Unmasking the Orchestra: Quantifying the Impact of Coordinated Accounts in Online Debates

As the threat from coordinated accounts that seek to manipulate online discussions continues to grow, the availability of publicly accessible social media data offers an invaluable resource for analyzing and understanding the impact of such harmful activities. This data provides a detailed view of how coordinated efforts influence public discourse, allowing us to uncover patterns, detect malicious behavior, and assess the broader implications of these tactics. In this thesis, we investigate the impact of coordinated accounts on online debates by analyzing Twitter discussions surrounding the 2015 COP21 and 2021 COP26 climate change conferences. Current approaches for detecting coordinated accounts often lack fine-grained metrics to quantify their impact on online discourse over time. To address this, we propose an integrated approach that evaluates a range of measurements, including user position in information cascades, action delay, descendant counts, hashtag co-occurrences, and user interaction metrics. By combining these measurements, we develop a comprehensive set of methods to identify and assess the evolving impact of coordinated accounts in online debates. This approach improves our understanding of how these accounts influence public debates and contribute to polarization, informing the design of systems, algorithms, and policies to mitigate these threats. Through a series of temporal analyses, we measure changes and trends in coordination rate, the relationship between coordination and polarization, and the evolution of debate diversity and toxicity. Additionally, we compare these dynamics between COP21 and COP26, and explore shifts in skepticism among the groups.