Do you think it’s true? Diversity of expression regimes in the face of fake news and mechanisms of conversational self-regulation

This thesis aims to explain two paradoxes: (1) Why do most empirical studies show that fake news represents only a small proportion of the total information consulted and shared by social media users, even though they are neither subject to editorial control nor bound by journalistic ethics? (2) How can we understand the rise of political polarization, given that users do not seem to be particularly receptive to fake news? To address these questions, two studies were conducted on Twitter and Facebook. Each combines quantitative analyses of digital traces with online observations and interviews. This hybrid methodological approach made it possible not to reduce the users studied to their reaction to a single fake news item on a particular social network, but to examine the variety of their practices in different interaction situations (both online and offline), while identifying some of their socio- demographic characteristics. The first study identified all users who shared at least one piece of content classified as fake news by fact-checkers in the French Twittersphere. Drawing on a corpus of content flagged as fake news by Facebook users, the second study went beyond the issue of factuality to examine users' reactions to statements whose epistemic quality is uncertain. Three main findings emerge from the thesis. First, the sharing of fake news does not affect all social media users equally or indiscriminately; it is actually limited to a small group of internet users. These individuals are not less educated or less cognitively skilled than others, but they are more politicized and critical of institutions. Although they are in the minority, these users are likely to facilitate the agenda-setting of their political camp's opinions in the public debate due to their hyperactivity online and the vast amount of news they share. Second, social media users exposed to fake news can deploy forms of critical distance to varying degrees, depending on their position in the social space and the interaction norms of the situations they are in, either by exercising "prudence énonciative" (discursive caution) or by expressing "points d'arrêt”, i.e., intervening in the flow of a conversation to formulate disagreements or corrections. Third, these forms of critical distance rarely lead to genuine deliberative debates or the expression of agonistic pluralism but instead result in "dialogues of the deaf" among a minority of particularly active online users. These conclusions call for future academic studies, as well as the public debate, to shift their focus away from the sole issue of fake news, so as not to overlook other information and communication troubles, such as the manipulation of the political agenda or the brutalization of public debate by a minority of users and the spiral of silence mechanisms that follow.