Work in Progress
`The Cursed Equilibria of Algorithmic Traumatization'
Abstract - This paper explores the phenomenon of algorithmic traumatization within the context of modern digital platforms, where the mechanisms of content recommendation algorithms pose unique challenges to individual mental health and well-being. Algorithmic traumatization materializes in the psychological harm suffered by users as a result of interacting with algorithmically curated and targeted content, especially when such content results from users’ profiling, reinforcing harmful obsessions or exposing to triggering material. Using an economic framework, the paper proposes a theory of harm based on (1) the higher probability that vulnerable users (i.e. users with obsessive harmful thoughts) will engage with certain content and (2) the incentive and ability of platforms to provide harmful personalization. This theory will argue that, ignoring long-term consequences of potentially harmful recommendations, the interaction of users and platform can result in ‘cursed’ equilibria, namely stable states in which both are engaged in a suboptimal cycle of mutually detrimental outcomes. In particular, the model will highlight that, for vulnerable users (1) individually harmless content can cumulatively harm mental well-being when shown in excessive quantities and (2) a reduction in the variety of recommended content amplifies long-term risk of harm. Therefore, the paper will recommend platforms and regulators to ensure users are offered and/or recommended an optimal level of content variety, regardless of the collected information about them. Finally, the paper will examine strengths and weaknesses of current regulatory frameworks that can shape platforms’ conducts and prevent algorithmic traumatization from happening, providing some recommendations based on this algorithmic theory of harm.
`The Two Faces of Personalization: Online Consumer Search and the Role of Advertising'
Abstract - This paper explores the impact of personalization, through targeted and contextual advertising, on online consumer search behaviors and purchase decisions. Personalization and targeted advertising have significantly influenced digital platforms, enhancing sales and user engagement metrics across major online retailers and content providers. However, concerns regarding their potential negative consequences on consumers have prompted regulatory scrutiny globally. Leveraging data from a large field experiment involving 1,200 participants over three months, the research intends to compare online search behaviors, product prices, and characteristics under different ad exposure conditions. The study challenges traditional consumer search models by considering the minimal costs and unique challenges of online environments, such as choice overload and the psychological impacts of targeted advertising. The work proposes a comprehensive methodology to explore search costs, the salience of product attributes due to advertising cues, and the nature of consideration sets in shaping consumer behavior online. This approach aims to provide new insights into the influence of digital platforms and advertising on consumer decisions, potentially informing platform conduct discussions and supporting designs that enhance consumer welfare. The broader implications of this research could contribute to updating industrial organization models, informing debates on large platform conducts, and shaping data protection regulations by understanding the net benefits of personalization for users.
`Should I Post it? Privacy and Motivations Behind Online Sharing'
Abstract - The decision of sharing content online is complex and it is driven by the interplay of many factors, which motivate the existence of a discrepancy between stated and revealed preferences of sharing personal information online. A quasi-experimental survey has been administered to 168 Cornell University students who own an Instagram account, eliciting preferences for online sharing, social norms, quality of former experiences in social networks, and actual posting behavior. This study suggests a content-type level approach that evaluates the interplay of a number of factors, and tests the effect of contextual cues for sharing private information. Cues enhance the influence of the willingness-to-share (WTS) on social norms, but the impact of the fear of being misjudged and the lack of confidence in physical appearance have a notable effect for sharing self-depicting pictures. While most of the content is considered appropriate, many users feel uncomfortable sharing it and they eventually choose not to share it. Former privacy violations and feeling prevented from sharing online are more frequent among users with high WTS and higher appropriateness rates.
Published Papers
2022 - Applying Behavioural Economics to Firms
with Tim Hogg and Leon Fields | The Behavioral Economics Guide (2022) & Concurrences Writing Awards (2022)