Thomas Schelling, polarization and the web


Thomas Schelling has died. Schelling’s pathbreaking work in game theory had enormous influence during the Cold War and ultimately earned him a Nobel Prize. It also helps illuminate some of the unexpected consequences of the internet as a medium for information-gathering and conversation — in particular the technology’s tendency to breed ideological polarization (a tendency that shaped political discourse during the recent presidential campaign). In my 2008 book The Big Switch, I discussed how one of Schelling’s papers, “Dynamic Models of Segregation,” holds important lessons for making sense of social dynamics online:

In 1971, the economist Thomas Schelling performed a simple experiment that had a very surprising result. He was curious about the persistence of extreme racial segregation in the country. He knew that most Americans are not racists or bigots, that we’re generally happy to be around people who don’t look or think the same way we do. At the same time, he knew that we’re not entirely unbiased in the choices we make about where we live and whom we associate with. Most of us have a preference, if only a slight one, to be near at least some people who are similar to ourselves. We don’t want to be the only black person or white person, or the only liberal or conservative, on the block. Schelling wondered whether such small biases might, over the long run, influence the makeup of neighborhoods.

He began his experiment by drawing a grid of squares on a piece of paper, creating a pattern resembling an oversized checkerboard. Each square represented a house lot. He then randomly placed a black or a white marker in some of the squares. Each marker represented either a black or a white family. Schelling assumed that each family desired to live in a racially mixed neighborhood, and that’s exactly what his grid showed at the start: the white families and the black families were spread across the grid in an entirely arbitrary fashion. It was a fully integrated community. He then made a further assumption: that each family would prefer to have some nearby neighbors of the same color as themselves. If the percentage of neighbors of the same color fell beneath 50 percent, a family would have a tendency to move to a new house.

On the internet, making a community-defining decision
is as simple as clicking a link.

On the basis of that one simple rule, Schelling began shifting the markers around the grid. If a black marker’s neighbors were more than 50 percent white or if a white marker’s neighbors were more than 50 percent black, he’d move the marker to the closest unoccupied square. He continued moving the pieces until no marker had neighbors that were more than 50 percent of the other color. At that point, to Schelling’s astonishment, the grid had become completely segregated. All the white markers had congregated in one area, and all the black markers had congregated in another. A modest, natural preference to live near at least a few people sharing a similar characteristic had the effect, as it influenced many individual decisions, of producing a dramatic divide in the population. “In some cases,” Schelling explained, “small incentives, almost imperceptible differentials, can lead to strikingly polarized results.”

It was a profound insight, one that, years later, would be cited by the Royal Swedish Society of Sciences when it presented Schelling with the 2005 Nobel Prize in Economics. Mark Buchanan, in his book Nexus, summarized the broader lesson of the experiment well: “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.”

Just as it’s assumed that the Internet will create a rich and diverse culture, it’s also assumed that it will bring people into greater harmony, that it will breed greater understanding and help ameliorate political and social tensions. On the face of it, that expectation seems entirely reasonable. After all, the Internet erases the physical boundaries that separate us, allows the free exchange of information about the thoughts and lives of others, and provides an egalitarian forum in which all views can get an airing. The optimistic view was perhaps best expressed by Nicholas Negroponte, the head of MIT’s Media Lab, in his 1995 bestseller Being Digital. “While the politicians struggle with the baggage of history, a new generation is emerging from the digital landscape free of many of the old prejudices,” he wrote. “Digital technology can be a natural force drawing people into greater world harmony.”

But Schelling’s experiment calls this view into question. Not only will the process of polarization tend to play out in virtual communities in the same way it does in neighborhoods, but it seems likely to proceed much more quickly online. In the real world, with its mortgages and schools and jobs, the mechanical forces of segregation move slowly. There are brakes on the speed with which we pull up stakes and move to a new house. Internet communities have no such constraints. Making a community-defining decision is as simple as clicking a link. Every time we subscribe to a blog, add a friend to our social network, categorize an e-mail message as spam, or even choose a site from a list of search results, we are making a decision that defines, in some small way, whom we associate with and what information we pay attention to. Given the presence of even a slight bias to be connected to people similar to ourselves—ones who share, say, our political views or our cultural preferences—we would, like Schelling’s hypothetical homeowners, end up in ever more polarized and homogeneous communities. We would click our way to a fractured society.

Image: “Checkers” by Emily Cline.