Today, Madan has his hands full with a new start-up, Ginger .io, which is applying telemetric signaling and predictive analytics to health. This arguably is a more noble cause than matchmaker tech. And the market for relationship software to help couples digitally analyze how they speak to each other doesn’t really exist. But the utility of telemetric communication analysis is being proven. The testing ground just happens to be someplace other than love.
There’s a reason people sometimes claim to feel married to their job. Marriage takes work, yes, but our work life has a lot in common with a long-term romantic relationship. Collaboration styles have a huge influence on outcome and performance. But there’s a key difference: poor collaboration between two people in a workplace can hurt an entire organization, resulting in lost revenue or worse. This is why its employers are leading the way in developing techniques to actually collect real-time relationship data.
In the big data present, the honest signals that occur between people, the inaudible notes that make up the tone and character of our interaction beyond what is literally said, are mostly lost. In the naked future that ability spreads to more people and more couples. Suddenly, a lot of people can become much smarter about what effect their words and actions will have on the person they’re with. This future is visible today in the way that a few ambitious organizations and companies are measuring collaboration dynamics.
How Office Relations Lead to Nuclear Meltdowns
In 2012, Cindy Caldwell, Christopher Larmey, and statistician Brett Matzke of the Pacific Northwest National Laboratory (PNNL) set out to try to predict where workplace accidents were going to occur around the lab and which teams of employees (or work groups) would be involved. An accident at PNNL is a bit more serious than a stubbed toe or a sprained wrist. The lab, which does work for the U.S. departments of Defense and Energy, is involved in cutting-edge research on nuclear fission, natural gas development, and weapons research. Employees handle volatile, radioactive, poisonous, and highly classified material on a daily basis. A bad day at the lab is a bad day indeed.
Matzke plotted all the accidents that had occurred in the lab during the previous year. He and his coresearchers had to consider all sorts of mishaps, from the ones involving explosive material to more mundane types involving vehicles, falling from ladders, or just misfiling paperwork. They wanted to see if there existed some common feature among them that predicted their occurrence.
They discovered that employees who indicated (via survey) that their relationship with their supervisor was strained, who felt they weren’t well listened to, that their concerns weren’t shared, and, as a result, weren’t engaged in their workplace were much more likely to have an accident.
When Caldwell and her fellow researchers added together the scores for (A) whether the group worked with hazardous materials (note: they found that a work group will have 1.9 accidents a year just because they’re exposed to hazards); (B) worker engagement; and (C) past operational experience (defined as previous incidents, sick days taken, staff performance, hire and attrition rate for a group), they were able to almost perfectly predict the number of incidents that each work group would experience that year.28 Considering that the work groups contained an average of just sixteen people, it’s a short leap from figuring out the weak link work groups to the particular weak link workers in the groups.
That number becomes far more useful if you can also predict when that accident will take place. When I asked Caldwell if there might be some way to do that, she acknowledged that constant telemetric monitoring of employee engagement would provide more actionable data than a once-a-year survey.
A couple of years ago a California company, e22 Alloy, began marketing a software-as-a-service (SaaS) application (called Alloy). The company was one of very few start-ups that could analyze the “continuous, objective data of the online activities of the workforce.” The distinction between continuous data collection and spying is a subtle but important one. Monitoring employee behavior without that employee’s knowledge and using collected information to punish employees can indeed be called spying. Forcing employees to submit to having all their computer activity watched is not spying if you tell them you’re doing it but that sort of petty office-tyrant behavior isn’t going to be good for morale and probably won’t be much of a productivity boost, either.
Company founder Josh Gold was very sensitive about the spying applications of his product. In his presentation at Strata 2012, he recommended that employers not use the program without the explicit permission of their staff, and that employees should be able to suspend tracking whenever they choose.
Used properly, this sort of app could provide supervisors with “advance warning” that a big project is headed off the tracks. For instance, if you’re a manager and one of your work teams starts communicating a lot more, but tangible work product decreases, that’s a warning sign, as is “changing activity patterns,” which could take the form of a lot more e-mails suddenly shooting back and forth at the end of the day and/or a lot of profile updating on LinkedIn.29
Back to love. In the same way that tone, timing, and particular changes in interoffice chatter can help predict project failure, communication changes between spouses can be indicative of buried problems. Marriage and work really do share a lot in common.
Here’s a case in point. My wife and I both work from home and are copartners in a little joint hobby we call “not letting the house become an apocalyptic hellscape.” Succeeding in this endeavor requires a certain amount of vacuuming, laundry folding, moving of trash and recycle bins to the curb, returning them to the side of the house, and dishwashing. It demands effort, management, and communication. In fact, it’s very much like a regular job, and each of us has one (or more) of those as well.
