Institute for Software Research
School of Computer Science, Carnegie Mellon University
Simulating Diffusion with Conflicting Knowledge
Kenneth Joseph, Kathleen M. Carley
Center for the Computational Analysis of Social and Organizational Systems
A simulation model is developed and used to better understand the diffusion of conflicting knowledge in a social network. Within the simulation, correct and incorrect knowledge diffuse simultaneously, however, simulation agents are not explicitly made aware that contradictory information exists. Instead, they may become aware of this by obtaining both correct and incorrect knowledge of the same piece of information. The agent must then determine what knowledge to accept. We implement various models of how people react to conflicting knowledge, which we refer to as intrapersonal conflict resolution strategies (ICRS). We analyze these ICRS with respect to the speed and breadth of correct and incorrect knowledge spread and how often agents find themselves holding conflicting knowledge. We also consider the effect of varying the odds of incorrect knowledge existing in the system, the network structure and the differentials in trust between agents. We find evidence that our model is consistent with many real-world notions of diffusion. Consequently, we make tentative conclusions about the effect of our various experimental conditions on the spread of conflicting knowledge in a social network. We note that conflicting knowledge spreads more slowly and persists longer than knowledge which cannot be conflicted.