New research shows how the people we are in social contact with can influence us for better – and worse. Governments should take heed.
From healthcare to climate change, governments today face a range of problems in which they must persuade people to change their behaviour. But maybe instead of relying on their powers of persuasion, governments should consider using the science of social networks as a tool to do this. True, many claims for the power of these are based on the hype surrounding major social networking websites. But the basic idea is simple: people can join together in groups with particular patterns of social links, and these patterns then have important effects on the way they behave.
The shape of these networks has surprising effects. Take an unlikely example: stage musicals. Brian Uzzi is a sociologist at Northwestern University in Chicago. He is also a big fan of musicals. Musicals have been big business for decades, but investors have to guess which shows will be a hit. Bye Bye Birdie, a profitable 1960 production, ran for 607 nights. Bring Back Birdie, its 1981 sequel, closed after just four. Intrigued, Uzzi used network science to find out why.
He investigated 321 American musicals that were launched between 1945 and 1989, paying particular attention to whether the team of director, producers, choreographers and writers had worked together before. After crunching the statistics, he discovered something remarkable. Teams who had never worked together, perhaps unsurprisingly, fared poorly: their ‘weak’ networks meant a lack of creative vision, and lots of failures. And at the other extreme, teams that had worked together successfully also tended to produce flops. Sometimes, lacking outside creative input, the team just rehashed the same ideas that worked the last time; sometimes, lacking newcomers, they ‘developed’ their vision in unwise ways. Either way, lightning rarely struck twice. But, in between, Uzzi found a point of balance. Groups with exactly the right mix of new and old participants reliably produced hits. This variation in the ‘density’ of the ties allowed easy communication and fostered greater creativity; new ideas from the outsiders meshed with the experience of the insiders. It made no difference whether a musical was about cats or roller-skating, or who starred in it – what mattered was the nature of the network binding its team together.
These insights have some fascinating consequences for policymaking-particularly in the area of health. According to network science, teenagers are more likely to adopt unhealthy eating habits if their friends’ friends consume junk food, even if they don’t personally know those individuals. Conversely, they may be encouraged to eat more healthily if their immediate friends do. This opens up opportunities for governments to reduce spending. For example, suppose a health authority has £100 and wants to improve the diets of ten people. If it invests £10 in each individual separately, it might only influence one or two. However, by investing the entire sum in persuading one or two key individuals who are central to a social network, the authority could trigger a cascade effect, influencing the entire group for a fraction of the cost. The same principle could be applied to other areas, such as encouraging physical exercise or quitting smoking.
However, the influence of networks is not always positive. The same mechanisms that can spread healthy behaviours can also amplify harmful ones. Financial panic, political extremism, and the rapid proliferation of unhealthy fads can all be accelerated by tightly-knit social networks. This presents a significant challenge for policymakers: how to harness the power of social connectivity for good while mitigating its potential for harm. Understanding the precise structure of these networks—identifying the key influencers and the strength of the connections between people—is therefore critical. Without this map of human relationships, any intervention risks being inefficient or could even produce unintended negative consequences.
