By Cass R. Sunstein, Professor at Harvard Law School
This post is part of a series on social media and democracy.
On balance, the question of whether social media platforms are good for democracy is easy. On balance, they are not merely good; they are terrific. For people to govern themselves, they need to have information. They also need to be able to convey it to others. Social media platforms make that tons easier.
There is a subtler point as well. When democracies are functioning properly, people’s sufferings and challenges are not entirely private matters. Social media platforms help us alert one another to a million and one different problems. In the process, the existence of social media can prod citizens to seek solutions.
Consider the remarkable finding, by the economist Amartya Sen, that in the history of the world, there has never been a famine in a system with a democratic press and free elections. A central reason is that famines are a product not only of a scarcity of food, but also a nation’s failure to provide solutions. When the press is free, and when leaders are elected, leaders have a strong incentive to help.
Mental illness, chronic pain, loss of employment, vulnerability to crime, drugs in the family – information about all these spread via social media, and they can be reduced with sensible policies. When people can talk to each other, and disclose what they know to public officials, the whole world might change in a hurry.
But celebrations can be awfully boring, so let’s hold the applause. Are automobiles good for transportation? Absolutely, but in the United States alone, over 35,000 people died in crashes in 2016.
Social media platforms are terrific for democracy in many ways, but pretty bad in others. And they remain a work-in-progress, not only because of new entrants, but also because the not-so-new ones (including Facebook) continue to evolve. What John Dewey said about my beloved country is true for social media as well: “The United States are not yet made; they are not a finished fact to be categorically assessed.”
For social media and democracy, the equivalents of car crashes include false reports (“fake news”) and the proliferation of information cocoons — and as a result, an increase in fragmentation, polarization and extremism. If you live in an information cocoon, you will believe many things that are false, and you will fail to learn countless things that are true. That’s awful for democracy. And as we have seen, those with specific interests — including politicians and nations, such as Russia, seeking to disrupt democratic processes — can use social media to promote those interests.
This problem is linked to the phenomenon of group polarization — which takes hold when like-minded people talk to one another and end up thinking a more extreme version of what they thought before they started to talk. In fact that’s a common outcome. At best, it’s a problem. At worst, it’s dangerous.
The Perils of Personalization
A little over a year ago, an important Facebook post stated, “The goal of News Feed is to show people the stories that are most relevant to them.” It drew attention to Facebook’s “ core values,” which require an emphasis on “what content is most important to you.”
The post emphasized, “Something that one person finds informative or interesting may be different from what another person finds informative or interesting.” That’s true. The post added, “as News Feed evolves, we’ll continue building easy-to-use and powerful tools to give you the most personalized experience.”
Really? I hope not. From the standpoint of democracy, that’s a nightmare.
Instead of creating “the most personalized experience,” consider three principles for the communications environment in a system that aspires to democratic self-government.
First, citizens should be exposed to materials that they would not have chosen in advance. Serendipity is a good thing. Unplanned, unanticipated encounters are central to democracy itself. Such encounters often involve topics and points of view that people have not sought out and perhaps find quite irritating – but that might change their lives in fundamental ways. They are important partly to ensure against fragmentation, polarization, and extremism, which are predictable outcomes of any situation in which like-minded people speak only with themselves.
Second, many or most citizens should have a wide range of common experiences. Without shared experiences, a heterogeneous society will have a much more difficult time in addressing social problems. People might see each other as strangers, foreigners, possibly even enemies. Common experiences, emphatically including the common experiences made possible by social media, provide a form of social glue. Societies need such things.
Third, citizens should be in a position to distinguish between truth and falsehood – and to know when democratic processes are being manipulated. In democracies, of course, it is fair for people to disagree about what the truth is. But if people are knowingly spreading lies, and if nations are attempting to disrupt other nations, some process should be in place to enable citizens to have access to the truth.
An Experiment in Colorado
To explore the issue of polarization, bear with me for a moment and consider a small experiment in democracy that I conducted with some colleagues over a decade ago. We brought about sixty American citizens together and
assembled them into groups, generally consisting of six people.
We couldn’t have known it at the time, but in a way, we were testing the same effect that social media has on the political process. Members of each group were asked to deliberate on three of the most controversial issues of the day: Should states allow same-sex couples to enter into civil unions? Should employers engage in affirmative action by giving a preference to members of traditionally disadvantaged groups? Should the United States sign an international treaty to combat global warming?
As the experiment was designed, some groups were “liberal” and others were “conservative” — drawn from residents of Boulder and Colorado Springs. (There was no mixing.) It is widely known that Boulder tends to be left-of-center and that Colorado Springs tends to be right-of-center. The groups were screened to ensure that their members conformed to these stereotypes. People were asked to state their opinions individually and anonymously (by writing them down in private) both before and after fifteen minutes of group discussion, and also to try to reach a public verdict before making their final anonymous statements as individuals.
The results were simple and disturbing. In almost every group, members ended up with more extreme positions after they spoke with one another. That’s group polarization in action.
