The Chilling: Identifying Strategic Antisocial Behavior Online and Examining the Impact on Journalists

We introduce ConvTreeTrans, a new model to analyze online conversation trees. We find that attackers use coordinated strategies while supporters do not, creating a 'chilling effect' that causes journalists to delay their posting.

Identifying Preferences Is Harder Than You Think

Discover how a simple "organize-then-vote" step transforms choice overload into confident, authentic decisions.

Adversarial Machine UNlearning (AMUN)

We present AMUN, a new method for machine unlearning that leverages adversarial examples to remove specific training samples from a model while preserving accuracy. Our approach outperforms prior state-of-the-art unlearning methods, especially when access to remaining training data is limited.

Algorithmic Collective Action With Two Collectives

Many socio-technical systems rely on user interactions and information to recommend content to other users. Here we study how collectives of users may act together to influence these systems.