More important, neither of us can agree on the definition of “hellscape.” I fall more on the literal side. In fact, if I’m deep in a project, I may not notice that an interdimensional portal has opened over the cat-litter box until Yog-Sothoth the Outer God wraps his cold tentacle around my neck. My wife, meanwhile, knows by sight how many days it’s been since someone ran the vacuum.
In our interactions, we’ll fall into the role types outlined by Pentland’s research. “Do you think we should run the vacuum?” she will ask, taking the explorer role, which my wife uses to talk to me about vacations to faraway places, contemporary issues, and our mutual acquaintances. I do not want to explore the issue of running the vacuum. I will respond that I am too busy and prattle off all the items on my to-do list I feel are more relevant. In doing so, I will speak calmly and evenly, taking the leadership position. My to-do list is something I can speak on with authority. I will win this exchange, but in doing so, I will lose. My wife will run the vacuum but not feel good about it. She has her own to-do list that is as long as mine, but she can make time to play “not letting the house become an apocalyptic hellscape” twice as hard. I detect her feelings and reflect them, rationalizing my own feelings of resentment.
People in long-term relationships approach exchanges with residual notions and emotions from the last exchange. Over time this can erode a person’s ability to objectively perceive what’s fair or logical in terms of the division of household labor, expenses, goals, and so on. Communication telemetry could fix this problem. Now imagine the workforce telemetry solution described above in place at home.
If I’m deeply involved in a project, running up on a deadline, then my communication patterns, my Internet usage, my verbal exchanges with my wife will indicate this just as it does when a workplace manager is floundering on a project. In those instances where I truly am too busy, I won’t have to tell my wife I don’t see the need to vacuum; she will actually be able to verify it herself. I, likewise, could use data from her communication patterns to reach a better understanding of her current stress level, which would speak to engagement with the house. I don’t ever have to see the house in the
same way that she does to remember to run the vacuum. All I have to see is her current stress level and then, without questioning, bring out the Dyson.
This isn’t a perfect solution to the problems that arise in long-term cohabitation, but it does strike at one of the biggest unspoken problems in modern marriage. After a certain period, we expect the person we are with to be able to anticipate our moods. This expectation is not born of any rational or objective understanding of the way we communicate but of simple exhaustion. We become tired of explaining ourselves. The only solution is to develop the capacity to say more without exerting more effort, and telemetry can help with that.
In the years ahead, if more managers begin to experiment with telemetric solutions and if those experiments prove to be successful, we may become more accustomed to the idea of digging deeper into the secret signals inside our private communication. Not everyone will want to monitor their relationship signals for warning signs but for those who do, the process will become easier and cheaper. More of our talking, chatting, and signaling is taking place online and will be retrievable later at lower cost.
The last predictor of relationship longevity, according to Finkel, could be called the stress test. The late psychologist Reuben Hill first proposed it in 1949 after surveying couples who had been separated during World War II. What he found was that how a couple deals with unexpected emergencies—the loss of a child, critical injury, a sudden drop in wealth—portends strongly for the future of that couple. In much the same way reactions to stressful events predict future health states for individuals, couples who were able to survive high stress events grew more stable and became less likely to break up.30
Stress is the fiery crucible in which truly stable marriages are formed. We’ve long known this, but today the effects of stress on husbands and wives—or potential husbands and wives—can be modeled or run through a simulation, in the same way we simulate hurricanes, floods, and credit defaults to test the resilience of infrastructure and institutions. Some of the most promising work in treating post-traumatic stress disorder (PTSD) right now is based on running simulations of the traumatic events. There’s no reason why a couple who was really serious about forging the strongest long-term relationship possible couldn’t run simulations or game traumatic events beforehand to see how such an event would impact their relationship. If current experiments treating PTSD with simulations continue to prove effective, marriage counselors could recommend traumatic event role-playing as a means to better ensure relationship health.
Currently, scenario testing traumatic events is not an action that people consider when planning a future with someone. There just never seems to be a good time to tell the person that you are with, “Before we take this any further, let’s do a few virtual reality natural disaster simulations!” In other words, there is no science or data yet on the effects of stress testing on marital relationship longevity. That’s just not the way we think of love, but we know that stress tests and simulations in business and engineering are effective in finding problems before those problems blow up. No one would get on a bridge that was marked with a sign reading THIS BRIDGE HAS NEVER BEEN LOAD-TESTED. Yet we carry on for years in relationships to which we’ve never applied any sort of objective strength test outside a Cosmo quiz. When we learn to approach personal relationship decisions with the same seriousness that we collectively approach issues of public safety, then all of us will experience fewer relationship disasters. The idea might seem far-fetched but a few years ago so did the notion that a majority of singles would turn to the Internet to find love before heading out to a local bar. As stress tests and virtual reality simulations prove their utility in other areas of life, we will eventually get around to applying them to love and then the last component of the formula will be in place. We can finally create our soul-mate predictor app.
