Chapter 2: Networks introduced
2.3 Cybernetics, the study of behaviour; Systems theory, the study of
Networks are a subset of systems that emerged within the Information Age (see chapter 2.2) therefore cybernetics and systems theory, or collectively systems studies, play a key role in the history of networks in the twentieth century. Systems studies in the twenty-first century continue to have much to offer the study of networks as it remains a highly active area. However, it is not the interdisciplinary unified field of study that Wiener envisaged. Instead, starting with different origins, cybernetics and systems theory have remained distinct and today have many more variations conceived to suit different problems, application and schools of thought (Umpleby and Dent, 1997, pp. 2–3). Why do cybernetics and systems theory continue to exist separately when one of their primary aims was to develop an interdisciplinary field that unified researchers with a holistic methodology as an alternative to the division of labour?28
This section addresses this question through a detailed comparative discussion of cybernetics and systems theory as it has a bearing on the proposed networked art's attitude on transdisciplinarity. The section moves the discussion forward from the technological and organismic distinction defined by Wiener and von Bertalanffy within the context of the mechanist-vitalist controversy in chapter 2.1. Through the
finalisation of a definition of systems, the section focuses on current thinking about key points of systems studies and how these reflect concerns of the Information Age, part of the context of the proposed networked art. In the process, two key concepts are revealed as describing the key distinction between cybernetics and systems theory. These concepts will inform discussions of the antecedents of networked art in chapter 3 and form crucial components of the framework of networked art
discussed in chapter 4.
Some scholars maintain that systems theory is a subset of cybernetics while others
28 Wiener, von Bertalanffy and many other key figures in cybernetics and systems theory initially came together in a number of meetings between 1942 and 1947 with precisely this aim of unifying researchers under one field of study. Many of the meetings were supported by the Josiah Macy Foundation and would later become known as the Macy Conferences (Wiener, 1949, pp. 19–33).
that cybernetics is a subset of systems theory (Arnold, 2014, p. 46; Scott, 2004, p.
1369). Gordon Pask, however, states that distinctions between the two are of no significance (Institute for the Study of Coherence and Emergence, 2012). All of these perspectives are in a sense true because each one is based on a scholar's disciplinary origin. Both Wiener and von Bertalanffy's technological and organismic definition of systems (see chapter 2.2) conform to this. Their perspectives are based on their disciplinary origins. In order to understand how cybernetics and systems theory relate to each other and accommodate different perspectives, a partial agreement with Pask's attitude in this matter is required. Therefore, as Pask states there exist distinctions between cybernetics and systems theory and these will be discussed in this section as the reason for their maintained separation. Theory will not, however, be shaped as a means to rationalise differences or widen the
distinctions between cybernetics and systems theory (ibid). Unlike Pask, this research considers the distinctions of “practical significance” (ibid) because it enables networked art to position itself historically and philosophically in relation to network and systems studies.
Stafford Beer, a theorist and consultant who applied cybernetics to management, describes cybernetics in Decision and Control: The Meaning of Operational Research and Management Cybernetics as studying:
“the flow of information round a system, and the way in which this
information is used by the system as a means of controlling itself: it does this for animate and inanimate systems indifferently. For cybernetics is an interdisciplinary science, owing as much to biology as to physics, as much to the study of the brain as to the study of computers, and owing also a great deal to the formal languages of science for providing tools with which the behaviour of all these systems can be objectively described” (1994, p. 254).
William Ross Ashby, a psychiatrist who became one of the leading pioneers in cybernetics, describes cybernetics in An Introduction to Cybernetics as treating “not things, but ways of behaving. It does not ask, 'What is this thing?' but 'what does it do?'…. It is thus essentially functional and behavioristic … The materiality is irrelevant, and so is the holding or not of the ordinary law of physics” (1956, p. 1).
Beer's and Ashby's definitions together with Wiener's definition of cybernetics as being “the entire field of control and communication theory, whether in the machine or in the animal” (1949, p. 19) have clear commonalities. There is no mention of any mesh-like patterns, which had previously applied to networks up until the twentieth century; instead, each definition refers to either material non-specificity or the complete absence of materiality.
