Chapter 2 - Systems Concepts

As any poet knows, a system is a way of looking at the world. The system is a point of view - natural for a poet, yet terrifying for a scientist! (Weinberg 1975, p 52)

The intent of this chapter is to introduce and discuss a variety of key concepts that I believe are useful for understanding the complexities of natural and social interactions relevant to sustainability. As noted in the introduction, some of these concepts are characterized by a lack of consensus on definition, by variation among theoretical explanations, and by variable weighting applied to different aspects. In this case, discrepancies arise from common vs. 'scientific' definitions, and others from the pre-paradigmatic nature of the studies. Broad consensus has not yet been reached with respect to many complex self-organizing systems concepts. To help clarify this, I begin by outlining the key concepts, and some relations among them. I consider variations among the definitions relevant to this discussion, although these do not necessarily represent full consideration of the variations expressed or possible.

The motivating concern for developing the systems concepts discussed in this thesis was recognition that we draw boundaries around systems, as a means of both ecological and political identification, even when systems lack spatial and temporal boundaries. Rather than force identities on such systems to make analysis, understanding, and planning easier, I advocate developing new concepts that force us to recognize the complexities, contradictions, and uncertainties that exist. As a starting point, I advocate relinquishing boundaries. The discussion in this chapter, is a move toward reconceptualizing systems on a different basis.

In the first section, I briefly discuss the concept of system, emphasizing a definition based on relations rather than on boundary delineation.
In the second section, I introduce and discuss the distinction between autopoietic and sympoietic systems that forms the central theme of this thesis. I begin with the concept of autopoiesis (self-production) introduced by Maturana and Varela (1980, also Varela et al. 1974). Their precise definition has minimized debate on definitional issues, however, application of the concept is a contentious issue. This is primarily noted in the following chapter on human systems, the subject area where it receives the most debate (Fleishaker 1992, Mingers 1995).
I consider autopoiesis to be a more specific form of self-organization, referring to systems with the capacity for producing as well as organizing the components essential for their own continuation. Living systems such as cells and organisms provide examples.
As a contrast, and due to my concerns that we attempt to bound that which has no boundaries, I introduce and discuss sympoietic (collectively-producing) systems (Dempster 1995). These are self-producing systems that do not have clearly defined spatial and temporal boundaries. Ecosystems provide a good example.
Discussion in this section is focused on identifying and explaining the distinction between the two types of poietic systems - a distinction that forms the basis for the critique of planning. Of particular importance to the discussion are two concepts defined by Maturana and Varela:

Organizational closure - refers to the degree of self-containment a system has with respect to its organization and is consequently critical regarding informational issues.

Structural Coupling - refers to the structural relationship a system has with its environment and is a concept critical to discussing interdependence among systems.
Differences between the systems with respect to these characteristics lead to numerous other differing properties, behaviours, and characteristics that are important for understanding.
In the third section I discuss the concept of self-organization, which I introduce as a means for understanding complex systems and for identifying systems without boundaries. While a key concept for the discussion, it is subject to considerable confusion and debate. Since there is a lack of consensus on meaning, I again provide a continuum for definition. The general notion is that the generation and continuation of self-organizing systems involves internally, rather than externally, governed processes and distinctions. The notion proves useful for conceptualizing sympoietic systems. Since these systems cannot be defined by their boundaries, they must be defined by their self-organizing characteristics. Four factors involved in the process are highlighted and discussed: global-local influences, feedback and recursion, points of tension, and complexity characteristics.
In the final section, I consider the implications of the concepts, including a description of the evolution of complexity, and of the possibilities for future causality. Because of my stated interest in sustainability, I consider the implications of the system characteristics as they relate to the attributes and elements of sustainability discussed in Section 1.2.2 above.

It is important to recognize that the following descriptions represent the "ideal" system characteristics - the lenses through which we can look at and understand the world. As with any other categorization, they represent the ends of a continuum. Any real system will sit in the vast grey area in between. The distinction between autopoiesis and sympoiesis, however, is useful for drawing attention to particular aspects of systems characteristics that I believe have been neglected.

2.1 Systems

As noted in the introduction, system is a term that is open to considerable variation among definitions. Figure 2.1 illustrates one possible definitional continuum, drawing attention to the way in which systems are conceived: as bounded entities or as connections among elements.

