The effects of kinship bias on multi-agent environments: Studies on theoretical models and the Internet access problems

Multi-agent systems have been an important focus of research in many academic disciplines. Researchers have been studying general macro behaviors of rational agents that try to maximize their own utilities. When a substantial number of agents interacts locally without global control, there still are certain behaviors that can be observed on a macro level. This research focuses on the emergence of cooperation among rational agents.; In the first part of the dissertation, we develop several theoretical models to study the effects of similarity-based cooperation, also known as kinship-biased cooperation. In these models, in which agents with similar characteristics are encouraged to interact frequently, persistent global cooperation emerges.; In the second part, we focus on a practical problem known as the Multiple Access Problem (MAP), which considers the Internet information search strategies that effectively retrieve required information without overloading information servers.; Studying the MAP requires a theoretical model that captures the essence of such a problem. We prove mathematically that the MAP is the n-Person Prisoner's Dilemma Game. With the n-IPD formalism, we find the optimal interval---the lower bound and upper bound---for the number of information sites for a given number of information-seeking agents. When the optimal interval is satisfied, cooperation among agents is effective, meaning that resources---i.e., information sites are optimally shared. When the optimal interval is not satisfied, cooperation is not effective since there is no dilemma; that is, either the resource is not enough for the given number of agents no matter what the agents do or cooperation is not necessary since the resource is abundant.; The dissertation then presents empirical studies on the MAP. In the empirical studies, we observe the effects of several mechanisms that improve the global performance of agents in retrieving information being sought. We also discover that cooperation in the MAP is very difficult to promote. The dissertation presents inherent difficulties involving the MAP as well as possible solutions.