To offer a little more detail: Discussion made same-sex unions more popular among liberals; discussion made such unions less popular among conservatives. Liberals favored an international treaty to control climate change before discussion; they favored it more strongly after discussion. Conservatives were neutral on that treaty before discussion; they strongly opposed it after discussion.
Mildly favorable toward affirmative action before discussion, liberals became strongly favorable toward affirmative action after discussion. Firmly negative about affirmative action before discussion, conservatives became even more negative about affirmative action after discussion.
The experiment also made both liberal and conservative groups more ideologically homogeneous — and thus squelched internal diversity. Even in their anonymous statements, group members showed far more consensus after discussion than before. It follows that discussion helped to widen the rift between liberals and conservatives on all three issues. Before discussion, some liberal groups were, on some issues, fairly close to some conservative groups. The result of discussion was to divide them far more sharply.
Here’s my point: Every minute of every day, the Colorado experiment is being replicated on social media, and in countless nations. Your Facebook friends may be a lot like one of the Colorado groups (only a lot bigger). On your Twitter feed, you might follow people who think like you do. As you read what they have to say, you’ll end up more entrenched in your position. For many users, social media platforms are creating the equivalent of the Colorado experiment, and with damaging results for democracy.
Of course that isn’t happening to everyone. Many social media feeds have a diversity of view; many people don’t use social media to engage about politics at all. But if something like the Colorado experiments is happening to hundreds of millions of people, we might not understand each other — and the project of self-government, which is always hard, will get a lot harder. In many nations, that’s happening today, and it will happen tomorrow. And while social media platforms are hardly responsible for the problem, they are not doing enough to help. What’s more, the problem is compounded if self-interested people, companies, and nations are knowingly spreading lies to those who – they think – are especially likely to believe them.
Better Architecture
These claims should not be misunderstood. By emphasizing the problems posed by knowing falsehoods, polarization, and information cocoons, I do not mean to suggest that things are worse now than they were in 1960, 1860, 1560, 1260, or the year before or after the birth of Jesus Christ. Information cocoons are as old as human history. The concern is not that things are getting worse. It is that the increased technological capacity for self-sorting and for personalization is creating serious problems. What social media platforms do is to make certain kinds of targeting and certain kinds of self-sorting, and especially self-sorting among hundreds, thousands, or millions of strangers a lot easier — easier than it has ever been. We have had plenty of targeting and echo chambers before, but targeting people who are especially likely to believe specific falsehoods, and one-click echo chambers, are something new.
Nor do I mean to suggest that with respect to polarization, social media are worse than newspapers, television stations, social clubs, sports teams, or neighborhoods. Empirical work continues to try to compare various sources of polarization, and it would be reckless to suggest that social media do the most damage. Countless people try to find diverse topics, and multiple points of view, and they use their Facebook pages and Twitter feeds for exactly that purpose. But still, countless people don’t.
The good news is that social media platforms are hardly a finished fact to be categorically assessed. They are very much a work in progress.
The uses of Facebook, Twitter, Instagram, and others depends on a variety of factors, including individual choices, relevant algorithms, social norms, and the architectural judgments of the platform designers themselves. A News Feed, or anything like it, can promote information cocoons, or discourage them. Platforms can provide safeguards in the event that democratic processes are being intentionally disrupted or harmful falsehoods are spreading; it can help people find out what is true. (Of late, Facebook has been doing exactly that .)
It could do much more. It could continue to focus on reducing personalization and more on producing information to people, expanding their horizons and potentially counteracting polarization.
It could find ways to work on issues of civility. It could promote a focus on substance, rather than who is on whose team, or which team is best. It could change its News Feed in ways that could combat rather than promote fragmentation. Recent work such as “Related Articles” points in exactly the right direction, with the inspiring, even defining words: “One of our main goals is to support an informed community on Facebook.” There’s much more to consider, but the recent work is an impressive start.
Much more is being done in this vein. To take just one example, Read Across the Aisle , an app, gives you an assortment of diverse news sources. As you read, it tells you whether you are going “blue” or “red” — and the slider moves more to the left, or to the right, if you’re living in an echo chamber. The whole goal is to help people to live outside it, and to escape any kind of filter bubble.
In the coming years, we will inevitably see a lot of experiments designed to help social media to counteract the recent threats and to make democracy work better. No one can predict what is on the horizon. That’s excellent news – and it has the advantage of being true.
As with automobiles, so with social media: We’re a lot better off with them than without them, but aggregate judgments are an obstacle to improvement. So John Dewey gets the last word: “I would not minimize the advance scored in substitution of methods of discussion and conference for the method of arbitrary rule. But the better is too often the enemy of the still better.”
Cass R. Sunstein is the Robert Walmsley University Professor at Harvard University. He is the author of many books, including #Republic (2017), Nudge: Improving Decisions About Health, Wealth, and Happiness (with Richard H. Thaler 2008), and Simpler (2013). AMONG his latest books are The World According to Star Wars (2016) and The Ethics of Influence (2016). The author has occasionally served as a consultant for Facebook on issues connected to those in this essay.