Bubbles as Clusters in opposing Network Cascades
https://www.theatlantic.com/business/archive/2017/01/america-bubbles/514385/
It has become more and more obvious that we exist in bubbles. Most obviously, we exist in political bubbles which is cited as a big reason why the outcome of the most recent election was so surprising to so many people. Besides politics, American exist in bubbles in almost every aspect of their lives. Even at Cornell, I’ve noticed alike people tend to hang out in cliques. The article attributes this clustering to a sociological concept called “homophily”. This means that people seek similarities in their lives, which includes in social circles, workplaces and even marriages. Also, even before people make decisions themselves that lead to bubble behavior, they are likely already in a bubble. Factors such as socioeconomic status, education and political views determine one’s course throughout life and these factors tend to connect people who are alike rather than different. This phenomenon makes it even easier for people to bubble. Finally, bubbling is most common in smaller groups where connections are more important. In scenarios where individuals get to choose many participants, like hiring for a job or admissions to schools, they are more “fair” and “don’t mind” selecting some people who don’t share similarities with them. However, in smaller cases where there is a selection of smaller size, people tend to choose others like them because it matters more. Unfortunately, this “small club” phenomenon is has a direct impact on all our lives, since the more elitist, powerful or prestigious a club is the smaller and more exclusive it will be. People in these positions are the most able to effect change and in a better world they would be more diverse.
Much of this behavior can actually be explained by cascading behavior in network relations. We assume that there is benefit in homogenous relations and smaller or no benefit in heterogenous relations. If we model this as a relation network with ideologies A and B we can obviously see that people want to have connections with others with similar ideologies. Since A-A and B-B relations offer more benefit than A-B and B-A relations, people gain more from being in a bubble or cluster than not. Furthermore, this is more impactful in smaller groups. Consider a person as a node and connections as edges. For a node with many edges to homogenous nodes, the impact of one heterogeneous connection is minimal. However, if this node has only a few connections, heterogeneous connections are take away more proportionally from total value in this scenario than in the previous one. This explains the small-club behavior above. Finally, the reason bubbles will likely stay bubbles has to do with cluster density. Because there is only benefit so relations in clusters and it is extremely easy to form these relations, bubbles modeled as networks often have a high density. Thus, this prevents cascade behavior because it requires a lower conversion threshold for the cascade to happen. While this can be seen as a problem, the solution lies in making more heterogeneous connections. We have assumed that there still is benefit to these connection, just not as much homogeneous ones. It is understandable to choose homogeneous connections over heterogeneous ones, but that should not stop an individual from making these connections separately. These relationships do provide value and if more people make them they would reduce the density of clusters. As density is reduced, conversion thresholds don’t need to be as high and people see more benefit from considering alternative viewpoints. This encourages more complete information and reasoning, which ultimately leads to a more whole society.