Countering Fake News: A Comparison of Possible Solutions Regarding User Acceptance and Effectiveness

Since the emergence of so-called fake news on the internet and in social media, platforms such as Facebook have started to take countermeasures, and researchers have begun looking into this phenomenon from a variety of perspectives. A large number of scientific work has investigated ways to detect fake news automatically. Less attention has been paid to the subsequent step, i.e., what to do when you are aware of the inaccuracy of claims in social media. This work takes a user-centered approach on means to counter identified mis- and disinformation in social media. We conduct a three-step study design on how approaches in social media should be presented to respect the users' needs and experiences and how effective they are. As our first step, in an online survey representative for some factors to the German adult population, we enquire regarding their strategies on handling information in social media, and their opinion regarding possible solutions --- focusing on the approach of displaying a warning on inaccurate posts. In a second step, we present five potential approaches for countermeasures identified in related work to interviewees for qualitative input. We discuss (1) warning, (2) related articles, (3) reducing the size, (4) covering, and (5) requiring confirmation. Based on the interview feedback, as the third step of this study, we select, improve, and examine four promising approaches on how to counter misinformation. We conduct an online experiment to test their effectiveness on the perceived accuracy of false headlines and also ask for the users' preferences. In this study, we find that users welcome warning-based approaches to counter fake news and are somewhat critical with less transparent methods. Moreover, users want social media platforms to explain why a post was marked as disputed. The results regarding effectiveness are similar: Warning-based approaches are shown to be effective in reducing the perceived accuracy of false headlines. Moreover, adding an explanation to the warning leads to the most significant results. In contrast, we could not find a significant effect on one of Facebook's current approaches (reduced post size and fact-checks in related articles).