The Love Machine
So what is the future of love? We know that personality profiles can help you predict who will or won’t be a great date; sociometrics can tell you how well a date will go. A data set of sociometric scores going back for years will reveal how your personality, and that of someone else, might interact over a period of years. Trauma simulation can even give you a sense of how your marriage will weather life’s big storms. An ensemble of these scores won’t tell you if you’re in love, but you can predict arguments and resolve them in advance. You can get a window into the future of your relationship with someone. You can find a mate who is indeed scientifically suited to you. More important, you can use science to make your relationship stronger.
The first step toward a better science of dating is getting customers to give up more data about themselves and how they date, beyond simple information about the sort of person whom they’re interested in.
What Yagan wants is a lot more data from his users, not just information on how they answer questions about what they’re looking for in a relationship but also Amazon reviews that provide a sense of why some people find some products are superior to others, Facebook and Foursquare information about comings and goings, Fitbit data to measure the beating heart. These signals, formerly inaudible but now detectable online, make up what Yagan calls “true identity” and leveraging true identity “broaches a line that no site has managed to cross.”
Sandy Pentland reached a similar realization early on in his work on sociometric data; that “by adjusting for personal characteristics and the history of previous interactions, we can dramatically increase our ability to predict people’s behavior.”31 With enough data a naked future emerges, a profile that is more living and thus predictive than any survey questionnaire because it is assembled from action, because it changes, as do you and I.
But would we dare call this love?
Perhaps we need to change our definition of what love is. We tend to view it as something we own and thus can lose, something we want and are entitled to, and something we lend in the hope of getting back. Perhaps love is more fluid, less connected to who we are and more firmly attached to what we do. Love is more than dopamine (at first) and oxytocin (later). It’s a decision that, if we are lucky, we are called upon to make over and over again. We make hundreds of decisions in our relationships every day. If we could develop the ability to pick up just a few more of the signals that the person we love sends out continuously, then that decision making would improve. Love becomes less work.
Though the Brahmins understood little about the makeup of the universe compared with what we know today, they understood that idea well enough.
After describing why Vedic astrology is an expert practice worthy of admiration, Paramahansa Yogananda, in his Autobiography of a Yogi, effectively devalues the entire endeavor to predict the future and launches an eloquent defense of free will: “The message boldly blazoned across the heavens at the moment of birth is not meant to emphasize fate, the result of past good and evil, but to arouse man’s will to escape from his universal thralldom. What he has done, he can undo. None other than himself was the instigator of the causes of whatever effects are now prevalent in his life. He can overcome any limitation, because he created it by his own actions in the first place.”32
CHAPTER 9
Crime Prediction: The Where and the When
WHEN crack first got to Pittsburgh, Pennsylvania, in the late 1980s, police attacked the problem the only way they knew how: busting dealers who were working out in the open, performing sting operations, and planting patrols on blocks where they had disrupted drug traffic before. Getting a dealer off a particular block was a big victory. If you’re a Pittsburgh crack pusher, you can’t just send a letter to your clients with your new address. And naturally, as anyone who has ever seen The Wire knows, whenever a dealer is forced to relocate to a new block he runs the risk of encroaching on territory that belongs to another dealer, which can lead to . . . disagreement. The police understood that clearing and holding blocks were crucial to slowing the spread of crack but they didn’
t have the resources to clear and hold every neighborhood. Some dealers were going to relocate. Finding a new market that isn’t yet occupied buys a drug dealer a lot of time to set up. If the police could anticipate which neighborhoods were the most conducive to drug dealing and why, theoretically they could predict where the dealers were going to go set up.
How to figure this out? The most well-established approach to predicting which neighborhoods were going to experience an uptick in crime was called broken-windows theory. In 1982 researchers James Q. Wilson and George L. Kelling observed a correlation between neighborhood dereliction, vandalism, vacancy, little lifestyle crimes like prostitution and panhandling, and broad crime increases. Neighborhood dereliction took the form of broken windows. To this day, the theory remains the basis for the zero-tolerance police efforts in places such as New York under the Giuliani administration and, to a lesser extent, Baltimore under former mayor Martin O’Malley. While it offered an effective if controversial approach for mayors looking to appear tough on crime, it was a lousy tool for predicting what sorts of crimes were going to take place and where and when they were likely to occur. Exactly how many windows have to be broken in your neighborhood before a crack dealer sets up on your corner?1
The Naked Future Page 21