Autopoiesis and the science of administration: essence, sense and nonsense

Autopoiesis and the science of administration: essence, sense and nonsense – public

Walter J.M. Kickert


A remarkable number of applications of the concept of ‘autopoiesis’ have appeared in various fields of the social sciences. The popularity of ‘autopoiesis’ in the social sciences seems to stem from the work of the sociologist N. Luhmann in the mid-eighties (Luhmann 1984). He used the model of a living system — autopoiesis — which was formulated by Chilean biologists (Varela, Maturana and Uribe 1974) to describe how a living system is able to generate and regenerate its own organization, in the development of a more general theory of self-referential systems centred around the key concept of ‘communication’. The law theorist G. Teubner, who was struggling with paradoxes of self-reference in modern methodological approaches, picked up Luhmann’s theory of self-referential systems and communication and developed an autopoietic approach to law (Teubner 1988). The Marxist political scientist B. Jessop, interested in the problem of why the capitalist system could survive despite its tendency for imminent crisis and continuing class struggle, used the autopoietic theories of Luhmann to develop a model for radical autonomy in societies (Jessop 1990). The organization scientist G. Morgan used autopoiesis in one of his metaphors about organizations (Morgan 1986). Morgan created an interpretation of autopoiesis which led him to intriguing ideas on self-referential closure, ego-centrism and self-reflective evolution of organizations.

The remarkable popularity of autopoiesis in quite a number of highly different fields of social science, makes it worthwhile to have a close look at it.

The model seems particularly interesting for the advancement of the science of (public) policy and administration because it seems to offer new insights into the self-governance of social organizations. At a time when top-down direct government steering is increasingly being replaced by self-responsibility and autonomy of social institutions, and theories on central governance from super-ordinated positions are being replaced by theories on inter-organizational networks and non-compulsive cooperation, the need for theoretical insights into self-governance is apparent. The autopoiesis model seems to offer these.

Moreover, the autopoiesis model seems to open a fundamentally different perspective on the relationship between organization and environment. The usual organization science perspective that an organization adapts to its environment, or is at least influenced by it, is fundamentally turned around. An ‘autopoietic’ organization, on the contrary, is self-referentially closed. It only perceives its environment as a projection of its self-identity. It only functions in order to survive, and to maintain its self-identity. The usual relationship between organization and environment apparently is radically reversed in the autopoietic perspective. Thus from the organization science point of view, the model seems intriguing.

In this paper, the applicability of the autopoiesis model to administrative science will be discussed.

The model has already stimulated some outstanding social and administrative scientists to creative thinking about analogies and possible implications. The model apparently triggers lateral thinking and seems to be a rich source of creativity. In order to assess the applicability of ‘autopoiesis’, it is, however, not sufficient to judge whether or not the various interpretations are ‘creative’. The question then arises of whether or not these ideas have much to do with the original model.

In order to apply a theory to the social sciences that has been derived from the natural sciences, it is necessary to fully understand the original, and to be cautious in converting and applying it. I had some doubts about the full understanding and cautious conversion when I considered some social scientific ‘applications’ of the biological model of a living organism: autopoiesis. As a matter of fact, the originators of the model themselves explicitly opposed the idea of stretching the model beyond the area of biology (Varela 1981). It is therefore probably wise to start our consideration of the applicability of autopoiesis by first explaining what the natural scientific model originally meant. The primary fields of origin of autopoiesis are biology and systems theory. Only after this exploration of the homelands of the model will we discuss the various applications of autopoiesis to different fields of the social sciences and consider its usefulness for (public) administration.

The origin of the model in biology, its development in systems theory, and its application in administrative science reflect to a great extent my own career in science, being a physicist by training, having arrived via control and systems theory at a dissertation in organization science, and thereafter proceeding to the science and practice of public administration. This enables me to understand both the ‘hard science’ aspects of the model and assess its usefulness for the ‘soft’ administrative science.

It should be kept in mind though that, from the viewpoint of usefulness, it is not so important whether a useful idea is an accurate translation of the original natural scientific model, but rather whether the idea is interesting and relevant for administrative science.


The Original Model

In 1974, two Chilean biologists Varela, Maturana and the systems theorist Uribe, published an article on living systems. They were in search of the essential characteristic which distinguishes such systems as ‘living’ systems. In other words, they were in search of the definition of ‘living’. Contrary to the prevalent biological definition of living, the ability to reproduce, they argued that a living organization can only be characterized by the network of interactions of components which constitute a living system as a whole. These relations define a complex system as a unity. They constitute its organization. Instead of looking at what makes a living system reproduce the parts of the system, its components, they looked at the organization of the living system that is reproduced. In the manner of true holistic systems theorists, they did not emphasize the separate parts but concentrated on the whole.

In their view, it is not reproduction as such, but rather the reproduction of the organization of a living system, the network of interactions between the components, which makes it ‘living’. In this manner they arrived at their definition of an autopoietic system as a ‘network of productions of components which (i) participate recursively in the same network which produced them and (ii) realize the network of productions as a unity’ (Varela et al. 1974).