Although a focus on components and boundaries is typical when identifying or understanding systems, the relations among these components are critical for establishing a system, its behaviour, and its degree of complexity. Consider three balls connected by rods (Figure 2.2). Arranged as a triangle they will be very stable. As a pendulum connected to a pendulum - same components, different relations - the system will have unpredictable behaviour.*1 Organisms provide another example. It is not the molecules contained in an individual plant or human that make such systems complex, but the set of relations holding the molecules together. I, therefore, define a system as a set of relations that describes some sense of connectedness.

A key question to ask, then, is: How are these relations established? What creates the pattern of organization that describes any particular system? These questions are especially relevant for the large complex natural and social systems that are the concern and subject of this thesis. The purpose of this chapter is to consider these questions by focusing on concepts relating to complex self-producing and self-organizing systems.

2.2 Autopoiesis and Sympoiesis*2

To establish what they believed to be the essential quality that differentiates living systems from non-living systems, Maturana and Varela (1980 and Varela et al. 1974) pointed to the self-producing capacity of living systems. (See Box 2.1.) Such systems continuously and recurringly produce relations among their components in a manner that allows them to continually reproduce the same pattern of relations. Consider a living system at its most basic: a bundle of complex molecules arranged by a complicated set of relations. These systems use energy to organize physical matter into particular arrangements. The arrangements, in turn, are capable of producing the necessary components arranged in the necessary pattern to allow for their own continuation.

Maturana and Varela chose the term autopoiesis, from the Greek words for self and production, to emphasize their representation as a new, different, and specific concept. A key aspect of autopoietic systems' self-producing ability is the production of their own boundaries through "preferential neighbourhood interactions" (Maturana and Varela 1980). The phrase is used to emphasize that it is the system's interactions that form the boundary, not external forces.

Maturana and Varela provide a precise definition:*3

A dynamic system that is defined as a composite unity as a network of productions of components that, a) through their interactions recursively regenerate the network of productions that produced them, and b) realize this network as a unity in the space in which they exist by constituting and specifying its boundaries as surfaces of cleavage from the background through their preferential interactions within the network, is an autopoietic system.

This organization that defines an autopoietic system as a composite unity is the autopoietic organization, and we claim that an autopoietic system in the physical space, that is, an autopoietic system realized as a composite unity by components that define the physical space by satisfying the thermodynamic requirements of physical phenomena, is a living system. (Maturana 1980, p 29)

They also list a set of criteria for defining autopoietic systems which are listed in Box 2.2. Due to their self-defined boundaries and self-referential nature, autopoietic systems are autonomous units, separated from their environment on an organizational level. As will be explained further below, this does not mean that the systems are totally independent of their environment. They are only autonomous in the sense of being self-governing, not being independent.

Many complex living systems, however, do not match these characteristics - especially regarding boundary production. Typical ecosystem definitions, for example, note the fuzzy nature of ecosystem boundaries in both spatial and temporal dimensions (e.g. Golley 1993, Noss 1995, Agee 1996).

The term sympoiesis, from the Greek words for collective and production, was chosen to describe systems which, in contrast to autopoietic systems, are characterized by cooperative, amorphous qualities (Dempster 1995). These systems recurringly produce a self-similar pattern of relations through continued complex interactions among their many different components. My aim in characterizing sympoiesis is to move away from the necessity of delineating boundaries. Interactions between components and the self-organizing capabilities of a system are recognized as the constitutive and defining qualities. 'System-hood' does not depend on production of boundaries, but on the continuing complex relations among components. The concept emphasizes linkages, feedback, cooperation, and synergistic behaviour rather than boundaries.

While I have previously described sympoietic systems as self-producing (Dempster 1995, 1996, 1997), they are better understood as positioned on a continuum between self-organizing systems and autopoietic systems (Figure 2.3). While they are self-producing to some degree, without clearly defined boundaries and with the continual input of new information, they cannot be described as self-producing in the same sense as autopoietic systems. I will use the term poietic to refer to both systems and to their continuing self-similar production.

