Inadvertent leaks: exploration via agent-based dynamic network simulation

Abstract

A significant volume of information leaks in organizations are inadvertent, a form of information spillage. Because the leakage of information is driven by the complex interaction of technology, social, and behavioral factors, we use a hybrid agent-based and dynamic network model, Construct, to simulate the flow of sensitive information in knowledge-driven organizations. Because interaction patterns often change when an organization is under stress, we simulate stress to the organization with effect-based (reliability and integrity) crisis scenarios. Using a virtual experiment, we vary the crisis scenarios, organization’s structure, IT connections, and pressure to separate personnel based on security ratings. Our experiment suggests that the organization’s structure, IT connections, separation pressure, and typical performance all influence how much an organization suffers from inadvertent leakage. In evaluating how organizations respond to crisis, organizations with stove-piped IT tend to clamp down on leakage in response to the crisis, while organizations with Mesh IT tend to have more leakage. Integrity crises tend to decrease leakage; while reliability crises tend to increase leakage in organizations, especially those with Mesh-based IT.