Diffusion of competing innovations in influence networks

Although an influence network is a crucial factor affecting diffusion processes, it is generally fixed or assumed to follow some typical topologies in traditional diffusion research. The purpose of this study is to examine how innovation diffusion is changed by different influence relationship structures existing among individuals. We introduce an extended version of the regular digraph that can represent various influence relationships among influentials and followers. We focus on three key features of influence relationships (i.e. monopolization, localization, and diversification of opinions) and examine how they are associated with certain macroscopic behavioral regularities by employing agent-based simulation. The simulation results show that market becomes “locked-in” to a single product when influences are monopolized by few influentials. We also find that when influence relationship becomes complex as influentials and followers increase, market cannot be categorized by a single typology, but becomes random and unpredictable. Our model demonstrates successfully an underlying principle of collective behavior that uniformity of behavior is promoted under monopolization of opinions and random, unpredictable behavioral patterns emerge from diversification of opinions.