It’s only natural to think that a revolutionary communications technology like the internet will help break down barriers between people and bring the world closer together. But that’s not the only scenario, or even the most likely one. The internet turns everything, from knowledge-gathering to community-building, into a series of tiny transactions – clicks – that are simple in isolation yet extraordinarily complicated in the aggregate. Research shows that very small biases, when magnified through thousands or millions or billions of choices, can turn into profound schisms. There’s reason to believe, or at least to fear, that this effect, inherent in large networks, may end up turning the internet into a polarizing force rather than a unifying one.
In a 1971 article titled “Dynamic Models of Segregation,” Thomas Schelling, winner of the 2005 Nobel Prize for economics, offered a fascinating reappraisal of the segregation of communities along racial lines, illustrating the way biases are magnified through a kind of network effect. If asked what lies behind racial segregation, most of us would likely point to prejudice and discrimination. But Schelling, through a simple experiment, showed that extreme segregation may have a much more innocent cause. Mark Buchanan summarized Schelling’s findings in his 2002 book Nexus:
Schelling began by imagining a society in which most people truly wish to live in balanced and racially integrated communities, with just one minor stipulation: most people would prefer not to end up living in a neighborhood in which they would be in the extreme minority. A white man might have black friends and colleagues and might be happy to live in a predominantly black neighborhood. Just the same, he might prefer not to be one of the only white people living there. This attitude is hardly racist and may indeed be one that many people – black, white, Hispanic, Chinese, or what have you – share. People naturally enjoy living among others with similar tastes, backgrounds, and values.
Nevertheless, innocent individual preferences of this sort can have startling effects, as Schelling discovered by drawing a grid of squares on a piece of paper and playing an illuminating game. On his grid, he first placed at random an equal number of black and white pieces, to depict an integrated society of two races mingling uniformly. He then supposed that every piece would prefer not to live in a minority of less than, say, 30 percent. So, taking one piece at a time, Schelling checked to see if less than 30 percent of its neighbors were of the same color, and if this was the case, he let that piece migrate to the nearest open square. He then repeated this procedure over and over until finally no piece lived in a local minority of less than 30 percent. To his surprise, Schelling discovered that at this point the black and white pieces not only had become less uniformly mixed but also had come to live in entirely distinct enclaves. In other words, the slight preference of the individual to avoid an extreme minority has the paradoxical but inexorable effect of obliterating mixed communities altogether.
Buchanan sums up the lesson of Schelling’s experiment: “Social realities are fashioned not only by the desires of people but also by the action of blind and more or less mechanical forces – in this case forces that can amplify slight and seemingly harmless personal preferences into dramatic and troubling consequences.” (You can download a piece of Windows-only software to perform the Schelling experiment yourself.) In the real world, with its mortgages and schools and jobs and moving vans, the “mechanical forces” of segregation move fairly slowly; there are brakes on the speed with which we pull up stakes and change where we live. In internet communities, there are no such constraints. Making a community-defining decision is as simple as clicking on a link – adding a feed to your blog reader, say, or a friend to your social network. Given the presence of a slight bias to be connected to people similar to ourselves, the segregation effect would thus tend to happen much faster – and with even more extreme consequences – on the internet.
This is all theoretical, of course, but it’s easy to see how it follows logically from Schelling’s findings. And there is other evidence that the Internet may end up being a polarizing force. In a recent academic paper, called “Global Village or Cyber-Balkans? Modeling and Measuring the Integration of Electronic Communities,” Eric Brynjolfsson, of MIT, and Marshall Van Alstyne, of Boston University, describe the results of a model that measured how individuals’ online choices influence community affiliation. “Although the conventional wisdom has stressed the integrating effects of [internet] technology,” they write, in introducing their study, “we examine critically the claim that a global village is the inexorable result of increased connectivity and develop a suite of formal measures to address this question.”
They note that, because there are limits to how much information we can process and how many people we can communicate with (we have “bounded rationality,” to use the academic jargon), we naturally have to use filters to screen out ideas and contacts. On the internet, these filters are becoming ever more sophisticated, which means we can focus our attention – and our communities – ever more precisely. “Our analysis,” Brynjolfsson and Van Alstyne write, “suggests that automatic search tools and filters that route communications among people based on their views, reputations, past statements or personal characteristics are not necessarily benign in their effects.” Diversity in the physical world “can give way to virtual homogeneity as specialized communities coalesce across geographic boundaries.”
They stress that “balkanization” is not the only possible result of filtering. “On the other hand,” they write, “preferences for broader knowledge, or even randomized information, can also be indulged. In the presence of [information technology], a taste for diverse interaction leads to greater integration – underscoring how the technology serves mainly to amplify individual preferences. IT does not predetermine one outcome.” Nevertheless, they write that their model indicates, in an echo of Schelling’s findings, that “other factors being equal, all that is required to reduce integration in most cases is that preferred interactions are more focused than existing interactions.” If, in other words, we have even a small inclination to prefer like-minded views and people, we will tend toward creating balkanized online communities.
Such fragmentation of association tends to lead to an ever-greater polarization of thinking, which in turn can erode civic cohesiveness, as the authors explain:
With the customized access and search capabilities of IT, individuals can focus their attention on career interests, music and entertainment that already match their defined profiles, and they can arrange to read only news and analysis that align with their preferences. Individuals empowered to screen out material that does not conform to their existing preferences may form virtual cliques, insulate themselves from opposing points of view, and reinforce their biases. Authors of collaborative filtering technology have long recognized its ability to both foster tribalism as well as a global village.
Indulging these preferences can have the perverse effect of intensifying and hardening pre-existing biases. Thus people who oppose free trade are likely, after talking to one another, to oppose it more fiercely; people who fear gun control appear, after discussion, more likely to take action; and juries that want to send a message seem, after deliberation, to set higher damage awards. The reasons include information cascades and oversampled arguments. In one, an accumulating, and unchallenged, body of evidence leads members to adopt group views in lieu of their own. In the other, members of a limited argument pool are unwilling or unable to construct persuasive counterarguments that would lead to more balanced views. The effect is not merely a tendency for members to conform to the group average but a radicalization in which this average moves toward extremes.
Increasing the number of information sources available may worsen this effect, as may increasing the attention paid to these information sources … Internet users can seek out interactions with like-minded individuals who have similar values and thus become less likely to trust important decisions to people whose values differ from their own. This voluntary balkanization and the loss of shared experiences and values may be harmful to the structure of democratic societies as well as decentralized organizations.
It’s too early in the history of the internet to know whether this disturbing scenario will come to pass, a point that the authors emphasize. But we need only look at, say, the tendency toward extremism – and distrust of those holding opposing views – among the most popular political bloggers to get a sense of how balkanization and polarization can emerge in online communities. Brynjolfsson and Van Alstyne end on this note: “We can, and should, explicitly consider what we value as we shape the nature of our networks and infrastructure – with no illusions that a greater sense of community will inexorably result.” Personally, I’m even more fatalistic. I’m not sure we’ll be able to influence the progression of internet communities by tinkering with “our networks and infrastructure.” What will happen will happen. It’s written in our clicks.