Did filter bubbles allow Trump to win the presidency?
I want to talk about the election of Donald Trump last week. Everyone was shocked that he won – Trump was leading in 1 out of 10 polls the day before the election, giving him a slim margin to win. Even the renowned Nate Silver at FiveThirtyEight only gave him a 29% chance of winning (source), and under a popular vote system he wouldn’t have come close to winning, with a several hundred thousand vote deficit relative to Hillary Clinton. So then, how in the world did Trump manage to overcome such odds?
Over the past week, I have seen and heard many potential reasons. “Rural Americans were underestimated in the polls and showed up in higher numbers than we expected.” This isn’t compelling because rural Americans tend to vote conservative – and Trump got nearly the same number of votes Mitt Romney did back in 2012. “No one wanted to vote between the lesser of two evils.” A more compelling reason personally, but it doesn’t explain the turnout rate. “Third party candidates drew votes away from the major party candidates.” A potential explanation, but it only works if one party lost more candidates than the other, which there is little evidence for.
I would like to examine the effects of networks in the presidential election, specifically in relation to filter bubbles, which arise when a subnetwork of people share articles that fail to challenge their beliefs. People stuck within a filter bubble are more likely to see exclusively articles that support their own worldview – which many argue merely strengthens that worldview. This can have many effects, predominantly dealing with political polarization. Not only that, the polarization can be aggressive and demeaning towards political opponents – oftentimes offensive or unjustified, further dividing the network into bubbles. Unfortunately, Facebook has outright denied anything of the sort – see here.
In terms of networks, we can think about this as a (social) network becoming extremely disconnected. Information will only flow internally to bubbles, but rarely outside of them. Liberal networks will only share information with themselves and conservative networks will only share information with themselves. This can lead to a vast array of misinformation, and consequently bad information cascades. In this example, a liberal filter bubble believed that Hillary Clinton was certainly going to win, and therefore did not feel obliged to vote. The few swing voters who were essential to a Clinton victory did not see the point of voting because they heavily believed Clinton was going to win, but they personally could not justify voting for her entirely. In such a scenario, voting would be a waste of time, regardless of what you think of the other candidates. This would have been the main effect of a filter bubble in the election.
Although speculative, I believe it is also possible that filter bubbles helped alienate a substantial portion of the Democratic voter base. Personally, I have seen some aggressive articles on my own Facebook wall that would lead me to believe that the Democratic party no longer represents the working class American it has long pledged to represent – as evidenced by derogatory tones and simple neglect. As a protest against the “establishment”, the same voters who elected Barack Obama in 2008 and 2012 may have finally cracked and voted this time against Clinton. Of course, this is speculative, but could be a potential result of liberal filter bubbles encouraging alienation of the voter base.