| Popis: |
Social capital has been defined as an attribute of individuals that facilitates cooperation to achieve mutual benefit, and enhances a group's capability to solve collective action problems. In this paper, we formalise a new computational framework for optimising self-organised collective action using electronic social capital. This framework comprises event handlers which update multivariate forms of social capital (trustworthiness, social network, and institutions), and a set of metrics over these forms that provide inputs to social decision-making processes. We implement an experimental multi-agent test bed where a number of agents iteratively play simultaneous n-player games, and use the social capital framework for their action-selection. Our results show that social capital optimises outcomes (in terms of long-term satisfaction and utility), reduces the complexity of decision making, and allows the system to scale to support self-organising collective action in 'large' groups. |
