Collective design and innovation processes have become a necessity for the development of successful solutions for most real-world problems. Such large-scale design processes typically involve individuals with diverse knowledge, expertise and behaviors, and the organizational structures under which the collective design takes place are often complex and dynamic. Here we investigate, using a computational agent-based model, how the diversity of knowledge, expertise, and behaviors of individual members and the organizational network structure will affect the effectiveness of design and innovation processes at collective levels. Our agent-based model consists of (1) multi-dimensional (= multi-domain) problem space with nonlinear utility landscape, (2) individual agents with different domains of expertise/confidence and utility perceptions, and (3) organizational network structure in which the agents collaborate with each other. The agents use evolutionary operators as the means of exploration, exploitation, and information sharing to develop and refine their respective idea pools and to collaboratively produce a high-utility solution. Two experimental parameters were varied: organizational network structure and diversity of agents’ domains of expertise. Simulation results showed that the increase of diversity of expertise had a positive impact on the quality of final solutions particularly on small-world networks, and that small-world networks produced solutions much more efficiently than other networks. It was also found that the increase of network density did not contribute to the quality of final solutions, but instead it reduced the volume of the explored problem space. We plan to test these theoretical findings empirically through online human-subject experiments later this year.