Such a system is autonomous: it reproduces itself, its parts and its organization. An autopoietic system is organizationally closed with no apparent inputs and outputs. This is in contrast to an ‘allopoietic’ system, such as a mechanism in which the product of the operation is different from itself. An allopoietic system is open, receives inputs and produces outputs.

An autopoietic system is an autonomous and self-sustaining unity with a network of component-producing processes such that the components in interaction recursively generate the same network of processes which produced them. The process of continuous reproduction of the systemic unity is called autopoiesis.

In summary, the original model of autopoiesis has the following essential characteristics:

* It is a definition of what makes a living system ‘living’.

* It is an extension of the usual definition of living — a living system is capable of reproducing itself — in a holistic systems direction. Not the parts, but the organization of the living system which makes it a unity, a whole, is what matters.

* Living means the reproduction of the organization of the living system, the network of interactions between its components.

* An autopoietic system is an autonomous unity which consists of a network of component-producing processes such that the interactive components recursively generate the same network of processes.

* Such a system is organizationally closed with no apparent inputs or outputs.

The first paper on autopoiesis in English (1974) was based on a book by Maturana and Varela written in 1973 which was later translated into English from Spanish (Maturana and Varela 1980). This remarkable paper, which attracted wide attention in systems theorists’ circles, was followed by a number of publications by the same authors (e.g. Maturana 1975, 1981; Varela 1979, 1981; Uribe 1981) in which various aspects of the model were further clarified and elaborated. In particular, their articles in the volume edited by Zeleny (1981) emphasize how autopoiesis should not be interpreted. According to Maturana (1981), it is not identical to autonomy or to self-organizing. In his treatment of organizational closure — recursively organized systems — Varela (1981) stresses the limits of autopoiesis: the concept should not be used beyond the field of biology. Apparently the originators of the model were quite reluctant about its wider applicability.

Autopoiesis and Biology

The illustrative model of an autopoietic system that Varela et al. (1974) used clearly resembles the model of a biological cell. Their simple simulation model of autopoiesis consists of a substrate in which a catalyst is capable of producing a link which is able to bond with other links or to disintegrate into substrate. The computer simulation shows the spontaneous generation and reproduction of chains of bonded links around a catalyst. Biologically stated, it is a cell in which a catalyst nucleus generates and reproduces a membrane around itself.

Genetic theories from biology about cells, DNA molecules and genes have undoubtedly contributed to the invention of the autopoietic model. A single gene in a DNA molecule contains everything necessary to generate and reproduce a certain property of a living being.

It is doubtful, however, whether genetic biologists would consider the autopoietic definition of ‘living’ as new. In genetics, reproduction has everything to do with structure. A cell has a nucleus. A nucleus contains chromosomes. A chromosome consists of strings of DNA and these DNA strings are composed of genes, which code certain properties of the living being. A gene is a particular order of certain nucleic acids. The basic coding element for the properties of a living being, the gene, is thus defined by its particular ordering, in other words, its structure or organization as a unique unity. Reproduction has to do with structurally composed wholes. This is common knowledge to biologists, and not a new contribution of autopoiesis.

Epigons of Autopoiesis in Systems Theory

The implication of the model for the advancement of systems theory — its secondary field of origin — is more relevant.

In the late seventies, the idea of autopoiesis stimulated a number of systems theorists to further thought and reflection. The mathematical elaboration of autopoietic systems, using the automata theory of computer science, received relatively early attention (Zeleny 1977; Goguen and Varela 1979). The over-enthusiastic sweeping remarks about the wide applicability of the concept to human relations are mostly ‘soft talk by hard scientists’. In fact, these remarks are examples of the exaggerated pretentions of universal applicability which system theorists were often blamed for, a view which has considerably contributed to the decline in appreciation of the systems theoretical ‘paradigm’ in the administrative sciences since the seventies.

In some systems theoretical circles, the development of the model of autopoiesis was welcomed as a possible breakthrough which could restore the declining esteem for systems theory in the ‘applied’ sciences.

Apart from mathematical elaborations of autopoiesis — which, for our purpose, are neither interesting nor relevant — the model has stimulated other systems theorists as well. In the volume edited by Zeleny (1981), a number of articles which extend the original model are noticeable.

Jantsch (1981) extends self-reproduction outside the realm of living. He formulated a general dynamic systems theory with ‘dissipative self-organization’ as the unifying perspective. Jantsch was afraid that autopoiesis would otherwise become sterile. He distinguished two forms of self-organization. The first consists of conservative attractive and repelling forces leading to stable equilibrium. The second — called ‘dissipative structures’ — is a self-organizing dynamic order maintained by the continuous exchange of energy with the environment, spontaneously arising in far-from-equilibrium conditions, establishing a certain autonomy. Continuous autopoiesis depends on the continuous maintenance of non-equilibria. Jantsch’s interpretation of autopoiesis — called a ‘misunderstanding’ by Maturana and Valera — is related to notions from chaos theory, e.g. about ‘order out of chaos’ (Prigogine and Stengers 1985) which we will discuss later on.

