Skip to content

Computational modeling

Computational modeling and simulation are essential tools for understanding misinformation dynamics, recommender system behavior, and information spread on social platforms. These approaches range from agent-based models simulating individual user behavior to network models capturing information flow to mathematical frameworks analyzing equilibria and outcomes.

Agent-based models

Agent-based models (ABMs) simulate heterogeneous users with individual preferences, behaviors, and interactions. They are particularly valuable for understanding emergent phenomena—outcomes that arise from local interactions and are not easily predicted from high-level rules. In the misinformation context, ABMs help explain platform dynamics, recommendation effects, and radicalization pathways.

Key papers

  • The Amplification Paradox in Recommender Systems — Agent-based model with realistic user preferences explaining the "amplification paradox": why algorithmic audits find recommendation amplification of extreme content yet real user logs show recommendations don't drive extreme consumption. Demonstrates how collaborative filtering plus content nicheness alone explain contradictory empirical findings.