Researchers model misinformation as a shared trust commons problem

A new paper by researcher Vrinda Malhotra, accepted at the Annual Modeling and Simulation Conference 2026, models misinformation as a commons problem—where trust is a shared collective resource that degrades when too much attention goes to low-credibility content. The agent-based simulation treats attention as a scarce private budget each agent allocates across sources. When aggregate attention shifts toward unreliable content, the trust environment deteriorates, making it harder for users to process and correct credible information. Across experiments, the system settles into four distinct modes: credible stability (high trust, reliable information dominates), misinformation dominance (trust collapses, false claims take over), polarization (two opposing trust clusters emerge), and a mixed baseline (fluctuating but no clear winner).

The paper separates two critical control problems for policy simulation. First, the balance between trust repair and harm largely determines whether the system recovers from misinformation or collapses into a low-trust state. Second, homophily (the tendency to connect with similar others) and rewiring (dynamic network changes) dictate whether disagreement remains integrated across groups or solidifies into persistent, isolated clusters. This framework provides a transparent, reproducible testbed for comparative experiments on interventions—such as fact-checking, credibility nudges, or cross-cutting exposure—that must address both trust restoration and the structural conditions that sustain polarization. The work bridges computer science and social science, offering concrete levers for policymakers to combat the systemic erosion of shared truth.