Groupthink, where incorrect opinions take hold in communities, may be less of a problem than often claimed. Instead, collections of people are more likely to come to the correct conclusion when they share their knowledge with each other than when making decisions in isolation, a new study suggests.
In a world where data is expanding at a dizzying rate, social scientists are increasingly focusing on how people make decisions. If one of the great challenges of our age is how to best process the wealth of information we have, learning how we currently do it is essential.
One popular theory, widely publicized in the book The Wisdom of Crowds, holds that large masses of people are more likely to get to the right answer than so-called expert individuals. However, new research by the University of Pennsylvania's Dr Damon Centola adds a twist, undermining one of the theory's key elements.
Wisdom of crowds advocates argue that groups of people can only be trusted if making decisions independently. Those who talk to each other may be swayed by influential individuals, particularly anyone seen as an expert, and the wisdom transformed to groupthink.
There are certainly examples one can point to that appear to support this theory, most notably how commentators anticipating Hilary Clinton's election shaped public expectations, including in the betting markets. However, cherry-picking such cases is the antithesis of science, so Centola conducted experiments where people were placed into one of three organizational structures and asked to make numerical estimations of various quantities. His results have been published in Proceedings of the National Academy of Sciences.
In Centola's centralized groups, a single individual was connected to everyone, rather like a newspaper or television commentator, positioned to shape multiple opinions. “Egalitarian networks”, on the other hand, replicated social media, with all members given the opportunity to share their views with everyone else. Members of a control group made their judgments independently.
Each group got three chances to undertake challenges, such as estimating the number of calories in a plate of food, with the second and third rounds made after seeing what those they were connected to had said the previous times. Participants were paid by performance to ensure they tried.
Centralized groups got slightly worse on the later rounds, but egalitarian networks consistently improved. “While opinion leaders can sometimes improve things, they were statistically more likely to make the group worse off,” Centola said in a statement. "Where everyone is equally influential, people can help to correct each other's mistakes.”
Despite the artificial set-up, the result casts doubt on fears that online social networks are leading to disaster. Centola is testing his findings in the more realistic environment of doctors assessing hospital patients.