Of numerous ways of defining complex systems, this one seems useful and relevant to system engineering:
A complex system is a system in which there are non-trivial relationships between cause and effect: each effect may be due to multiple causes; each cause may contribute to multiple effects; causes and effects may be related as feedback loops, both positive and negative; and cause-effect chains are cyclic and highly entangled rather than linear and separable.
The non-trivial nature of the relationships in a complex system make the whole system non-deterministic, ambiguous or chaotic (in the mathematical sense that a very small change in initial conditions may produce a very large change in outcome), even if the individual relationships within the system are well understood.
Complexity as defined above is a property of the system of interest. Complexity is also created in the wider system comprising the system of interest and its stakeholders when the system is not fully understood, and when different stakeholders have different partial understandings of the system and of other stakeholders’ concerns. A major goal of Systems Engineering is to reduce this “perceived complexity” by establishing shared and valid models of the system, in order to improve stakeholders’ knowledge and understanding of the system and its context.
An example of cyclic cause and effect is the biological process of mutualistic symbiosis, in which each of a pair of systems uses the other’s waste as raw material for its own processes. The systems import energy from the environment to sustain the symbiotic processes, so the second law of thermodynamics is not violated. The “Circular Economy” takes this concept and applies it to industrial value chains, to turn them into value loops that are closed cycle apart from import of energy. Waste from one process is the feedstock for the next. If the energy comes from the sun, the value loop can be sustainable as long as energy is available from the sun.
The INCOSE Complexity Primer (https://www.incose.org/docs/default-source/ProductsPublications/a-complexity-primer-for-systems-engineers.pdf) provides a concise introduction to complex systems.
The difference between Complicated and Complex is discussed in, for example, Snowden and Boone (2007), and the INCOSE Complexity Primer (INCOSE, 2015). Complicated systems can be viewed as knowable and deterministic, and once developed their configuration can be “frozen”; whereas complex systems are not fully knowable or deterministic, may be dynamically reconfigurable, and continue to co-evolve with their environment throughout their lifecycle.
INCOSE (2015) Complexity Primer for Systems Engineers, INCOSE, 2015
Snowden and Boone (2007), A Leader’s Framework for Decision Making, Harvard Business Review, Nov 2007