System behaviour, such as is expressed in the formulations on the foregoing pages, can always be related to the concept of complexity. The more complex a system, the more intricate its behaviour. It is, however, necessary to bear in mind that, given enough time and space, even the simplest system can produce quite unexpected and surprisingly complex phenomena. To emphasize the characteristics of a complex system, the following comparison between simple and complex systems has been made by R. Flood and M. Jackson (1991).
Simple systems are characterized by
- a small number of elements
- few interactions between the elements
- attributes of the elements are predetermined
- interaction between elements is highly organized
- well-defined laws govern behaviour
- the system does not evolve over time
- subsystems do not pursue their own goals
- the system is unaffected by behavioural influences
- the system is largely closed to the environment
Complex systems are characterized by
- a large number of elements
- many interactions between the elements
- attributes of the elements are not predetermined
- interaction between elements is loosely organized
- they are probabilistic in their behaviour
- the system evolves over time
- subsystems are purposeful and generate their own goals
- the system is subject to behavioural influences
- the system is largely open to the environment
A large system per se normally signifies a greater complexity inasmuch as more subsystems and more processes are in operation simultaneously. The degree of organization inherent in the system, defined as predetermined rules guiding the interaction, is another basic determinant. Non-linear and stochastic processes with many feedback loops of higher order and time delays are also important.
A complex system often behaves in an unexpected manner and the relations between cause and effect are often difficult to understand. Measures taken to understand or control may sometimes yield the opposite of our intentions. Measures seemingly reasonable in the short- term often prove to be harmful in the long run. Human interference with delicate regulation mechanisms may cause changes which lead quite abruptly into a new state, essentially irreversible and continuing for a very long time.
A system is generally less sensitive to external structural influences than to internal. The complex system can nonetheless be unsusceptible to changes in the internal parameters. Increases or decreases in their values are neutralized by many kinds of negative feedback and these changes have little influence on system behaviour. Linear negativefeedback systems are either stable or unstable regardless of the input signal applied. Non-linear feedback systems may be stable for some inputs but unstable for others.
The general unifying forces keeping a systemic hierarchy together vary according to their evolutionary direction. Systems which evolve upward possess strong cohesive forces joining the subsystems and are thereby less easily disrupted. Lower levels generally function better than the later established suprasystem. In systems which evolve downward, developing specialized subunits, the suprasystem is stronger than the younger subsystems.
All feedback systems are apt to oscillate, affecting the behaviour, to a lesser or greater extent. Within the existing network of coupled variables each variable has a highest and lowest threshold. Within these limits the system can vary freely; if they are exceeded, disorder and finally collapse will occur. If they do not maintain a defined value, other variables will then occupy the available variation space and take over. For the systems researcher it is a constant problem to know whether certain feedback loops reveal significant differences, or merely amplify insignificant ones.
In Figure 2.21, A shows a state of stable oscillations where the input is a feedback from the output. This occurs when the feedback (thin line) has a phase which is the opposite to the system disturbance and is of equal amplitude. In B, the oscillations are damped and diminish when the feedback is less than the output. Finally, a feedback signal inducing corrective action greater than the error will amplify the same, causing growing oscillations and instability, according to C in Figure 2.21. A special kind of system behaviour is associated with system growth and adaptation. The introduction of a unique input at some critical time may suddenly permit a semi-organized system to organize itself into a hierarchy and to grow. The general unifying forces keeping a systemic hierarchy together vary according to their evolutionary direction. From these facts follows the inevitable conflict in all growing systems, that existing between the supra and subsystems.
Figure 2.21 Different oscillation patterns in feedback Systems (from Tustitt 1955).
The fact that living systems per se tend to complicate their interactions with their surroundings over time, is an inherent feature of their growth. This applies also to non-living systems — the growth of a computer network is a good example. Growth is mostly a consequence of an adaptation; bigger systems survive better than small ones. Inasmuch as it is not possible to adapt to everything, a system is prevented from growing to infinite size. Beyond a certain point integration and communication problems within the system exceed the benefits of large size. To maintain diversity and balance, un organism may not exceed the norm of its species. It eventually reaches its maximum stage, beyond a growth without deterioration is not possible.
Structural growth and the associated increase of size demand a specialization or modification of some components which in turn produce emergent systemic properties. A wider range of functions in a system with specialized subsystems makes it better equipped to cope with even unforeseen difficulties. The total system may thus have a longer life than its subsystems.
Source: Skyttner Lars (2006), General Systems Theory: Problems, Perspectives, Practice, Wspc, 2nd Edition.