PHILADELPHIA — Conventional wisdom holds that it's not what you know, it's who you know. But now scientists studying networking are starting to realize that when it comes to much in life, it's also who the people you know know, and perhaps also who those people know.
Drawing from computer science, math, sociology and other disciplines, researchers are starting to figure out how those branching thickets of human social networks are shaping our tastes, our purchases, how we vote, and even our health and happiness.
At the University of Pennsylvania, Michael Kearns is using controlled voting experiments to show how a small minority view can win over an overwhelming majority.
Kearns, a computer scientist and expert on machine learning and game theory, examines the connections between networks and human behavior in settings as diverse as voting and the vulnerability of the Internet to terrorism.
People are also reading…
His human experiments and others like it could overturn our notion of the way trends and influence spread through society, said Duncan Watts, a physicist and networking expert at Yahoo.
Watts said the marketing field and much of the public have embraced the idea that humanity is run by a minority of well-connected "influentials" who help ideas spread like infectious viruses.
It's an idea popularized by books such as Malcolm Gladwell's "The Tipping Point." But nobody knows if it really works this way, Watts said. "For all this discussion about influentials and how they drive word-of-mouth, there's no empirical evidence — no real theory."
Penn's Kearns, he said, is starting to bring a more hard-science approach to bear on the issue.
For Kearns' most recently published experiment, Kearns created a network from a group of 36 subjects. He put each one at a work station linked to between two and 18 of the others.
They were asked to vote for red or blue. If everyone in the group could agree on the same color within one minute, everyone would get rewarded with money. If they failed to reach consensus, they would get nothing.
But he gave the subjects different preferences. Some were told they'd get paid $1.50 for each round that red won and only 50 cents if blue won. For others, the incentive was reversed.
"There's this tension between all of them wanting to collectively agree but selfishly wanting everyone to agree on their preferred color," he said.
One real-world analogy would be the recent Democratic presidential primaries, he said. Many voters passionately backed Hillary Rodham Clinton or Barack Obama, but worried that split opinion would cause the whole party to lose.
Behind the scenes, Kearns rigged the experiment in different ways, sometimes mixing up the incentives so that some students got only $1.25 for pushing their color on the group and 75 cents if they went the other way.
Despite the short deadline, he said, people came to some agreement in 55 out of 81 separate trials.
He found that sometimes a tiny minority could rule. In the most extreme cases, red won when only six subjects preferred it, the other 30 wanting blue. All the members of the minority needed was "influence" — that is, more connections within the group than the people they competed against.
" 'Influential' people can determine the outcome to their liking," Kearns said, even if the majority has a strong incentive to go the other way. In this case, having lots of connections made a subject influential.