There is some difficulty in using the term self-producing, even for autopoietic systems. Although the systems produce the components essential for their own continuation, they do not produce the components of those components. This is a slippery slope argument, however, since ever smaller constituent parts appear to exist. Since the term autopoiesis is somewhat entrenched, I will continue using it, although "self-constructing" (see Kay 1984) may be more accurate. Poiesis, then, will be used in reference to the continual production of a self-similar set of relations, more similar in the case of autopoietic systems than in sympoietic systems.

To gain an understanding of the two system types and the significance of their differences, I will compare descriptions of some of the key characteristics that are listed in Table 2.1.

Table 2.1 - Comparison of Poietic System Characteristics
AUTOPOIETIC SYSTEMS	SYMPOIETIC SYSTEMS
Defining Characteristics
self-produced boundaries	lacking boundaries
organizationally closed	organizationally ajar
external structural coupling	internal and external structural coupling
Characteristic Tendencies
autonomous units	complex, amorphous entities
central control	distributed control
'packaged,' same information	distributed, different information
reproduction by copy	amorphous reproduction
evolution between systems	evolution within system
growth/development oriented	evolutionary orientation
homeostatic balance	balance by dynamic tension
steady state	potentially dramatic, surprising change
finite temporal trajectories	potentially infinite temporal trajectories
predictable	unpredictable
Examples
cells, organisms	ecosystems, cultural systems

Before considering these characteristics, however, some specific definitions applied by Maturana and Varela (1980 and Varela et al. 1974, Maturana 1980) are important for the discussion.*4

The pattern of organization*5 of a system is the relations among components that define a system as a specific type of system. The pattern of organization of a tree, for example, is the relationship between the leaves, trunk, roots, and other components. If the pattern of organization is changed, it becomes a different type of system, such as moss (no trunk) or an elephant (no leaves).
The structure*6 of a system is the actual relations and components that constitute a particular system in a particular domain. A tree, for example, exists in the physical domain, so its structure will be the actual physical arrangement of the components that make it a particular tree. A spruce has a different structure than a maple.*7
Any pattern of organization can be manifest in many different structures.

Distinguishing between these two aspects of a system allows definition of the following concepts:

Organizational closure refers to the degree of self-containment a system has with respect to its pattern of organization. A system can be organizationally closed, open or ajar.*8 A system which organizes the relations among its components in such a manner as to ensure the continuation of its own pattern of organization is organizationally closed. Consider a tree which has a specific pattern of organization that governs the receipt of energy and nutrients to perpetuate its structure and subsequently its pattern of organization as a tree. Although the system receives external inputs, it maintains a pattern appropriate for its own development and continuation.
A system which relies completely on external sources for determining its pattern of organization, such as a human made artifact, is organizationally open.
A system which relies on external sources, yet limits these inputs in a self-determined manner, is organizationally ajar. An ecosystem which limits the introduction of new species is an example of the latter.

Structural coupling refers to the recognition that a system's structural relationship with its environment will determine its responses to disturbances or triggers. For example, a house plant moved to a different location will not survive if it no longer receives the structural inputs (energy and material) that it requires even though its pattern of organization has not changed. Allergic reactions also illustrate the concept. All people have the same pattern of organization, yet each individual has a specific structure. An individual's response to peanuts, for example, will vary according to their structure.*9 Continued system survival depends on having a structure which can respond suitably to the system's environment.

Poiesis refers to the ability of a system to continually and recursively produce its own set of relations.*10 A particular pattern of organization will produce a particular structure. To be sustaining, this structure must in turn produce a pattern of organization that will ensure continued production of a continually suitable structure. This recurring cyclic production does not mean that the structure and organization cannot change. Figure 2.4 illustrates this recursive process noting the need for structural inputs, structural coupling, and the differences between autopoietic and sympoietic systems.

In Section 1.3.3 above, I have noted three essential elements of sustainability: maintainability, adaptability and environmental change. The last two systems aspects, structural coupling and poiesis, are the most obviously relevant to the possibility of sustainability. As I will illustrate, however, the degree of organizational closure is critical to the trade-off between maintainability and adaptability and a system's consequent ability to persist despite environmental change. It is here that the distinction between the two types of systems becomes most relevant.