In the same volume, Ben-Eli (1981) took a much more classical cybernetic stance towards autopoiesis. Equilibrium, stability, regulation and control were key notions in his elaboration of autopoiesis into a notion of evolution. The concept of self-organization was seen as a basis for the idea of stability in viable dynamic systems. A self-organizing system consists of both the idea of consistent identity and the idea of dynamic variations essential for its continuous viability. Notice that in the same volume, Maturana (1981) stated that if the invariance of organization is the essence of an autopoietic system, such a system cannot undergo a change in its organization, that is, the system cannot be ‘self-organizing’ in an evolutionary sense.

Elsewhere (Geyer and Van der Zouwen 1986) other renowned systems theorists presented their ideas about autopoiesis. Laszlo (1986) elaborated on the cybernetics of social evolution. He distinguished between stability and ‘resilience’, the latter being more important for the long-term persistence of ecosystems. Resilience is the ability to absorb change and disturbance within a wide range of large and unexpected fluctuations. A system might be stable with respect to small deviations, but unstable for large ones.

The aforegoing incomplete sample of systems theoretical publications related to autopoiesis clearly shows that, even in its own field of origin, elaborations and applications of autopoiesis use broad interpretations of the original model, the validity and the accuracy of which is contested by the original authors.

How does this relate to the introductory statement that, in order to apply a natural scientific model, one has to be cautious in converting and applying it?

If one accepts the self-imposed restriction by Varela that autopoiesis should not be used beyond biology, our discussion about application and transposition is pointless. The same holds if the restriction of mathematical formality of the model is imposed. If one drops biology and mathematics, the essentials of the model come down to the self-reproduction of the organization of the autopoietic system.

However, what if even that essence is dropped and replaced by the much broader idea of self-organizing? A self-organizing system does not necessarily invariably reproduce one and the same organization. Ben-Eli (1981) and Laszlo (1986) use self-organization as an idea for their theories on evolution. Evolution implies a long-term change in organization. Jantsch’ (1981) ideas on order out of fluctuations are even less compatible with regeneration of the same organization.

Autopoiesis then seems reduced to some very broad idea about self-referentiality, and as such shows striking similarities with certain modern methodological approaches in the social sciences. As we will see from the following discussion on its applications in social science, that is indeed more or less what is left of the cautious transportation of the original model of autopoiesis.

Autopoiesis in Social Sciences

Luhmann’s Autopoiesis of Social Systems

Following the popularity of autopoiesis amongst systems theorists in the late seventies and early eighties, the concept broke through to the social sciences in the mid-eighties with the publications of the renowned German sociologist N. Luhmann (1984, 1986). In his book on ‘social systems’ Luhmann (1984) introduced the idea of a paradigmatic shift in systems theory from the holistic notion on parts and whole, via the distinction of system and environment, towards a theory on self-referential systems. In the constitution of their elements and operations, systems refer to themselves. They are self-referentially closed. This new theory of self-referential systems emerged from theories on self-organizing systems developed in the early sixties. Autopoiesis is one of the more recent systems theories he used.

He extended autopoiesis, a biological model of living systems, to the realm of social systems by interpreting autopoiesis as a form of system-building using self-referential closure (Luhmann 1986). In Luhmann’s view, communications are the basic elements of social systems. They are recursively produced and reproduced by a network of communications. Communication unities consist of a synthesis of information, utterance and understanding. This synthesis is not produced by some outside force like language, but by the network of communication itself. It requires self-reference. Luhmann saw the maintenance of social systems as a self-referential production. Autopoiesis presupposes a continuous need for renewal. Continuing dissolution of the system becomes the cause of its autopoietic reproduction. Maintenance is not a matter of reproducing the same patterns in similar circumstances, but rather the production of subsequent elements different from previous ones. According to Luhmann, an important contribution of autopoiesis was its epistemological consequence of distinguishing autopoiesis and observation, observing systems being themselves autopoietic systems.

Luhmann’s interpretation of autopoiesis can indeed be characterized as very broad. Self-referentiality is broader than self-organization, which is broader than self-reproduction, which is what seems left of autopoiesis if one drops biology. Moreover, Luhmann uses self-referentiality primarily as a methodological analogue for his epistemology. So, autopoiesis has in fact become an abstract philosophy. Furthermore, one has to realize that the notion of communication plays a much more central role in his social systems theory.

