Simulation models bring up the significance of interactions in the outcomes. For example, in the simulation of cat pictures and cats (Figure 6.2), we observed a potential for self-enforcing loops. The more cats there were, the more cat pictures, which would lead to an increase number of cats. The loop is complete. Thus, by adding only one cat into the world, the outcome could be a rapid increase of cat ownership. Such phenomena are often difficult to grasp, as they include nonlinearity, feedback loops and emergence at a group level. These aspects are common for complex systems.
Many different kinds of phenomena can be seen as complex, or having characteristics such as heterogeneity of interacting parts, unexpected or unpredictable emergence or sensitivity to initial conditions (see Table 6.2) for details). San Miguel et al. (2012) suggests that cities, companies, the climate, political markets, traffic jams, dinner parties and housing markets - among many other things - can be considered complex systems. Complex systems provide a framework to approach various phenomena (see Table 6.2 for a summary of this framework). Similar to networks (see Section 4.1), complex systems approach serves as a paradigm to approach a problem. They bear another similarity with networks as well. They have a deeply rooted history in mathematics and many years of evolution supporting its purposes.
Within the complex systems field, there are many legitimate ways to use these ideas. Sometimes these theories are used to generate categories that describe qualitative data. For example, Keshavarz et al. (2010) argues that schools are complex systems by qualitatively analysing 26 staff interviews, 18 management plans and 18 annual reports. In their analysis, they use ideas from major theories of complexity, such as diversity of agents in the system, feedback loops, interaction between agents and unpredictability to highlight how a school has characteristics of complex systems. Others use the ideas from complex adaptive systems to structure thinking. For example, Schneider and Somers (2006) formulate leadership in organisations by acknowledging that organisations are complex systems. This helps them to pinpoint pains in the leadership studies. Some have even argued that complexity approaches are overly used extensively without proper testing, thus leading to a fad (Mckelvey, 2002; Maguire and McKelvey, 1999). Similar to simulation models, there are calls to develop a stronger test between a link of theory and a model (Mckelvey, 2002). These criticisms are fruitful to remember when engaging with complex systems.
However, from the more computational perspective, complex systems provide a framework and jargon to describe the phenomena. These may be helpful even for modelling activities: seeking to understand how different characteristics of complex systems may present themselves in the phenomena. It can help to explicate what takes place in the model. For example, Figure 6.1 demonstrated emergence, one of the core features of complex systems. Thus, using complex system theory can help in thinking about what should be in the simulations.