Self-organization theory deals with how global coordination within a system can result out of the local interactions between member parts.1 The theory of self-organization has grown out of many different areas from computer science to ecology and economics. Out of these areas has emerged a core set of concepts that are designed to be applicable to all self-organizing systems, from galaxies to living cells.
Organization is a highly abstract concept but we can loosely equate it to the idea of order, with its opposite being what is called entropy or disorder. Order and entropy are typically measured by scientists in terms of information, that is, the more information it takes to describe something the more disordered the system is said to be.2 An example of this might be a piece of metallic substance consisting of tiny magnets called spins. Each spin has a particular magnetic orientation and in general, they are randomly directed, and thus cancel each other out. This disordered configuration is due to its heat energy causing the random movements of the molecules in the material. When we cool the material down, the spins will spontaneously align themselves so that they all point in the same direction. To describe the state of the spins in this ordered system would involve far less information relative to its original state, a description of which requires unique values for each randomly directed spin.
This process of magnetization is often cited as an example of self-organization, that is the spontaneous appearance of order or global coordination out of local level interactions.3 As we cooled the material down there was some area that had by chance some spins pointing in the same direction. Their alignment generated an increased magnetic force that was exerted upon its neighbors, creating what is called an attractor state which attracts other spins to this configuration. Each time another spin aligns itself with this particular attractor state, it augments the force that is exerted upon other spins through what is called a positive feedback loop that would cascade through the system until all elements were aligned within this new regime.
Another example of self-organization through positive feedback is what is called the network effect, where the more people that use a product or service the greater its value becomes. The telephone and Facebook are such examples, becoming more useful as more users join. In this way, local connections between individuals can rapidly form into global patterns. The network effect illustrates the positive relations or synergies between elements that can come about when they coordinate. It is due to the presence of these synergistic relations that the system as an entirety can become more than the sum of its parts in a process called emergence.
Ant colonies are a classical example given of emergence.4 Despite being governed by very simple rules and only local interactions, ants can through their combined activities generate colonies that exhibit complex structures and behavior that far exceeds the intelligence or capability of any individual ant. These colonies are thus said to have emergent properties. Ant colonies also illustrate the decentralized structure of self-organizing systems. The queen does not tell the ants what to do. Instead, each ant reacts to stimuli in the form of chemical scent exchanged with other ants. In this way, organization is distributed over the whole of the system. All parts contribute evenly to the resulting arrangement.
As opposed to centralized structure exhibited by most social organizations that are dependent upon a single coordinator, this decentralized structure that is inherent to self-organized systems gives them resilience and robustness. This results from the advantage that any damaged element can simply be replaced by any other, giving them huge redundancy. This applies to all self-organizing system from social institutions to technologies and ecosystems. For it to sustain itself over time, it must be able to withstand change and interventions from its environment, requiring the system to be both robust to these perturbations and capable of adapting to changes. The generation of noise and variation within the system is a classical mechanism for achieving this. Without diversity, a system can become rigid and develop into what is called a critical state. An example of self-organized criticality could be an economy whose many industries have developed a dependency upon petrol-chemical fuels. This lack of diversity of energy sources means a small disruption in the supply of petroleum from the system’s environment can have large global consequences.
Inversely, systems with a high degree of diversity between elements will be more robust as the variety between elements makes them more effective at absorbing change. Ecosystems are a classical example of this, generating a large variety of species that make it capable of surviving significant changes within its environment. Thus, we can see how evolution is a core concept in understanding the dynamics of self-organizing systems, whereby attractor states and feedback loops generate the system while periodical perturbations from its environment work to select the most adapted or fittest elements. As information technology is enabling new forms of human organization, people within many domains are faced with practical challenges of how to design and manage self-organizing systems, such as computer networks and new forms of social collaboratives, all of which are making self-organization theory particularly relevant to challenges of the 21st century.