Teubner’s Autopoietic Law

In his development of a constructivist epistemology of law, Teubner (1988, 1989) referred extensively to Luhmann’s epistemological interpretations of autopoiesis. In his journey towards an epistemology centred around ‘social construction of reality’ and ‘decentring of the subject’, Teubner discussed the poststructuralism of Foucault and the critical theory of Habermas. In his view, both Habermas and Foucault became entrapped in paradoxes of self-reference, be it infinite regression, circularity or tautology. Luhmann’s theory of autopoiesis offered him a way out in the sense that autopoiesis does not treat self-referentiality as a paradox to avoid, but as its useful essence. Autopoietic systems are based on that very self-referentiality. Teubner derived a theory of law as an autopoietic system, a network of communications that recursively reproduces itself. Following Luhmann, he considered law to be made up of legal communications. These do not derive their meaning from some outside ‘real’ world. The outside world itself is a social construction by the law. Persons are mere constructs, semantic artifacts produced by the legal discourse itself. Autopoiesis provides Teubner with the epistemological basis to support these theses.

Note that Teubner primarily used Luhmann’s interpretation of autopoiesis as an epistemological foundation. Hence, my previous comments on Luhmann’s interpretation also apply here. The level of abstraction reached by Teubner seems light-years remote from the original model of autopoiesis.

Jessop’s Autopoiesis in Politics and the State

Luhmann’s autopoietic ideas on communicative processes also formed the basis of Jessop’s (1990) ideas. Jessop viewed autopoiesis as a condition of radical autonomy. An autopoietic system defines its own boundaries, develops its own unifying code, reproduces its own elements, and obeys its own law of motion. It is not controlled from outside and there is no superordinate governing centre; the interactions between different autopoietic systems leads to blind co-evolution.

Referring to Luhmann, Jessop remarked that the political system in modern societies is no longer a paramount power. It is differentiated and involved in complex interdependencies with other autonomous autopoietic systems. It is hard to maintain the myth that the state can solve societal problems. The state is a collective fiction, a semantic artifact, which serves as a self-reference of political responsibility in a complex power circuit. Jessop gave the example of welfare politics as a self-referential and closed discourse.

Jessop also discussed the problem of societal guidance. If social systems are radically autonomous autopoietic systems, what is the unity of a society? A society would then be less than the sum of its parts; a series of self-closed systems engaged in a blind co-evolution. Jessop searched for a solution using the idea of ‘structural coupling’ between mutually indifferent systems which co-exist and co-evolve in the same eco-system. Such a mechanism could explain how radically autonomous systems can co-evolve and develop into stable, coupled blocs.

Assessing the merits of (Luhmann’s) autopoiesis from a Marxist perspective, Jessop concluded that the notion of operational self-closure of an autopoietic system and the mechanism of structural coupling between autopoietic systems have shed new light on the crucial concept of autonomy and its relativity. Autopoiesis serves to support, on the one hand, the view that society cannot be organized and controlled from a single superordinated centre — contrary to usual Marxist notions of class — because it consists of autonomous part systems and on the other hand, helps to recognize that there are stable networks in co-evolution in society.

As Jessop also only refers to Luhmann’s interpretations, a report of earlier comments is unnecessary. From the viewpoint of usefulness for administrative sciences, Jessop’s interpretations are more in line with modern administrative theories on the limitations of central government steering; the departure from a monocentric towards a polycentric theory of the state; the recognition that policy-making usually consists of a network of nearly autonomous actors; inter-organizational network theories, etc. These issues will be discussed in a later section.

Morgan’s Logic of Self-producing Systems

One metaphor in Morgan’s (1986) well-known collection of metaphors of organizations is that of the organization as ‘flux and transformation’. In search of the logics of transformation and change for the basic dynamics that generate and sustain organizations, Morgan explored three different images, one of which was the idea of autopoiesis. In contrast to the aforementioned authors, he derived his ideas directly from the original publications of Maturana and Varela. Well aware of their strong reservations about applying a biological model to the social world, he used it only as a metaphor. Morgan introduced the idea of autopoiesis as a basic challenge to the traditional approach in organization theory in which an organization is typically viewed as an open system in interaction with its changing environment, which has to be adaptive, in order to survive environmental challenges. An autopoietic system, on the contrary, is organizationally closed, autonomous and only self-referential. Autonomy, circularity and self-reference are its key features. This opens radically different perspectives on the relationship between organization and environment. In an autopoietic system, relations with the environment are internally determined. This has important implications for the maintenance of the organization’s identity. Morgan’s creative interpretation of autopoiesis led him to three intriguing conclusions. The first is self-referential closure, the attempt by organizations to interact with their environment as projections of themselves. The second is what he called ‘egocentrism’, the attempt by organizations to try to maintain their own identity against a threatening outside world. Organizations become preoccupied with, and over-emphasize, the importance of themselves, underplaying the significance of their wider context. The third is self-reflective evolution, the process of organizational change as an evolution of self-identity in relation to the wider world.

Although Morgan regards the theory of autopoiesis as a major challenge to understanding organizations, he also repeatedly underscores for its dubious implications. Egocentric corporate cultures may be successful in the short run, but often at the expense of their context, and they run the danger of destroying the whole. Morgan derived as the motto of this approach, ‘think and act systematically: more self-reflection, less self-centredness’.

