Von Foerster (1960) is one of several seminal accounts of self-organisation in the early cybernetics literature. See also for example Ashby (1966), Pask (1963), Beer (1967, chapter 14). The Pask paper is a lengthy review essay that develops the discussion from first principles and serves as an excellent introduction to the topic. The paper goes on to develop generalisations and interpretations for human psychological and sociological domains that anticipate Pask's later articulation of a cybernetic theory of conversations (Pask and Scott, 1972, Pask 1976, Scott 1982a, 1982b, 1993). Beer's chapter is a beautiful account of the complexity of organisation found in highly evolved natural systems. Beer is reflexively aware that he, the writer, is an example of such a system. He is at pains to insist that it is as observers that we distinguish their complex architectures as having specific structures and functions but the systems are wholes that have evolved. He teasingly suggests to the reader that the "goal" of a self-organising system is for it to "learn to be what it is".
Von Foerster (1960) defines a self-organising system as one in which, as measured from the observer's frame of reference, the rate of increase of redundancy is always positive. Redundancy refers to the degree of order or regularity in the system. The greater the redundancy the greater the more "complex" is the system as an organised, integral whole. An immediate consequence of von Foerster's definition is that, as time passes and the system grows or evolves, the observer is obliged to update and enlarge his description of the system and its possible states and behaviours. In so doing, he may also eventually be obliged to modify his frame of reference, adding new categories or dimensions. There is a logical problem in all this: "If the system is changing, what makes it the same system (other than just by edict of the observer)?" Ashby, von Foerster and other cyberneticians of their generation were well aware of this problem of system identity and acknowledged that the concept self-organising system is contradictory. However, the term continues to be employed.
Von Foerster's definition may seem quite innocent and elegant in its simplicity. However, there are deep implications. "Circular causality" implies that we have a system that is "information tight" or "organisationally closed". As Maturana later emphasised with his concept of autopoiesis, it is this circularity of organisation (processes produce products that embody the processes) that characterises the living from the non-living (Maturana and Varela, 1980). As noted in the previous section, only those adaptations that leave the integrity of the circularity undisturbed are allowed but note that, in principle, the set of possibilities is unbounded. In practice, real evolving systems may commit themselves to a particular adaptive solution that rules out others but there may always be a backtracking to the choice point. The concepts of growth, evolution and self-organisation add to the account of the last section. As long as their joint outcome is to preserve the integrity of the organism, the functions F and G that map sense to action and vice versa may change and evolve as the organism learns and comes to know.
From a first-order perspective, self-organising systems are embedded in the larger universe in which the second law of thermodynamics is at work: entropy, as disorder, increases as energy is dissipated. Ordered structures appear as pockets of "negative entropy", where, temporarily, the second law appears to be reversed and redundancy increases. Eddies in a stream are a relatively simple example. Schroedinger (1943) is an early account of these processes. Prigogine (1980) refers to the formation of any such "dissipative" structures as "self-organisation". That there is such "system emergence" is a reflection of a deeper "lawfulness" at work. Engels (1940), following Marx and Hegel, enunciated a general cosmic principle, the law of "quantity to quality", which states that increase in any variable quantity will eventually lead to a qualitative change. This is, of course, a very general "law', encompassing phase transitions in physical systems, Prigogine type self-organisation in chemical systems and von Foerster type self-organisation in biological and social systems. Indeed, the whole cosmos may be viewed as a self-organising system. Von Foerster's definition emphasises processes of growth and evolution and the limits imposed on an observer who wishes to model those processes. What is not generally recognised is that the existence of these limits also implies that the concept of a "causal mechanism" is itself held in question. McCulloch (1965) refers to "that great fiction, causality". Bateson (1972) stresses that in cybernetics, the question asked of complex systems is not what is causing what but rather, why did that particular event come to be, given the many alternative possibilities?
Paul Davies (1992), in a discussion of the evolution of complexity, points out that it is possible both for the cosmos to be determined (as by a set of well-specified equations) but for it to be in principle impossible for it to successfully simulate itself and hence predict its own next state. "Causality" at work in the cosmos leads to a creative becoming.
These points are perhaps helpful in understanding the cybernetic enterprise. There are those who perceive that cybernetics is about "machines" and assume the discipline has a materialist, mechanistic, reductionist, determinist, physicalist metaphysics, whereas, in fact, as Gunther has stressed, cybernetics is not committed to a particular ontology. Rather, cybernetics studies the world "as if" it were a mechanism (see Glanville (1994) for an elegant recent statement of this position). It is in its rigorous analysis of "all possible machines" that cybernetics reveals the untenability of a simplistic, mechanistic metaphysics and notes the "in principle undecideable" nature of ontological, essentialist questioning. Confusion is compounded because there are many workers in the physical, computing, biological and cognitive sciences who do hold mechanistic, physicalist views and who do seek to understand the "essential nature" of nature. To the outsider it may look as if all are "doing cybernetics" - and in a sense they are - but they are not cyberneticians in the tradition as described here.