The diffusion of true and false information on the internet
http://science.sciencemag.org/content/359/6380/1146
This article concerns the findings of three researchers : Vosoughi, Roy, and Aral. Utilizing a “data set of rumor cascades”, these researchers were able to come to the conclusion that false news actually diffuses through networks to greater depth, and faster than, true news. “The top 1 percent of false news cascades diffused to between 1000 and 100,000 people, whereas the truth rarely diffused to more than 1000 people. In terms of specifics of the study, they classified news and being true and false via “six independent fact-checking organizations”. Internet bots spread both news at the same rate, so the effects of this study were primarily due to news spreading faster through human networks, as opposed to automated ones.
Thinking about the results of this study in terms of the network diffusion material of this class, there could be several reasons why fake news is reaching more people in these networks than true news. One possible reason involves the relative payoffs of spreading true news vs fake news. Firstly, lets assume that there are two possible, binary behaviors : A – spreading fake news, B- posting your original content. Most people start off with posting original content, and some few individuals first spreading fake news, maybe due to personal interest or gullibility. As we covered in class, one’s desire to switch to behavior A depends on whether at least a b(a+b) fraction of your friends are doing it(where b and a are the payoffs for behavior B and A respectively). What might be taking place is that behavior A actually has a very high payoff, maybe in the form of social clout, or just intrinsically feeling more rewarded because you’re posting something relatively outlandish. Thus, because behavior A is very attractive, the proportion of friends needed for someone to adopt A might be pretty low. Thus, people are very easily convinced to spread fake news,and there are relatively few clusters that won’t adopt this behavior. Hence, fake news could reach over a 100,000 people. In the exact same scenario, but now labeling behavior A as spreading true news, the payoff for spreading true news might be a lot smaller than it was for fake news. Thus, the required adoption threshold is a lot higher, resulting in the spread of true news not being able to permeate through some clusters.