Autopoiesis Reconsidered

Although the above-mentioned interpretations of autopoiesis from the respective fields of application undeniably led to synergetic surplus values, one should realize that the various translations and uses of the model of autopoiesis were indeed rather creative. The transpositions differ so substantially from the original that one might wonder if they really were correct applications at all. Although it can be argued that from the viewpoint of the field of applications, it is not so important to ascertain whether a good and useful idea is an accurate transposition of an original idea from the natural sciences, the focus of this paper, however, is also to assess the applicability of the autopoietic model. In this section we will therefore first return to the question of what essential innovations autopoiesis has contributed to systems theory. What is the surplus value of autopoiesis and what are its limitations?

System Maintenance, Stability and Autopoiesis

Maintenance is at the heart of the concept of ‘homeostasis’, and is the cybernetic counterpart of the control engineering concept of ‘servomechanism’. A homeostatic system is capable of self-maintenance and recovery from disturbances. Not only in technical engineering does stability play a crucial role, but in many other fields of both natural and social sciences as well. It should not be hard for political and administrative scientists to imagine situations in which one would be interested in knowing whether a particular system will be able to survive, to maintain itself in the midst of disturbing events such as a crisis, or whether it will, in fact, disintegrate.

The concept of ‘stability’ is defined as the capacity of a system to regain an initial state after a disturbance. The formal definition of stability is that the difference between initial and actual state will in time become zero. This formal definition consists of two elements, deviation and time. Using these two elements, it is possible to distinguish between several types of stability. The time factor provides the distinction between short- and long-term stability. Although the statement ‘in the long run we will all be dead’ holds true, people might be interested to know what is going to happen in the short run. The factor deviation provides the distinction between stability and equilibrium, and the distinction between a deviation which is ‘near’ to a state of equilibrium, and one which is ‘far from’ it. An equilibrium is not necessarily stable when disturbed. A ball on the top of a hill will roll down the hill when touched. Moreover, it makes a difference to what extent an equilibrium is disturbed. The system might be stable for small deviations but unstable for large ones. In other words, the system can be stable for a limited range of deviations. It is this kind of distinction which Laszlo (1986) used in pointing out the difference between stability and ‘resilience’, the ability to absorb change and disturbance within a wide range of large and unexpected fluctuations.

The innovative aspect of autopoiesis, given the tradition in the thinking about maintenance, continuity, stability and so forth, is that it models the self-maintenance of the organization of the system rather than the variables, which are the elements of the system. The autopoietic model pictures a system that generates and regenerates its organization. The model shows that a certain structural configuration is generated and that once it exists, notwithstanding perturbations by the rise or decay of elements, the configuration is maintained.

It is not the usual model of stability of system variables, but a model of organizational stability. It is a stability model at the next higher (meta) level of the organization of the system.

Organizational Closure

A serious limitation of the original formal model of autopoiesis (Varela et al. 1974) is its notion of closedness, albeit closure on a higher meta-level of organization. The expose on system stability given above made it clear that stability should not be confused with closedness of a system. It is the very existence of environmental disturbances to the system, that is, the openness of the system, which gives rise to discussions of stability. Yet the model of autopoiesis unfortunately suggested that organizational stability had something to do with closure.

A closed system has no inputs from, and outputs to, its environment. It does not interact with the environment. Indeed the original formal autopoietic model lacked inputs and outputs. All elements needed to generate and reproduce the system existed within the system.

Later clarifications and interpretations of autopoiesis and its organizational closure have loosened this restriction. Varela’s (1981) explanation of organizational closure no longer suggests that strict closeness. Jantsch (1981) goes much further and considers the exchange of matter and energy with the environment — the openness of the autopoietic system – as crucial. His interpretation of self-organizing in terms of dissipative structures is that of the organization of a dynamic order in the midst of fluctuations in far-from-equilibrium conditions. Luhmann (1986) stated that the main contribution of autopoiesis is for it to be understood as the recursively closed organization of an open system. It does not return to the old notion of ‘closed’ versus ‘open’ systems. It links closure and openness. In Luhmann’s view, the theory on self-referential systems represents a step forward from the old system’s theoretical distinction between ‘open’ and ‘closed’; self-referential closure is only possible in an environment where it produces openness (Luhmann 1984).

Indeed, in social reality, closed systems do not exist. Most social systems are, by definition, open. In the political system, a ministry cannot isolate itself from its environment: parliament, interest and pressure groups, social institutions, citizens, etc. This also holds true for the organization of a ministry. The organization of a ministry is a reflection of the needs, interests and institutions in a particular policy area. Strict organizational closure is an unrealistic notion.

Structural Meta-Stability and Chaos

An autopoietic system is a model with organizational stability, that is, stability at the meta-level of the structure of the system. Whatever the perturbations are at the object-level of the elements of the system, at the meta-level, the system maintains its structure. In general broader terms, at the object-level the system can even be unstable, while at the meta-level, the organization of the system remains stable. Chaos can reign in an organizationally stable system.

To a certain extent, this can actually happen in biological reality. Some living organisms can be wounded or cut in parts, and will nevertheless recover. Elements of the organism can ‘die’ but the organism as a whole ‘survives’.

