“Are you embedded in your social surroundings?” This is the “core” question, says Alex “Sandy” Pentland, head of the Human Dynamics group at MIT Media Lab.
It may seem like an odd question for a technologist who has had a major influence on the development of the Fitbit and Google Glass, devices that insert technology between us and the world. Pentland, however, argues it’s a natural extension of his 40 years of experience in the field.
He believes that if technology made it possible for people to know how much, or how little, they invested in each other and their communities, they might give more priority to what’s important (other people) and less to what they think is important in the moment (work, bingeing Netflix). That data would inform not just how people plan their weekends or workouts, but also how they design cities and businesses to maximize belonging and safety—factors his team has found contribute significantly to productivity, innovation, and perhaps most importantly, happiness.
Pentland has spent 40 years developing tools to study how groups work. By using sophisticated mobile sensors and big data to detect hidden patterns of human interaction, he says businesses, governments, and individuals can design for more meaningful interaction. “I liken it to Jane Goodall watching the apes in the trees,” he says. “Does everybody get up at the same time? There are all these patterns we humans are not that conscious of but they dictate a huge fraction of our lives.”
These simple yet profound questions about the nature of our communities have serious implications for our health and well-bring. The virtue of using wearable tech to answer them is that rather than ask people what they do, or bring them into a lab to study them like specimens, sensors track what people do, and with whom.
Humans have always strived to measure things about themselves: IQ, REM sleep, steps taken, heart rate, and likes on Instagram. Pentland has a more radical proposition: study how we act in groups to unpack who we are as humans. “We find that 50-90% of behavior is habitual and predicted by your network of social relationships,” he explains. “Most of human behavior fits into the local community—your community—what you value and choose, so studying groups is important because that accounts for the majority of what you do.”
This requires new metrics. He draws an analogy to child development, where there are domains designed to measure how children are developing, including cognitive, social, emotional, and motor skills. He says adults also need domains for social connection: measuring how much they invest in their immediate communities and how much they explore new ones.
Investing in social capital
Pentland says there are two kinds of social capital that people need, the stuff that has bound together human societies for time eternal: bonding capital, which ties you to your nuclear group, and bridging capital, which connects you to the world and new ideas.
There is tension between the two types of capital. “This is one of the fundamental conflicts of the human nature,” he says. “People need that tribe that reinforces them, but new opportunities and things come from exploring,” he said.
The people who build up bridging capital by pushing themselves out of their comfort zones and getting away from the tribe tend to do better in life (pdf). They learn different perspectives, see more opportunities, and make more money. “They just live better,” Pentland says. “But it is uncomfortable.”
The key is to invest in both types of capital. But few are aware of how they divide their time between the two (or that they even exist). Pentland’s hope is that awareness of how people live might incentivize them to live better, so he’s spent 20 years developing metrics to quantify social connections.
Just as the Fitbit fostered a maniacal focus on steps taken and calories burned, a social Fitbit would warn against excessive insularity, or remind us to bridge out into new territory. Brexit, the rise of nationalism and the increasing divisiveness of political and civil life suggests we might need some new tools. “We have a decrease in the bridging capital as well as the bonding capital, partly because we are more electronic than we used to be, more spread out, and because things move faster,” Pentland says.
The pace of our lives, in other words, makes it hard to see the big picture. “People always forget when things move fast, you are always out of kilter, you don’t know what to do, you don’t have time to build up that trust and belonging,” he says. But societies and democracies sort of depend on it.
Plenty of people, from evolutionary biologists to doctors, sociologists, and psychologists extol the virtues of social connection, but few have developed a good way to measure it. Having someone conspicuously observe a group, not surprisingly, can change how people act.
Pentland’s interest in all this goes back to 1973, when he was working part-time for NASA’s Environmental Research Institute. He was charged with tracking Canadian beavers from space, but since beavers are small and satellite imagery back then wasn’t great, he honed in on beaver ponds, which were visible from orbit. Since beaver families build their own ponds, he figured you could glean a lot of information from the ponds. Humans are also social creatures who communicate, cooperate, and create safety together: what we do in groups says a lot about who we are.
Over the years, Pentland’s lab pioneered the invention of “sociometers,” credit card-sized sensors with microphones, GPS, and other sensors that people wear around their necks to measure the diversity and density of interactions in a group. They track the percentage of time talking, energy level of voices, speaking rates, interruptions, how much one vocal pattern mimics another, and more. In studies with banks, call centers, and during salary negotiations and business pitches, sociometers can predict things like which groups have a strong sense of of belonging and trust (measured by frequency and nature of interaction), which in turn predicts innovation and productivity. In other words, it’s the group dynamic—not the number of Harvard graduates—that generates corporate success.
