
- Director: Dr Uri Simonsohn, Esade - Universitat Ramon Llull
- Date of defense: July 15th, 2022
Abstract
Lies, manipulation, or false information are nothing new. All have been around for centuries since the flow and content of information has been a source of power. However, in the age of the social platforms, the mechanism has changed. While broadcasters still exist, they compete with other information sources that rely on virality through the network for information diffusion. In this competition, traditional and institutionally created content suffer a disadvantage: they do not play with the same rules. Content that deviates from the deontological and stylistic reporting of reliable information providers leverages and hijacks psychological heuristics to become more viral. For example, they rely on emotional evocation (sentimental language and appeal to moral values) and rely on styles that require less cognitive effort to be processed (in terms of grammatical and lexical complexity). In this PhD thesis, I explore how misinformation sources take advantage of psychological heuristics to define their strategic choices. More specifically, the thesis is comprised of three studies: (1) Using Natural Language Processing, a quantitative exploration about how different misinformation categories (clickbait, conspiracy theories, fake news, hate speech, junk science, and rumors) differ in terms of sentiment, appeal to morality, readability, and lexical diversity; (2) By means of network analysis and Natural Language Processing, a demonstration of isomorphism strategies from misinformation sources towards reliable sources with the aim of mimicking high-status actors in the information ecosystem; and (3) a dynamic analysis of fake news vs reliable sources reporting about the Covid pandemic during 2020 and 2021. In general, my results indicate that there are significant structural differences between distinct misinformation categories and between misinformation sources and reliable sources (for example, fake news are, on average, 18 times more negative, 12 times less lexically diverse, 50% more appealing to morality and 13% easier to process than factual news. Also, I present evidence about how fake news, clickbait, and rumors employ a mimicking strategy towards reliable sources with the aim to be perceived as legitimate and high-status information providers. Moreover, I show how reliable and fake news sources behaved during the Covid pandemic, revealing a convoluted initial reaction followed by a misinformation strategy to create chaos and ambiguity by constantly shifting between topics. Finally, I discuss the implications of the thesis results for social media platforms, the media ecosystem, political elites, and Internet users.