This idea of the coexistence of chaos and structural meta-stability opens interesting perspectives analogous to the intriguing eye-openers which Gleick (1988) vividly describes in his best-seller on ‘chaos’. Scientists usually study linear ‘normal’ systems, where small perturbations have predictable ‘linear’ effects that are controllable. Chaos is usually considered ‘weird’ as a category of non-solvable problems.

As long as a dynamic system can be modelled by linear differential equations, problems can be solved, but non-linear differential equations cannot usually be solved. Using many illustrations from the fields of meteorology, ecology, astronomy, etc., Gleick pointed out that there can be regular patterns in chaos; that systems which at first sight appear utterly chaotic, if studied in a different way, might reveal a certain order, or structure. Instead of studying relatively undisturbed ‘linear’ developments, one might obtain intriguing insights by studying the possible structure in utterly chaotic ‘nonlinear’ phenomena. Search for ‘stability’ at a meta-level of a system, which is seemingly ‘unstable’ (chaotic) at the object level.

Autopoiesis is an example of stability at the meta-level of the organization of the system. Let us take the parallel with ‘order out of chaos’ (Prigogine and Stengers 1985) one step further. One of the interesting discoveries in chaos theory is the ‘bifurcation diagram’ which describes a system which has a single steady state up to some parameter threshold, and possesses two steady states above that threshold. Order out of chaos does not necessarily have to be constrained to one single order. Chaotic systems can reveal different types of ‘order’, different ‘(meta)stabilities’. Translating this into terms of organizations and structure, creates an interesting extension of the idea of autopoiesis. In an autopoietic system, the system always generates and regenerates one and the same ‘organization’. A ‘bifurcation’-like system can generate and regenerate several different ‘organizations’ in the midst of perturbations and chaos. The multi-stable self (re)production by the system of various organizations avoids the negative connotation of the mono-stable ‘autopoiesis’ with ‘dynamic conservatism’ of government which always regenerates the bureaucratic organization and therefore perpetuates itself (this is discussed later on). The multi-stability conception combines the attractiveness of order out of chaos with the desirability of a pluralist order.

Autopoiesis and Administration Science

After the previous theoretical reconsideration of the model of autopoiesis and its possible implications, let us now end with some attempts to perform a more or less ‘literal’ transposition to the science of (public) administration. The first attempt turns out to have rather controversial implications in public administration. The second possible application points in another direction.

Dynamic Conservatism of Bureaucracy

One possible translation leads to an ultimate conservatism of bureaucracy, or more precisely, to dynamic conservatism. That is a system which possesses the valued features of flexibility, adaptiveness, changeability, but uses them only to preserve its own existence. Change of policy objectives, change of functions and instrumental change all preserve the existence of the institution. The continuance of the institution, the organization, is the ultimate or only goal. Whatever happens, be it policy change or budgetary cutbacks, the bureaucracy will never die (Kaufmann 1976).

In fact, the autopoietic model describes precisely such dynamic conservatism. A system generates and reproduces itself, particularly its organization. In translation, the autopoietic model would show government institutions to be continuously reproducing their organization, that is, the bureaucratic form of their organization. The application of the concept of ‘living’ to public administration would seem to suggest that the bureaucracy possesses perpetual existence. Therefore, it can be expected that the model will hold a strong attraction for those individuals with a dislike for, or suspicion of, bureaucracies.

The objections given above do not imply any criticism of the model of autopoiesis. If the choice is made to describe bureaucracy as a self-reproducing system, the model cannot be blamed for that characteristic. A model is merely a representation of some elements and interactions in reality, into a formal system of variables and relations. The model cannot be held responsible for the hard facts of reality.

The negative interpretation of autopoiesis seems related to the strict interpretation of the organizational closure of the system. If interaction with a turbulent outside world, the essence of self-maintenance, is added to the model, the interpretation becomes substantially more positive, putting it in line with Morgan’s (1986) interpretations. Note that Morgan also recognized the dangers of self-centredness of organizations. If the autopoietic model is extended to a multi-stability model of ‘order out of chaos’ (see discussion before), the objection of conservatism also vanishes.

The presumed perpetuity of government organizations has received attention in practical politics and in administrative science. Kaufmann’s analysis showed that in a period of some fifty years almost no government organizations had ceased to exist, which prompted him to state that they seemed to be immortal (Kaufmann 1976). In practical politics this presumed perpetuity led to the introduction of time-limit clauses in policies and regulations. Examples are ‘sunset laws’ which expire on a predetermined date. Research on sunset legislation showed, however, that the presupposed expirations often did not take place and that continuations of supposed temporary regulations were the rule and not the exception (De Leon 1978; Behn 1978). In an attempt to refine the thesis on immortality, Hogwood and Peters (1982) made a distinction between policy maintenance, termination, succession and innovation. Real policy innovation or termination does not take place as often as policy succession — a new policy, programme or organization for an existing problem. Apparently the ‘living’ of government organizations cannot be regarded as a binary property (either continuation or termination).

