This article appeared in The New York Times "Opinion"
On social media, encountering opposing views can make people become even more wedded to their own.
By Christopher A. Bail
During his appearance before Congress on Wednesday, Twitter’s chief executive, Jack Dorsey, repeatedly denied that Twitter’s algorithms are biased against conservative voices. His denials echoed recent statements he has made about the importance of exposing people to opposing political views. Indeed, he announced last month that Twitter was experimenting with new features that would actively expose people to such views.
Mr. Dorsey’s goal of reducing political polarization is commendable. But his proposed solution — disrupting our social media “echo chambers” — may actually make things worse. Forcing Twitter users to encounter political views they disagree with, my research shows, can make them become even more wedded to their own.
Scholars were once optimistic that social media could increase bipartisan dialogue by allowing virtually anyone to engage in public debate about politics. Yet mounting evidence suggests that Facebook and Twitter have allowed Republicans and Democrats to further segregate themselves. A recent study indicates that 85 percent of retweets are made by people who share a political orientation.
The link between social media and political polarization, however, presents a classic chicken-or-egg question: Do the accounts we follow on platforms such as Twitter shape our political views, or do we mostly follow accounts that reflect our views? We cannot know whether social media is a cause or a symptom of our deeply divided politics, unless, of course, we conduct a controlled experiment.
In a study that was published last month in the journal Proceedings of the National Academy of Sciences, my colleagues and I did just that. We surveyed more than 1,200 Twitter-using Republicans and Democrats about their political views. Then we paid half of them to follow for one month a bot we created that retweeted messages from elected officials and other opinion leaders from the other political party.
Instead of reducing political polarization, being exposed to opposing ideas increased it. Republicans who followed a Democratic bot for one month expressed social policy views that were substantially more conservative at the conclusion of the study. Democrats who followed a Republican bot exhibited very slight increases in liberal attitudes about social issues, but those effects were not statistically significant.
Why did some social media users’ political views become more entrenchedafter we disrupted their echo chambers? One possibility is the structure of Twitter itself. Social psychologists have long argued that positive, intimate contact between members of rival groups across an extended period can produce compromise. But that is not what Twitter offers. Its character limits — combined with the anonymous, spontaneous nature of so many exchanges on the platform — simply may not be conducive to mutual understanding.
No single solution will reduce political polarization on social media. But a first step should be for Twitter to experiment with removing its character limits. Allowing people to voice their opinions in detail will not improve the civility of discourse by itself, but it may facilitate a better competition of ideas and increase the possibility for Democrats and Republicans to understand one another.
Second, as our research indicates, Twitter should not force its users to view messages from a political party they oppose. Instead, it should create an alert system that makes people aware when they are being exposed predominantly to one point of view. The most pernicious effect of social media echo chambers may be that most people are unaware of how much their political views are influenced by selective exposure to information.
Finally, if Twitter is resolved to expose users to opposing political views, it should focus on doing so with specific issues. Republican and Democratic Twitter users appear unready to have broad conversations about politics. But breaking up the echo chambers that prevent cross-party discussion about market-based solutions to climate change, for example, might be more successful.
Some argue that social media sites will never provide the type of forum necessary to reduce political polarization and that we should focus on offline dialogue. Yet as social media continues to grow as a forum for political debate in our country, such solutions are not practical. We need to try something. But disrupting our Twitter echo chambers will most likely backfire.
Christopher A. Bail is an associate professor of sociology and public policy at Duke University.
A version of this article appears in print on Sept. 9, 2018, on Page SR2 of the New York edition with the headline: Please, Twitter, Don’t Do It.