in more and more and more decisions where they don’t belong.’

Baveja raises question after question about the checks system. ‘I can tell you that 50 per cent of what you will find in a background check is going to be either traffic violations or drug convictions,’ he says. ‘But if I smoked a joint seven years ago, what does it really tell you about me? Does it mean I’m going to be a bad tenant?’

‘But surely, if I had been convicted of shooting someone, the check would flag this up?’ I ask.

‘No, a fail is a fail in a traditional background check. It’s just not precise,’ says Baveja.

But there is another problem: the system can miss people who really are criminals. It’s known as a false negative. Of the millions of checks done every year, between 1 and 2 per cent turn up a problem. Indeed, the vast majority of people who have done bad things pass their background checks. ‘How on earth did we get to a point where we rely on flawed data and processes that lack rigour to determine whether someone should get a job or will be a good tenant?’ says Baveja. ‘I mean, these are serious decisions, right?’

I admit to the Trooly team that even though their intentions seem entirely positive, the process still feels intrusive. ‘I knew you were going to ask us about privacy, everyone does,’ says Baveja. ‘I laugh when people talk about privacy. Not because it’s not a serious issue but think about how much information you give to the banks. Every bill, purchase, credit card, everything, and we just accept that, right?’ Trooly complies with applicable privacy and data protection laws, including the Fair Credit Reporting Act in the United States. So if I wanted to, I could opt out of the ratings altogether, check and challenge my profile data. My report is deleted after ninety days. ‘We’re not doing anything in the shadows here,’ says Baveja. ‘Any trust mechanism, even asking a friend about someone, involves some level of intrusiveness. I think the dividing line is how you do it. Are you doing it ethically? Are you doing it transparently? We bend over backwards to make sure what we are doing is way better, way fairer and way more predictive than any other trust mechanisms being used today.’

Here’s the thing. Baveja is reassuringly competent, open and honest. I trust him. And he clearly has positive intentions to disrupt the existing background-checks system, which is seriously flawed. I was still left, however, with the feeling that current privacy and data protection laws don’t offer a sufficient degree of protection when it comes to what Trooly customers could do with our data.

At the end of my day with the Trooly team, I tell them about Doris. Would the search have caught her if she had applied to UrbanSitter? The answer is a definitive ‘yes’. My parents would have known that she didn’t in fact belong to the Salvation Army, had no previous childcare experience and had a chequered criminal history. In other words, Doris would not have made the cut.

‘Trooly is helping us screen our caregivers in an objective and quantifiable way. But it’s only part of the equation,’ Lynn Perkins says. She makes an important admission: technology should never try to replace parental intuition. ‘If a sitter shows up at your door and you get a weird feeling, it doesn’t matter if they have passed checks, how well reviewed they were or what you thought about her online, say you are suddenly not feeling well and cancel. Go with your gut.’ There’s no question: at the end of the day, we’re the ones who have to decide where to place our trust.

It’s complicated. Our intuition may be strong but sometimes we tune into trust signals that are loud even though they are not in fact good indicators of reliability, honesty and competence. Doris, with her spectacles and Scottish accent, is a prime example. Someone looks the part, and it’s only later you discover that your kids have been looked after by an armed bank robber and drug dealer.

In the future, online trust can only get faster, smarter and more prevalent. And that has to be a good thing in terms of helping us make more informed choices–whether it’s about hiring a lawyer, selling our house or taking care of our kids. At the same time, we don’t want to lose what makes us human.

Making the odd mistake when it comes to trust, taking a leap, is sometimes how we open up new possibilities or find ourselves in unexpected situations, both exciting and dangerous. It’s how we place our faith in strangers, without knowing what might come of it. For that matter, my mother’s trust in Doris wasn’t entirely misplaced. The funny thing was that while Doris turned out to be an accomplished criminal, she was also a rather good nanny.