Table 1 Comparison of features between a complex system and a conventional system

Relationship of individual to system

System comprises discrete individuals, who are considered as distinct and statistically independent from each other, but who share the system environment

System comprises individuals, each interacting with others in the system; characteristics of the whole system emerge from these interactions

Context and culture

Context or culture seen as separate from the individuals and may be externally directed or imposed. Treated as a confounder or covariate in analysis

Context or culture seen as emergent properties of the system. In turn, these properties condition the interactions of individuals within the system

Predictability of response to events

Multiple independent responses to change produce a coherent average value response and an approximately normal distribution

Changes to the system are usually buffered by local interaction (so have minimal effect), but sometimes events spread through the system with unexpectedly large effects

Statistical Distributions

Normal distribution for continuous measures, Poisson distribution for events

Heavy-tailed distributions for events: typically inverse power law or log-normal