Few thoughts on an important paper that just appeared in Science, The spread of true and false news online. The paper received (and will receive) justified attention: it is massive (“~126,000 rumor cascades spread by ~3 million people more than 4.5 million times” in a long temporal window – from 2006 to 2017), it includes several detailed analyses (the authors did not only check basic metrics such as speed and size of diffusion, but they measured things like structural virality; the proportion of political versus non-political news; the role of bots; they run a sentiment analysis of the tweets, etc.), and it has a straightforward (and I guess welcome to many) take-home message: “fake” news are more successfull than “true” news in social media, at least in Twitter (*).
I am curious to see how this will be commented. To me, in fact, this study shows that false news does not spread because of social media, but because their content is, on average, more attractive than the content of true news. One of the main results of the study is indeed that what differentiates false from true news are certain emotional features, such as provoking more surprise or disgust. This should be familiar to psychologists, anthropologists, or researchers in cultural evolution. The same emotions have been found to favour the transmission of stories in experiments in which people talk to each other in “transmission chains”, the experimental equivalent of the Chinese whispers game (here a recent review.)
From an evolutionary point of view, it makes sense that individuals are not sensitive to an abstract notion of “truth”, but to various cues that point to the importance of the content (**), and that may be associated, only on average, to truthfulness. Thinking in this way, “false” news can exploit these features in an almost unconstrained way, while “true” news can not, exactly because they need to correspond to reality. In other words, fake news can be designed to spread, and the fact that they do it better than true news is not that surprising.
Said so, it is still an open question what the effect of the spreading of fake news is. One thing is if many people enjoy sharing silly stories, another if fake news have an important role in political and cultural dynamics. I do not think we have an answer but, against the doomsayers, there are a few papers that studied specifically the impact of fake news in political episodes such as the election of Donald Trump, and they found it generally limited (see here and here.)
I found revealing a buzzfeed.com article of few months ago, that reported on the 50 biggest fake news on Facebook in 2017. Below, from the article, the first ten:
When I linked to this article on my Twitter I commented “there is nothing new under the sun”, and I continue to have the same opinion (4 on 10 have the word “penis” or “vagina” in the title!)
My (clearly biased) take-home message from the Science article is that studies of spreading of online misinformation need to take advantage of researchers in cognitive sciences and cultural evolution. If we want to know why fake news spread on social media, perhaps we need to know why many of us enjoy them. Maybe the fault is not the internet, and Facebook and Twitter are just a magnifying mirror of what we are.
(**) While I do myself quantitative research on the cultural dynamics of emotions, I think that emotions are pointers to more specific features of content, as for example claimed here for threat-related material, as one of the specific features of a more general negative-emotion bias.