It鈥檚 the new politics. As the UK and other EU countries have been gearing up for the European parliamentary elections, Facebook was once again flooded with misinformation. In response, it has shut down dozens of suspicious pages and groups.
But new accounts can pop up quicker than old ones are shut down, so another approach to reduce the damage is to try to prevent misinformation from reverberating in so-called echo chambers 鈥� segregated networks of people with polarized opinions. A new analysis has found a few simple tweaks that could make a bit difference.
By simulating the making and breaking of connections between users of a social network and watching what that does to the flow of information, Kazutoshi Sasahara at Indiana University in Bloomington, USA, and his colleagues have shown that the structure of social networks makes echo chambers inevitable.
Advertisement
Data from Twitter, including political tweets from the 2010 US midterm elections, backs up out prediction that most networks of people will rapidly descend into isolated, like-minded communities, says Sasahara.
But the simulation also revealed a simple fix. The team found that the key mechanism that encourages the formation of echo chambers is allowing users to unfollow or unfriend whoever they like, which makes it easy to cut yourself off from people you disagree with. They also found that when social networks recommend that users follow friends of friends 鈥� people who are likely to share similar views 鈥� echo chambers formed more than twice as quickly.
Sasahara and his colleagues suggest that social networks could try to prevent echo chambers by making it harder for people to severe connections with people with different opinions. Users should be encouraged to 鈥渟nooze鈥� an account rather than stop following it entirely 鈥� something 鈥� or alerted when they are about to unfollow their last source of a particular type of information.
Similarly, social networks should avoid recommending new connections that are likely to reinforce existing opinions too strongly. However, any tweaks should still allow people to cut themselves off from abusive accounts. 鈥淭he most important thing here is that it鈥檚 users who finally decide whether or not to follow or unfollow someone, not algorithms,鈥� says Sasahara.
Reference:听arXiv,