Another way of looking at ‘living’ organizations worth mentioning here is the study of births and deaths of organizations based on the notions of population ecology in biology (McKelvey and Aldrich 1983). Here too ‘living’ is considered a binary property: an organization is either alive or dead. In reality, gradual change in government organizations occurs much more often than the formation of entirely new organizations (birth) or their total abolition (death).

Autonomy and Self-governance

A second possible application of autopoiesis is as a model for the autonomy and self-governance which many societal systems appear to possess.

The idea that government is able to exert direct control over the course of affairs in societal sectors, the idea of government steering society from a powerful control position above and apart from the rest of society, has been shown to be unrealistic (Den Hoed et al. 1983). Government is not the almighty controller of social processes. The control capacity of government is limited for a number of reasons: lack of legitimacy, complexity of policy processes, complexity and multitude of institutions concerned, etc. Government is only one of many actors that influence the course of events in a societal system. Government does not have enough power to impose its will on other actors. Other social institutions are, to a great extent, autonomous. They are not controlled by any single superordinated actor, not even the government. They largely control themselves. Autonomy not only implies freedom, it also implies self-responsibility. Autonomous systems have a much larger degree of freedom of self-governance. Deregulation, government withdrawal and steering at a distance (Kickert 1993) are all notions of less direct government regulation and control, which lead to more autonomy and self-governance for social institutions.

The need for theoretical concepts on autonomy and self-governance is self-evident. The autopoiesis model can be considered the next step in the direction of a theory of system autonomy and self-control.

Control is crucial in cybernetics. Cybernetics was originally defined by Wiener (1948) as ‘the science of control and communications’. Regulation and control are its basic themes (Ashby 1956). All ‘hard science’ approaches to control assume the hierarchical notion of a controller who controls a controlled system which interacts with an environment. The controller receives information from the system and sends back control measures in return. Hierarchical control often does not exist in social systems. Several entities have influence, which at the same time influence each other, and possess a degree of autonomy. The controllability of the system is restricted; there is no unilateral control relation and no single controller above others. The subsystems are capable of self-control.

The notion of a network of several nearly autonomous actors with different and often conflicting interests, in which no single actor has enough power to dominate the others and in which decision-making is a bargaining process and decisions are compromises, is fundamentally different from the mono-rational model of governance by a single controller which adjusts the subsystems in order to reach a collective objective (Hanf and Scharpf 1978; Kickert 1980).

The contribution of autopoiesis is that it not only describes how a network of autonomous (autopoietic) systems behave, but also how it is able to maintain its organization, its identity. Organizational meta-stability is possible. Notice that, apart from ideological differences, this is also what intrigued Jessop (1990) about the notion of autopoiesis.

The ability of a network of autonomous systems to maintain itself and its basic order when confronted with disturbances, and its ability to survive in a turbulent and complex environment, is a vital quality in an administrative situation where direct top-down control by government is being steadily replaced by a self-governance autonomy of social institutions. It is reassuring to know that these institutions might be able to survive any hostile storms that they may encounter.


The central ideas of the autopoietic model of living systems can be used in a number of different ways to create concepts and models useful in the social and administrative sciences. Reasoning by analogy is a perfectly legitimate source of creativity. The various creative interpretations of autopoiesis reviewed in this article have led to remarkably interesting and relevant ideas. From the social science point of view, one can be reassured as to its usefulness. The question is whether or not these interpretations have much to do with the original model of autopoiesis.

In this paper, the focus of interest was on the applicability of the autopoiesis model to the administrative sciences. Applying a natural scientific model to a social science is hazardous. Full understanding of the original model is necessary if a correct conversion is to take place and caution when applying the model is recommended.

This paper included an extensive description of the original model of autopoiesis and a discussion of the implications of the model for the systems theoretical concepts of stability. After an exploration of the homelands of the model, an extensive review of the various interpretations of autopoiesis in different fields of the social sciences was presented. Finally an attempt was made to perform a careful translation and cautious application.

In the social sciences, the popularity of the autopoiesis model more or less derives from the publications of the renowned German sociologist N. Luhmann. In the administrative sciences autopoiesis became known via one of the metaphors about organizations used by G. Morgan. The usual relationship between organization and environment is fundamentally reversed in the autopoietic perspective. It is a pity that this perspective, which from an organization science point of view seems quite intriguing, has not attracted much attention in this field of science.

The overall conclusion is that the possibilities of a strict conversion of the autopoiesis model into a valid model that can be used in the administrative sciences are limited. The usefulness of the model does not seem to lie in strict adherence to the original and literal translation, but rather in its power as a source of creative lateral thinking. It can inspire highly interesting and relevant ideas. From the social and administration science point of view, it is not so relevant whether or not an interesting and useful idea is an accurate translation of some natural scientific mode. One final point is that a ‘hard science’ model cannot be blamed for not accurately fitting a ‘soft science’ world.


