Everyone knows you want to tuck skinny jeans into your boots. Ten years ago, everyone knew you wanted bootcut jeans that went over your boots. But How does anyone know these things? How is one option selected as the default out of all possible alternatives and then universally accepted?
The genesis and emergence of social conventions has long fooled cognitive scientists, sociologists, linguists and philosophers. Prominent ideas assume that institutionalized mechanisms, such as centralized authority or incentives for collective bargaining, are necessary to allow shared conventions to take hold. In contrast, newer socio-evolutionary ideas have suggested that networks of locally interacting individuals can spontaneously and inadvertently organize themselves to achieve global coordination, even in the absence of formal institutions.
This kind of self-organization is very difficult to demonstrate, especially on a meaningful scale. Now a mathematician and a sociologist have teamed up to show that global social conventions can, in fact, arise spontaneously from local interactions, even though the people involved have no idea they are coordinating anything. There is only one condition: the people must be hyperconnected.
The researchers recruited people from the Internet to play The Name Game. Participants sat in front of a computer screen and were shown a photo of a blonde woman, whom they had to name within twenty seconds. They were matched with a partner they had no knowledge of. If the two partners agreed on a name for the woman on their screens, they each got fifty cents; if they did not, they were docked a quarter each. Then they moved on to another round with a new partner. While they were rewarded for local coordination, they had no incentives for – or information about – any sort of global consensus. A standard game was thirty rounds.
Three types of social networks were tested. One was a spatial network, in which partners were chosen only from neighbors or nearest neighbours; another was a random network, in which partners were not spatially limited, but limited to the same four degrees of separation as the spatial network; the latter was a homogeneous network in which all bets were off – anyone could be partners with anyone else. Participants knew nothing about the structure of the network they were in.
Participants in the spatial and random networks formed groups of agreement early on, but results were limited. At the end of the game, the most popular name was used by up to forty-five percent of the participants, while up to seventy-five percent of the partners agreed.
Success came more slowly on the homogeneously mixed network, but when it arrived it was more robust. Because people in this network were less likely to associate with the same partner repeatedly, insular “neighborhoods” could not develop using a particular name. This local failure only served to boost global coordination; one name was always widely accepted in this network, usually around twenty-two. Partners who had never interacted at this point could still get it. And this result held true, on the same timescale, when this network size was doubled to ninety-six participants.
The authors think their results “will be of interest to researchers exploring the effects of online connectedness on the emergence of new political, social and economic behaviors” and think it will be interesting “to examine the practical implications of the unintended effects of increasing social connectedness to the homogenization of behaviors and beliefs among large numbers of individuals who are not even aware that they are implicitly coordinating with each other.”
And that’s how we know which style of jeans to buy next year.
PNAS2014. DOI: 10.1073/pnas.1418838112 (About DOIs).