The Influence of Stubborn Agents in a Multi-Agent Network for Inter-Team Cooperation/Negotiation

When teams interact for cooperation or negotiation, there are unique dynamics that occur depending on the conditions. In this paper, a multi-agent system is used under the restrain of a network structure to model two teams of agents interacting for a common consensus, however with the presence of stubborn agents. The networks used were a minimum dumbbell network and two scale-free networks joined together. The network topology, which is a global characteristic, along with the presence of conflicting stubborn agents, can cause various conditions that affect teamwork in cooperation or negotiation. Notable characteristics revealed are boundary role persons (BRPs), lack of unity, need for a third party moderator, coalition formation, and loyalty of the BRP dependent on the distance from the core ideology of the team. Both local and global characteristics of network structures contribute to such phenomenon. The modeling method and corresponding simulation results provide valuable insight for predicting possible social dynamics and outcome when planning cooperation/negotiation tactics. Keywords—Multi-agent system; consensus problem; stubborn agents; complex network; dumbbell network; Laplacian matrix; boundary role person; coalition formation