This is a revised version of a paper delivered at the conference on ‘steering-autopoiesis-configuration’ organized by the Department of Public Administration of the Erasmus University Rotterdam on the 21st of November 1990, published in (In’t Veld et al. 1991).


Ashby, W. R. 1956 Introduction to cybernetics. London: Chapman and Hall.

Behn, R. D. 1978 ‘How to terminate a public policy; a dozen hints for a would-be terminator’. Policy Analysis 4: 393-413.

Ben-Eli, M. U. 1981 ‘Self-organization, autopoiesis and evolution’ in Autopoiesis: a theory of living organization. M. Zeleny (ed.), 169-181. New York: North Holland Publishers.

Geyer, F., and J. van der Zouwen, editors 1986 Sociocybernetic paradoxes. London: Sage.

Gleick, J. 1988 Chaos, making a new science. Harmondsworth: Penguin.

Goguen, J. A., and F. J. Varela 1979 ‘Systems and distinctions, duality and complementarity’. International Journal of General Systems 5: 31-43.

Hanf, H., and F. Scharpf, editors 1978 Interorganizational policy-making; limits to central coordination and control. London: Sage.

Hoed, den P. et al. 1983 Planning as enterprise (in Dutch). The Hague: Scientific Council for Government Policy.

Hogwood, B. W., and B. G. Peters 1982 ‘The dynamics of policy change: policy succession’. Policy Sciences 14: 225-245.

Jantsch, E. 1981 ‘Autopoiesis: a central aspect of dissipative self-organization’ in Autopoiesis: a theory of living organization. M. Zeleny (ed.), 65-88. New York: North Holland Publishers.

Jessop, B. 1990 State Theory. Cambridge: Polity Press.

Kaufmann, H. 1976 Are government organizations immortal? Washington: Brookings Institution.

Kickert, W. J. M. 1980 Organization of decision-making. Amsterdam: North Holland Publishers.

Kickert, W. J. M. 1993 ‘Steering at a distance; a new paradigm of governance in Dutch higher education’. Forthcoming in Governance.

Laszlo, E. 1986 ‘Systems and societies: the basic cybernetics of social evolution’ in Sociocybernetic paradoxes. F. Geyer and J. van der Zouwen (eds.), 145-171. London: Sage.

Leon, de P. 1978 ‘Policy termination’. Policy Analysis 3: 369-393.

Luhmann, N. 1984 Soziale Systeme (in German). Frankfurt: Suhrkamp.

Luhmann, N. 1986 ‘The autopoiesis of social systems’ in Sociocybernetic paradoxes. F. Geyer and J. van der Zouwen (eds.), 172-192. London: Sage.

Maturana, H. R. 1975 ‘The organization of the living’. International Journal of Man-Machine Studies 7/3: 313-332.

Maturana, H. R. 1981 ‘Autopoiesis’ in Autopoiesis: a theory of living organization. M. Zeleny (ed.), 21-33. New York: North Holland Publishers.

Maturana, H. R., and F. J. Varela 1980 Autopoiesis and cognition: the realisation of the living. Boston: Reidel.

McKelvey, B., and H. Aldrich 1983 ‘Populations, natural selection and applied organizational science’. Administrative Science Quarterly 28: 101-128.

Morgan, G. 1986 Images of organization. London: Sage.

Prigogine, I., and I. Stengers 1985 Order out of chaos. New York: Bantam.

Teubner, G., editor 1988 Autopoietic law; a new approach to law and society. Berlin: de Gruyter.

Teubner, G. 1989 ‘How the law thinks: towards a constructivist epistemology of law’. Law and Society 23/5: 727-759.

Uribe, R. B. 1981 ‘Modeling autopoiesis’ in Autopoiesis: a theory of living organization. M. Zeleny (ed.), 51-62. New York: North Holland Publishers.

Varela, F. G. 1979 Principles of biological autonomy. New York: Elsevier.

Varela, F. G. 1981 ‘Describing the logic of living’ in Autopoiesis: a theory of living organization. M. Zeleny (ed.), 36-48. New York: North Holland Publishers.

Varela, F. G., H. R. Maturana, and R. Uribe 1974 ‘Autopoiesis: the organization of living systems, its characterization and a model’. Biosystems 5: 187-196.

Veld, in’t R. J., L. Schaap, C. Termeer, M. van Twist, editors 1991 Autopoiesis and configuration theory: new approaches to societal steering. Dordrecht: Kluwer.

Wiener, N. 1948 Cybernetics. New York: Wiley.

Zeleny, M. 1977 ‘Self-organization of living systems: a formal model of autopoiesis’. International Journal of General Systems 4: 13-28.

Zeleny, M., editor 1981 Autopoiesis: a theory of living organization. New York: North Holland Publishers.

Walter J. M. Kickert

Department of Public Administration, Erasmus University, Rotterdam, The Netherlands

@ @

COPYRIGHT 1993 Sage Publications, Inc.

COPYRIGHT 2004 Gale Group