Multi-Agent-Based Simulation

TL;DR: The issue addressed in the MABS-2000 workshop and in this volume is the development of a methodology and technology to identify which techniques hold promise and which cannot possibly lead to useful applications for messy software or social systems.

Abstract: Messy systems have no clear boundaries; they are composed of so many natural and/or artificial entities with patterns of interaction so dense that they cannot be understood by inspection and system behaviour cannot be predicted by statistical or qualitative analysis. Obvious examples are real social systems and the Internet. Analysing and designing software to exploit such systems requires a different approach to software engineering and mechanism design. The issue addressed in the MABS-2000 workshop and in this volume is the development of a methodology and technology to identify which techniques hold promise and which cannot possibly lead to useful applications for messy software or social systems. 1 Messy Systems and Tidy Systems Much of AI and the great preponderance of papers at conferences such as ICMAS relate to software systems with clear boundaries and well understood relationships among known entities. The perceived development path is by means of step-by-step development of well understood formalisms and algorithms. These are tidy techniques for tidy systems. Real systems are frequently messy. The boundaries of action and opportunity are unclear and probably unknowable. Relationships among entities are either too complex to be understood or shifting faster than we can identify them. In short, we frequently do not know enough about the systems to make them amenable to tidy techniques for toy systems. The differences between messy and tidy systems have been recognised for some considerable time by the agents research community. Wooldridge and Jennings [20] argued that “[i]f a system contains many agents..., then the dynamics can become too complex to manage effectively. There are several techniques that one can use to try to manage a system in which there are many agents. First, one can place it under central control... Another way... is to severely restrict the way in which agents can interact... one can ensure that there are few channels of communication... [or] by restricting the way in which agents interact. Thus very