Knowledge diffusion in social networks under targeted attack and random failure: the resilience of communities

Knowledge diffusion is a complex and demanding process that requires coordination and collaboration between agents with different levels of knowledge, to establish fruitful learning interactions. In this paper, we develop an agent-based model to investigate how different behavioral/sociological rules can alter, strengthen, or weaken this process. We observe that, during normal times, different aggregation strategies are apparently irrelevant for determining differences in learning opportunities. However, under crisis, there is an observable outperformance of social structures with established communities, characterized by both strong ties (i.e., intense contacts within communities) and weak ties (i.e., knowledge spillover across communities). We further test system resilience, considering interruptions to the knowledge diffusion of expert agents and the random temporary removal of agents (simulating a viral outbreak). We discuss how these scenarios may explain economic phenomena and explore the implications for policies aimed at mitigating knowledge and economic inequalities.