Representing personal consumption is important to a range of expert decision makers in academic, government and corporate settings. We cite the possible contribution of studying a system model of the use of information in personal consumption when agents are members of a small world network. Information has properties in non-rival borrowing and sensitivity to network memberships that individual agents do not typically include in heuristics they use in budget allocation. We use computational methods to show how an expert system in which a super agent using a model that includes properties of information and network parameters can define welfare maximizing allocation and policy to bring the allocation of individual agents closer to a welfare-maximizing allocation. Implications of systems with the dynamics we propose for representing demand-side management in expert systems are noted.