Disciplined Exploration of Emergence Using Multi-Agent Simulation Framework

Abstract

In recent years the concept of emergence has gained much attention as ICT systems have started exhibiting properties usually associated with complex systems. Although emergence creates many problems for engineering complex ICT systems by introducing undesired behaviour, it also offers many possibilities for advance in the area of adaptive self-organizing systems. However, at the moment the inability to predict and control emergent phenomena prevents us from exploring its full potential or avoiding problems in existing complex systems. Towards this end, this paper proposes a framework for empirical study of complex systems exhibiting emergence. The framework relies on agent-oriented modelling and simulation as a tool for examination of specific manifestations of emergence. The main idea is to use an iterative simulation process in order to build a coarse taxonomy of causal relationships between the micro- and macro layers. In addition to the detailed description of the framework, the paper also discusses the corresponding verification and validation processes as important factor for the success of such a study.