In recent years, Pentland has focused on how to measure bridging and bonding capital. He and a team recently designed a measure of trust (a type of bonding capital) using a group of families at MIT. The group comprised 130 adults in a young-family community which he and some students tracked in 2010. “Tracked” is a bit of an understatement, as described in papers published about it:
The dataset includes continuous collection of over 25 phone-based signals— including location, accelerometry, Bluetooth-based device proximity, communication activities, installed applications, currently running applications, multimedia and file system information, and additional data generated by our experimental applications. In addition, we collect financial information through receipts and credit card statements, logs of Facebook socialization activities, daily polling of mood, stress, sleep, productivity, and socialization, as well as other health and wellness related information, standard psychological scales like personality tests, and many other types of manually entered data by the participants.
In the trust exercise, the researchers asked individuals in the group three questions (where “X” is the other participant in the experiment):
- Would you ask person X for help in sickness?
- Would you ask person X for a $100 loan?
- Would you ask person X for babysitting?
The results could be mapped out like this, with each person a node in the network:
When they looked at the data collected by devices, people who spoke on the phone with each other a lot were the same people who said they trusted each other when asked the three questions, with an average 94% accuracy. “We could predict the answers based on whether people had active reciprocal connections,” Pentland explains.
That people who talk on the phone trust each other hardly seems like rocket science. But the exercise helps to identify reliable and quantifiable measures. “The reason to choose to measure social capital is that some studies have shown that social connections, among various non-economic factors, are among the most influential factors on human well-being,” according to a paper produced as part of the project.
Privacy is clearly an issue. Sociometers were not developed to track people all the time, and they record voice and speech patterns instead of actual words. Social capital measures focus on individuals (or pairs), not groups.
Pentland has worked with the World Economic Forum to develop sophisticated data guidelines. He does not minimize the risks of having such troves of highly personal data unprotected. “It can be a little Black Mirror, right?” he says referring to the dystopian sci-fi televisions series. “To avoid autocratic control, citizens must have effective, direct control of data about themselves,” he wrote in a post for the World Economic Forum.
Will it matter?
By measuring both bonding capital (belonging, trust) and how often you speak to different communities (getting outside the silos) you are confronted with data about how you segregate your life. It’s far from perfect: someone from London who spends time in New York and Los Angeles may think they are building up bridging capital, when in fact the people they meet are more similar to them than people from different classes or backgrounds in the city where they live.
“These two ideas—bridging and bonding —are sort of fuzzy,” Pentland admits. “We haven’t worked out a mathematical formula for them.” But he says we know what’s important: “to seek people different than you, and not just see them but have a relationship with them, to walk in the other person’s shoes.”
We don’t design for diversity but rather concentration.
Pentland worries that the rise in polarization and echo chambers is in part due to our lack of awareness of how little bridging we do (humans are remarkably self-delusional). That’s in part due to the fact that we don’t design for diversity but rather concentration. Facebook and the New York Times seek to maximize like-minded thinkers, to concentrate them. But if we want to build and sustain rich communities, diversity is necessary. A test Pentland sometimes deploys to test this is to ask, “Do you know anybody who owns a pickup truck?” he says. “It’s the number-one selling vehicle in America, and if you don’t know people like that, that tells me you are out of touch with more than 50% of America,” he told the Edge.
While research is resolute on the benefits of exercise, which has paved the way for wearable technology to measure every step, heart beat, and calorie burned, the evidence is also clear that humans need each other.
When Harvard researchers studied a cohort of men for 90 years—tracking every life activity from jobs, kids, and spouses to weight, alcohol intake, and income—they found the primary determinant of their well being was the quality of their social interactions. Research shows that people who had weaker social ties had a 50% increased likelihood of dying early than those with stronger ones. Being disconnected, it showed, posed danger comparable to smoking 15 cigarettes a day, and was more predictive of early death than the effects of air pollution or physical inactivity.
That’s where Pentland’s devices come in. Having a social Fitbit won’t instantly make people happier or healthier, any more than millions of exercise-tracking devices already do. Humans know they need to do a lot of stuff that they don’t—sleep eight hours a night, eat fewer Pringles, drink less gin, do more yoga.
But even if we choose not to address our balance of bridging and bonding capital, it’s harder to ignore the nature of our community connections when we know precisely how much or little we are doing. “The one people most commonly fail to get is bridging capital,” Pentland says. “They live very segregated lives.” Evidence that you have not spoken to a new person in over a month should be as alarming as how few vegetables you eat or how few steps you take in a day.