Exploring agent-based simulation of causal maps: Toward a strategic decision support tool

This dissertation builds and tests a prototype system for strategic decision support. The key innovation in the research is the incorporation of multi-agent systems, a distributed artificial intelligence technique, into the innovative yet integrative causal mapping and dialectical system for scenario planning developed by Acar (1983). Strategic decision-making in firms remains more art than science, leading many to propose that managers faced with complex dynamic electronic market environments increasingly depend on intuition and "strategic thinking" to cope with rapid change. For firms heavily invested in information technology, every aspect of the value chain, both internal and external, is becoming integrated and dependent on information technology save one: support of strategic decision making. While a decision maker's intuitive thinking is a necessary part of strategy formulation, intuition alone is not fully reliable. Situations require strategic thinking take place in real time.; The research problem addressed by this dissertation is the exploratory development and testing of a prototyped tool for strategic decision support that can integrate with today's dynamic systems. The prototype incorporates an understanding of strategy as contained in the literature, integrating an approach to problem formulation and strategy development proposed by Acar (1983) with the latest distributed artificial intelligence technology. The dissertation discusses the underlying strategic theory and the requirements of a knowledge-based strategic decision support system, and derives system requirements. Simulation of strategic environments using complex causal maps containing multiple feedback loops presents an analytically intractable problem that can be solved using agent-based modeling and simulation. For so doing, an agent-oriented approach to causal mapping is developed. Usability studies test the user-friendliness of the interface and provide directions for further research and development of the system as a viable tool for strategic decision support.