Wiener's statement initially opens up the potential of cybernetics as being applied to all machines and animals, while Beer goes a step further by stating that it applies to all “animate and inanimate systems indifferently” (1994, p. 254). Ashby, however, eliminates any focus on a thing and states more clearly than others that “materiality is irrelevant” (1956, p. 1) in cybernetics. Instead, what is important is the behaviour that emerges from the relationship between things. By removing the focus from a thing, Ashby resolves the issue of disciplinary perspectives within cybernetics.
Cybernetics can be considered as the study of behaviour within systems; not what a system is but what a system enables. It can, therefore, be concluded that systems are not just technological or organismic (McLuhan, 1995), constructed or natural or what Wiener terms as machine or animal (1949, p. 19). Neither are they as Beer states animate or inanimate (1994, p. 254), a contrast based on locomotion. All of these are merely distinctions of materiality. Instead, cybernetic systems can be classified as encompassing all material and immaterial systems, or what is termed in this research as all 'real' and 'virtual', as well as possible systems.
In contrast to cybernetics study of systems, systems theory adopts a different approach to the study of systems. It is expressed pragmatically in the Encyclopedia of Systems and Cybernetics:
“no 'point of view' and no 'universe of discourse' on systems could exist without the simultaneous existence of something on which these subtle arts can be practised and which is generally called 'concrete system', or perhaps more imprudently 'real' system” (Institute for the Study of Coherence and Emergence, 2012).
Systems theory therefore prioritises the thing itself (Ashby, 1956, p. 1), a system as an entity (Ackoff, 1959 cited in von Bertalanffy, 1972, p. 9), how the arrangement of
its parts, their interaction and connectedness effectively leads to a structure or a whole that is more than the sum of its parts (Principia Cybernetica Web, 2002 b).
Unlike cybernetics, systems theory is only interested in systems that exist materially with manifestable applications and not the study of systems as a possible concept.
The distinction between cybernetics as studying the behaviour of all systems 'real', 'virtual' and possible, and systems theory studying the structure of only material systems is of key significance to each of their foundations, methods and research.
Cybernetics is a field with a basis in mathematics, physics (Arnold, 2013, p. 46;
Bateson, 1979, p. 227; Wiener, 1949, p. 19) and engineering. As such, it frequently uses logico-mathematical formulae, themselves systems of abstract language, to explain the concept of different types of systems.29 Systems theory, on the other hand, has a close relationship with applied fields such as biology and the social sciences, for example, von Bertalanffy was a biologist, which it explains through written discourse or illustrates in diagrammatic form (Gergely, 1979, pp. 45–46).
Additionally, cybernetics initially focused on isolated systems, commonly referred to inaccurately as closed systems. Closed systems are “able to absorb and emit energy and information, but not matter…it should clearly be distinguished from the isolated system concept [which] does not absorb or emit anything” (Institute for the Study of Coherence and Emergence, 2012). Isolated systems do not, therefore, interact with their environment through input or output. They tend to be only possible systems and lend themselves more easily to being described through systems of abstract language such as mathematics. Systems theory, however, focused on open systems from the outset, that is systems that do interact with their environment, because, as von Bertalanffy maintained, systems “by their very nature and definition are not closed systems” (1972, p. 39). From a systems theory perspective then all material systems are open as they exist in a defined environment. This is evident in biological systems for example, which take input such as nutrients, gases and so forth and produce output such as motion, heat and waste.
In the context of the discussion in chapter 2.1, systems theory can, therefore, be
29 For a discussion of this basis in mathematics and further interpretations of cybernetics that are not possible to explore here see 'The Different Meanings of Cybernetics' (Drozin, 1976).
seen to continue the long-established tradition of applying systems. It employs networks as material, functional systems, which can often be seen in their entirety (although increasingly this is not the case) or represented visually as a structure.
Cybernetics, on the other hand, considers the behaviour of a system as an emergent property; that is that complex behaviour, a non-visual or 'invisible' aspect of a system or network, can be created through the arrangement, interaction and connectedness of parts or nodes and is often represented symbolically. Because of its treatment of behaviour rather than a thing, cybernetics is more capable of conceptualising
systems, material, immaterial or possible, while systems theory, focused on the thing itself, demonstrates a more comprehensive understanding of a material system's structure and the relationship between its parts.
