global network

A Web of Connections

A half-century ago, the social psychologist Stanley Milgram conducted an experiment that vividly demonstrated the power of social networks. Milgram sent packages to 160 random individuals living in Wichita, Kansas and Omaha, Nebraska, asking them to forward the items to someone they knew on a first-name basis, with the ultimate intent of getting the mail to a stockbroker in Boston. On average, the packages passed through the hands of five intermediaries before they reached the intended recipient.

The conclusion of the “Small World” experiment—the idea that there are “six degrees of separation” between any two individuals on earth—has been roundly criticized in the ensuring decades, but there’s no disputing that Milgram was on to something: networks can function in astonishing ways.

Networks—both social and otherwise—surround us today. Large societal networks such as electrical power grids, communication networks, and transportation networks are all overlayed on social networks. The structure and properties of these large societal networks are governed by decisions made by individuals in the social networks. Other networks are as old as time: the synapses that wire our brains, the biological relationships that govern ecosystems around the planet. And at the intersection of physical, biological, and social networks are multi-layer networks that influence consumer choices, business decision-making, and workplace behaviors. These are particularly interesting to network science researchers like Mani.

"In the last 10 years, there has been a lot of focus on understanding how network structures can be used for previously unconsidered purposes: controlling epidemics, designing better transportation systems, increasing physical activity, reducing waste. We are finding broader and beneficial applications for network design."

—Ankur Mani, ISyE Professor

Mani, who teaches courses on network science for ISyE, says companies and small businesses alike have long relied on social networks to help bring them customers and generate sales. (Many businesses, for example, offer rewards to customers who refer friends.) The Internet, of course, significantly expanded business interest in networks: Today, companies use algorithms to analyze networks and use the findings to set prices and hone their marketing efforts.

But understanding how networks function can be tricky, Mani says. Social behaviors are predictable, but also complex. “Lots of companies give discounts to influencers on Instagram in the hope of getting them to promote their products and drive sales,” Mani says. “But does it really work?” In a 2019 paper titled “The Value of Price Discrimination in Large Random Networks,” Mani and his coauthors examined the practice and found that as the influencers’ networks grew bigger, the less beneficial the programs were, considering their cost and impact. What’s more, Mani says, “In some cases such programs lead to distrust among customers. If I find out my friend is getting a lower price than I am, I’m not going to be very happy. Considering the cost of implementation, the extra benefit you achieve is minimal.”

To understand how social networks function, researchers have to assess not only the nodes (i.e. the individuals) in the network, but also how they link to each other. These links, sometimes called edges, indicate how the nodes relate, says ISyE professor Krishnamurthy Iyer. “If the network represents parent-child relationships, for example, then the links are asymmetric,” Iyer says. “To understand a network, you have to understand how the components relate to and influence each other.” If you receive a referral for a psychologist from a close friend, for example, are you more likely to use it than if it came from a neighbor? Whose advice would you take regarding a stock tip—a stranger’s or a stock-broker’s?

Networks can be huge and multi-layered. Our ability to study them in a systematic way is fairly new—but the more we learn, the more we can understand the underlying patterns of behavior.

—Krishnamurthy Iyer, ISyE Associate Professor

Social networks are sometimes blamed for the spread of bad phenomena, like fake news or Ebola. But Mani observes that social networks can also be used to create good outcomes, such as energy conservation and improved physical activity. Traditional approaches to change behavior would involve individual incentives (Pigouvian mechanisms). “But this would not be as efficient as using peer pressure,” says Mani, who has studied the effects of inducing peer pressure to promote cooperation, such as in his paper titled “Inducing Peer Pressure to Promote Cooperation.” The more effective solution, he says, is to come up with a system where citizens are more likely to benefit if they can get other citizens to adopt good habits. “The peer pressure mechanisms are more efficient than individual incentives,” Mani explains.

Our understanding of how networks function has grown significantly in recent years, Iyer says, and will likely accelerate as our ability to collect data and harness computing power increases. “Networks can be huge and multi-layered,” Iyer says. “Our ability to study them in a systematic way is fairly new—but the more we learn, the more we can understand the underlying patterns of behavior.”

Joel Hoekstra is a Minneapolis-based writer.

This article appeared in the 2019 ISyE magazine.