Complex systems and Agent-based modelling
Complex Adaptive Systems
Complex adaptive systems theory challenges traditional notions of scientific research, social behaviour and policy making. Once social systems, such as a market, are conceptualized as complex systems, the challenge is not about management and control of such systems but how to participate and manage in them. No one is in overall control, though some may have far more influence than others. Also system outcomes are not a simple linear function of the behaviour of the parts. Driving forces in the system are the adaptive behaviour of individuals and feedback effects between the system and the individual level. Actors respond the actions of others who are responding to them ad infinitum.
One way of advancing research into complex systems is through the use of computer simulations, including agent-based methodologies (ABM), which are becoming ever more widely appreciated and used in the natural and social sciences. ABM represents a revolution in the way scientists can build models to replicate, analyse, test and predict the behaviour of complex systems, as it does not smooth over nonlinearities and can embrace major transitions and catastrophes. ABM simulates complex systems from the bottom-up, representing the actions and interactions of numerous individual agents in an artificial society.
Agent-based models are models formalized in computer code, representing a collection of recursive mathematical rules, applied to a clearly defined set of inputs. Such models are not restricted to general statistical models of behaviour, central driving equations or representative agents, but can represent explicitly the micro interactions taking place and how they produce macro structures and patterns of behaviour. Statistical and mathematical analysis and variable-based accounts of system behaviour are still relevant: they play an essential role in developing and testing ABM and in summarising their output, as they do for real-world social and economic systems.