Other points can and have been discussed as distinguishing between cybernetics and systems theory. For example, the consideration of possible or material systems termed indirect and actual by Tamás Gergely (1979, pp. 45–46), closed or open systems (Bertalanffy, 1972) and even mathematical or biological systems (Arnold, 2013, p. 46). Nevertheless, it is suggested that these are secondary differences influenced by the perspective of key figures, such as the mechanistic and
mathematical influence of Wiener on cybernetics and the biological influence of von Bertalanffy on systems theory, or are a result of cybernetics’ focus on system behaviour and systems theory's focus on system structure.
“we might say that systems theory has focused more on the structure of systems and their models, whereas cybernetics has focused more on how systems function, that is to say how they control their actions, how they communicate with other systems or with their own components … Since structure and function of a system cannot be understood in separation, it is clear that cybernetics and systems theory should be viewed as two facets of a single approach” (Principia Cybernetica Web, 2002 b, italics in original).
Cybernetics and systems theory should therefore not be considered as a split or division of labour, but rather as two allies with shared methodologies, research and overall objectives that only differ in focus within the project of systems studies (Gergely, 1979, pp. 46–47). As a result of this they, through combination, treat
systems as a whole and adopt a holistic methodology for studying systems. As such, distinctions between the two can and should be considered as a means of
distinguishing between their research.
Consequently, it is proposed that within the context of this research cybernetics should foremost be considered as a general study of system behaviour and systems theory as a general study of arrangement to form a system's structure. Viewed holistically they are “two scientific approaches, organically complementing each other” (Gergely, 1979, pp. 46–47) and pushing each other forward. In essence, systems theory identifies or builds systems and maps their structure while cybernetics studies the behaviour of systems and its emergent properties. This proposed relationship between systems theory and cybernetics raises several issues which all stem from or are influenced by the question, how precisely do arrangement and behaviour relate to each other within systems studies? The arrangement and behaviour relationship is cyclical with each part of the relationship influencing or feeding back on the other. Elements are arranged to act on each other, cumulatively creating behaviour and it is this which qualifies the arrangement of elements as a collection of nodes in a system or network (Backlund, 2000). The behaviour that emerges within the system is a result of the arrangement. Arrangement, to an extent, initially defines or shapes what behaviour can occur. In turn, behaviour can feed back into the system, completing the cycle, re-arranging the nodes as a means to a modified or new and perhaps more efficient or complex behaviour. This
adaptation can occur as a result of the system configuring itself internally, or if it is an open system as a response to external input. The identity of the system or network is therefore dependent on the collective behaviour of the nodes; it is in effect a holistic arrangement.
In order to understand the relationship between arrangement and behaviour, it is necessary to understand the process of identification of a collection of nodes as a system. It is mentioned above that what qualifies a collection of nodes as a system is their arrangement to act on each other. This statement is true. However, an initial factor must be considered. The identification of a system requires that “there must first be given an observer (or experimenter); a system is then defined as any set of
variables that he selects from those available” (Ashby, 1960, p. 16).30 It is in this that a significant complexity arises; if an observer selects nodes to be identified as a system or network, they have an initial influence or input on the system.
Consequently, the system must be open as von Bertalanffy postulates all material systems are (1972, p. 39). More importantly, since the observer has an input on the system, it questions whether the observer is, in fact, a part of the system and
whether the system remains the same following that input. Heinz von Foerster discusses this central issue.
“In working to derive functional models common to all systems, early cybernetic researchers quickly realized that their 'science of observed systems' cannot be divorced from 'a science of observing systems' — because it is we who observe” (1974 cited in Pangaro, 1990).
Alexander Backlund provides a sophisticated discussion of systems in a paper titled 'The definition of a system' (2000) that addresses and largely resolves this issue. In Backlund's definition, Ashby's observer is not automatically assumed to be part of a system because of input to the system. Firstly, a system should not be defined by
“any specific kind of relation (like interaction)” (ibid, p. 448), so the process of selecting elements to be nodes is not a sufficient criterion for the observer to be considered a part of a system. Secondly, “if an element affects parts of the system but is not affected by it [or] if an element is affected by parts of the system but does not affect any part of the system” (ibid) then it is outside the system.31 This,
therefore, allows only three possible types of elements: those that affect a system but are not affected by the system (Fig. 2.3.1); those that do not affect a system but are affected by the system (Fig. 2.3.1); and those that both affect and are affected by the system.
