Fake News on Facebook
http://www.vox.com/new-money/2016/11/16/13637310/facebook-fake-news-explained
Recently, especially in light of the presidential election, the issue of fake news on Facebook has gained a lot of attention and outcry. Many people are voicing their opinions about Facebook’s responsibility to bar fake news from spreading on its platform, and many people are also questioning whether the spreading of fake news in the time period leading up to the Presidential election could have affected the final results by a significant amount. Such fake news includes the fact that the Pope endorsed Trump, which is not true. Yet posts like these can gain a lot of attention on Facebook, even if some users are able to tell that they are false. Due to Facebook’s algorithm that prioritizes engagement when displaying posts to users, this allows these high-attention pieces to gain even more attention.
Although this is not a traditional disease, we can look at this situation through the lens of what we have been discussing in class about epidemics and their spread. To make the direct analogy, the “disease” here is some individual believing in a piece of fake news. As fake news spreads, users who share fake news and users who engage with those posts (thereby increasing its priority in newsfeeds) are the ways in which “infected” individuals come into contact with other individuals. The rate of infection, then, is the fraction of users who come into contact with fake news and actually believe them.
If we try to model this with a branching process, we calculate our R_0 by multiplying the number of people an infected individual comes into contact with by the rate of infection. In other words, our R_0 here is higher for the more people who see these fake news posts and for the more people who actually believe the fake news.
Thinking about it this way may shed some light on why fake news is considered such a big problem by many recently: the reach of Facebook posts to individuals is extremely high. This means that even if a vast majority of people wouldn’t believe headlines like “Pope endorses Trump”, the sheer number of people reached could make our R_0 value much greater than 1 – or at least definitely not less than 1. We know that R_0s of less than 1 are guaranteed to die out in a finite number of waves, but we cannot say the same for R_0s greater than 1–especially not for those much greater, as it likely may be in this case.