Trust domains in system models: algebra, logic, utility, and combinators

Abstract

Understanding the boundaries of trust is a key aspect of accurately modelling the structure and behaviour of multi-agent systems with heterogeneous motivating factors. Reasoning about these boundaries in highly interconnected, information-rich ecosystems is complex, and dependent upon modelling at the correct level of abstraction. Building on an established mathematical systems modelling framework that captures the classical view of distributed systems, we develop a modelling framework that incorporates both logical and cost-based descriptions of systems, which allows us to establish a definition of an agent's trust domain based on the satisfaction of logical properties at acceptable utility (handled here simply as cost) to the agent, of verification. In addition to the technical properties of the modelling framework itself, we establish a theory of logical combinators, including substitution, for composing trust domains to form relatively complex models of trust. We illustrate the ideas with examples throughout.