30 Ashby's statement implies that the observer is human however it should be noted that Pask defines observers as “men, animals, or machines able to learn about their environment and impelled to reduce their uncertainty about the events which occur in it, by dint of learning” (1968, p. 18). Both use masculine pronouns however it can be assumed this is of little importance to the discussion.
For a full definition of observer as it relates to this research, please refer to the glossary.
31 Whether an element is part of a system or not according to the conditions of affecting or being affected by, it conforms to Newton's third axiom or law of motion. Newton's law states that “to every action there is always opposed an equal reaction: or the mutual actions of two bodies upon each other are always equal, and directed to contrary parts” (Newton, 1687, p. 83).
Figure 2.3.1: Element (f) affects a system (a, b, c, d) but is not affected by the system. Element (e) does not affect a system (a, b, c, d) but is affected by the system (Backlund, 2000, p. 449).
Figure 2.3.2: The system (c, d) is a
subsystem of the suprasystem (a, b, c, d) as it affects and is affected by it. Element (a) affects (c, d) and is an example of an observer. Element (d) is affected by (c, d).
Neither (a) or (b) are part of (c, d) however they are part of (a, b, c, d) (Backlund, 2000, p. 450).
According to Backlund, it is only in type three that elements can be considered nodes within a system. Ashby's observer, therefore, can only be part of a system if affecting and being affected by the system; that is there must be bi-directional behaviour. It is clear that the process of selection by an observer is an affect on a system as it defines it, even if only initially. However, whether a system then affects an observer is more complicated. An observer, called S1, is both a biological system and a social being that is part of society as a system. As such, they are a system and are a part or node in at least one system. By selecting elements and identifying them as a system, called S2, S1 has already changed the identity of the elements. They are no longer independent. They are nodes; co-dependent parts of S2. This change of identity is fed back and registered by S1. It may inform subsequent behaviour by S1, if any, with S2 and so S2 is in turn affecting S1, initially and possibly in an ongoing manner. The process of identifying S2 and the resulting feedback of its status change to S1 creates yet another but larger system within which both S1 and S2 are incorporated. Backlund terms this system, called S3, a suprasystem (ibid, p. 449).
So in total, there are three systems: the observer S1 as a biological system, the system the observer identifies S2 and the new suprasystem S3 that incorporates both of the former systems as subsystems. In Backlund's diagrams (Figs. 2.3.1 and 2.3.2) these are respectively labelled as (a), (c, d) and (a, b, c, d). Each system may
continue to exist independently, merge or dissipate dependant on ongoing bi-directional behaviour. For example, if the process of creating S2 only entails its identification/feedback and S1 and S2 do not continue to affect each other, then S3 will cease to exist.
Continuing to use the terms observer and system as distinct entities should now be seen as highly problematic. While on the one hand, they do identify S1 and S2 as potentially two different types of systems, both are also nodes within S3 and
therefore in a sense equal. However, it is the meaning of the term observer, a term that typically defines a non-behaving role, which is most problematic. It can be viewed as an unfortunate terminological remnant of pre-systems studies methodologies, such as the division of labour, where an observer is classed as external, non-influencing and impartial to that which is observed; a position that is rarely achieved.
Pask identified issues surrounding human observers in pre-systems studies methodologies, in particular when the system being observed involved human participants. He proposed instead that a normative paradigm should be accepted where the observer can and quite often does act on the system causally (1975, p.
13–19); that is they create ongoing input in relation to the system’s output. He states that in reality, many “observers act as participants” (Pask, 1980, p. 399) and that “an observer who comes to know the system must be a participant in the system” (Pask,
13–19); that is they create ongoing input in relation to the system’s output. He states that in reality, many “observers act as participants” (Pask, 1980, p. 399) and that “an observer who comes to know the system must be a participant in the system” (Pask,