Misinformation, Persuasion, and News Media on Social Networks

ABSTRACT

Social media platforms have become a popular source of news and information for a large segment of the society across the political spectrum: Users receive information, share digital contents, or attend to some online publishers for latest news. However, the recent proliferating and fast-spreading misinformation and false news has affected people’s perception about the veracity of online information and in turn their social behavior. In such an environment of real and false information, this dissertation studies two aspects of user behavior through the lens of persuasion motives: (1) decision making of sharing online news, (2) choice on which news media for information.

The first module focuses on online news dissemination on a Twitter-like social network. Agents have heterogeneous subjective priors on an unobservable real-valued state of the world. There is a noisy observation (news) of the state and any agent receiving the news decides whether to share the news with her followers so as to assimilate their opinions to her own. I propose two frameworks: one considers computational agents and the other one is under game-theoretic setting.

In the first model, agents share the news based on whether it can move their followers’ beliefs closer to their own in aggregate and the current size of news spread without projecting news spreading in the future. I describe the dynamics of the news spread arising from the individual decision rule and uncover the mechanisms that lead to a sharing cascade. I elucidate an association between the news precision levels that maximize the probability of a cascade and the prior wisdom of the crowd.

The second model concerns a binary vote and rational agents who share news so as to make their followers cast the same vote as they do while strategically speculating on others’ sharing decisions and news spread in the future. I characterize the underlying news spread as an endogenous Susceptible-Infected (SI) epidemic process and derive agents’ sharing decisions and the size of the sharing cascade at the equilibrium of the game. I show that lower credibility news can result in a larger cascade than fully credible news provided that the network connectivity surpasses a connectivity limit. I further delineate the relationship between cascade size, network connectivity, and news credibility in terms of polarization and diversity in prior beliefs.

The second component of this dissertation investigates the choices of subscribers with diverse prior beliefs between two ideologically opposing news media (intermediaries) who are motivated to influence the public opinion, through their roles of news verification and selective disclosure. The news media may access some news about the state of the world; the news may or may not be informative and they can choose to verify it. The news media then decide whether to disclose the news, aiming to persuade their subscribers to take the optimal action about the state based on their own beliefs. I show that subscribers in the center of the ideological spectrum choose to subscribe to the intermediary with the opposing view, thereby exhibiting anti-homophily. In contrast, the extremists exhibit homophily and prefer the intermediary with aligned ideology.

COMMITTEE

Ali Jadbabaie (chair, supervisor), Muhamet Yildiz, Dirk Bergemann (Yale), Amir Ajorlou

EVENT INFORMATION

Hybrid event. To attend virtually, please contact the IDSS Academic Office (idss_academic_office@mit.edu) for connection information.