Faker Island. Profiling Misinformation Sharers on the French Twittersphere Using a Matching Procedure
What distinguishes users who share misinformation on social media from others? While experimental research suggests they exhibit lower analytical thinking, observational studies indicate a conservative bias and politically motivated behavior. However, both approaches present limitations: on the one hand, experimental research lacks ecological validity, whereas on the other hand, observational studies cannot infer causality. Furthermore, most research focuses on the U.S., a country characterized by a bi-partisan political system and a highly polarized media landscape, leaving open the question of who shares misinformation in other socio-political contexts. To address this gap, we analyze the French Twittersphere using ideological inference methods and a matching procedure. Our study is based on a dataset comprising four million tweets and compares 1,907 misinformation sharers with 958 users of similar political orientation who have not shared misinformation.
Our findings show that misinformation sharers are significantly more likely to (1) publicly display their political affiliation in their bios, (2) use pseudonyms, (3) share media content, (4) engage in hyperactive (re)tweeting, and (5) express negative emotions. However, their language does not indicate lower analytical thinking. These results challenge the idea that misinformation sharing stems from cognitive deficits and instead highlight the role of anger and political factors. By shifting the focus from misinformation susceptibility to the socio-political identities of sharers, this study contributes to ongoing theoretical debates and calls for further research into their activist practices and party affiliations.
Formal Explanations for Collective Decisions
Election results or the outcomes of participatory budgeting campaigns are typically presented to voters as a ranked list of alternatives based on scores. However, from a user or voter perspective, we believe this method fails to adequately explain why a particular candidate is elected or a project is approved. In this presentation, I will discuss our ongoing work on applying techniques used to explain black-box machine learning algorithms to transparent voting rules. Our explanations identify the smallest subsets of collected preference data that either support the winning candidate or, if altered, could change the outcome—these are known as counterfactual explanations. I will introduce algorithms for computing these formal explanations, along with refinements and bounds on their size, and share experimental results from real-world preference datasets. Our goal is to contribute to the establishment of solid foundations for explaining voting outcomes and to develop tools that provide voters with clear insights and engage them more in the collective decision-making process.
Building a Digital Commons for Democracy
Initially developed as a local digital platform, Decidim has grown into a global digital commons, used by governments, institutions, and communities to facilitate citizen engagement. We will explore the origins and history of the project, the principles that shape its governance and development, and how it has become a reference for open and democratic digital infrastructures. As an experiment in digital sovereignty, it continues to redefine the relationship between technology and democracy, offering an alternative to proprietary software in the realm of participatory processes. We will also delve into the current challenges related to sustainability, and the collective efforts that ensure Decidim remains a public and community-driven project.
Digital Democracy and the Bleak Economics of AI Slop
In this talk, I will discuss how the challenges and opportunities of digital democracy are shaped by the economic mechanisms and technological features characterising information goods, such as their tendency to produce concentrated markets, which in turn lead to inequalities in power and voice. We will focus on Artificial Intelligence (AI). As with past innovations that lowered entry and experimentation costs, AI now lowers the cost of information generation. Actors taking up the incentive to exploit that (commercially and/or politically) can use AI to flood the public domain with cheap and ubiquitous 'slop' that degrades and dilutes information quality and shifts the cost of search for relevant information on citizens and communities. As a result, AI acts as an accelerant of deeper societal dynamics, and can have a long-lasting impact on digital democracy (and democracy in general).