2.2.1 Comparison of Poietic Systems

The characterization of autopoiesis described here is based on the literature. However, it is essential to recognize that my description of the characteristics primarily results from making the distinction between autopoietic and sympoietic systems. Because of this, the descriptions of autopoietic systems will likely diverge from those offered by others. Any errors, contradictions, or misinterpretations of autopoietic system characteristics presented here are obviously my own. Key sources include: Maturana and Varela (1980), Maturana (1980), Varela (1981), Bednarz (1988), Fleischaker (1992), and Mingers (1995).

The characterization of sympoiesis arises from making the distinction between autopoietic and sympoietic systems. However, broader understanding of such systems has been generated through theoretical and observational discussions regarding complex and self-organizing systems from physical to social systems. Key sources include innumerable authors, especially those cited in the discussion of complex self-organizing systems (Section 2.3 below). In addition, insight has been gained from authors who articulate similar concepts, a few of which are noted in Section 2.4.3 and Box 2.8.

Differences among the organizational and structural aspects in the two system types lead to different characteristics. Although I discuss them under separate headings, these characteristics are intricately interconnected. In some sense the characteristics are causally connected; any particular characteristic arises because of the others. This is the first paradox of poietic systems; due to their recursive nature, no particular characteristic can be described as the cause.

Boundaries

I begin with boundaries since this issue was my point of entry.*11 I found autopoiesis - primarily the notion of self-production - to be a useful concept for understanding living systems. For application to ecosystems, however, the emphasis on self-produced boundaries ran counter to my belief that ecosystems had to be recognized as unbounded systems. Integration of the concepts and concerns led to definition of sympoietic systems.

I do not argue against recognition that useful, non-arbitrary boundaries for an ecosystem can be drawn by an observer using particular criteria at a particular scale. For the same observer, however, boundaries are effectively pre-drawn for autopoietic systems because the system, by producing its own boundaries, defines the relevant scale of observation. Since it autonomously organizes itself, anything less than the whole is obviously a part, and anything greater, is obviously 'environment.' Such clarity is not possible with sympoietic systems. This does not mean that autopoietic boundaries are clear and distinct when approached. As noted with categorizations in general, the closer one gets to any boundary, the less distinct containment becomes. For example, at an atomic level, organism boundaries are less substantial. The question of when food ingested by an organism becomes part of the organism also involves a fuzzy boundary. I maintain, however, that definition of boundaries for autopoietic and sympoietic systems is fundamentally different. I qualify this statement, however, by re-emphasizing that my description reflects 'ideals' and most systems will exist somewhere in between. In addition, I note that the boundary issue does not imply that autopoietic entities can (or should) only be studied as whole systems. Anything less or more, however, must contend with the system's self-defined boundaries.

Arbitrary distinctions do arise in discussing autopoietic systems, however, when defining a systems' pattern of organization. Earlier, I noted that particular patterns of organization can be manifest in different structures, using trees - spruce and maple - as examples. To illustrate the subjective nature of the classification, however, I note that it would be equally correct to name the latter two as patterns of organization, or to name all living systems as having the same pattern of organization. Either way, however, individual entities are manifest structurally. For autopoietic systems it is on this level, that boundaries, and consequently system identity, are defined. Although we cannot escape, completely, the arbitrary nature of naming or categorization, the advantage of the concept of poietic systems is that it emphasizes different aspects of naming and categorization.

Information

Due to their self-produced boundaries, autopoietic systems have control over system inputs and outputs. The systems, therefore, can maintain organizational closure by restricting undesirable information while keeping information essential for their continued production. It is important to note that on this point I diverge in an important manner from presentation of the concept by Maturana and Varela. They claim that since information is an extrinsically defined quality (see Box 2.3) it must be irrelevant to self-producing systems.*12 While recognizing the validity of their argument, I believe that without inclusion of information, autopoiesis is an impoverished concept. There is no possibility for following system history and recognizing the increased complexity of system structure and pattern of organization. The essential difference between an amoebae and a human relates to their information content. As I apply the concept of information then, I emphasize that I do so because it has heuristic value. It must be recognized simply as a tool for describing developmental and, especially, evolutionary change in successive autopoietic systems. "Nature is not about codes: we observers invent the codes in order to codify what nature is about" (Beer 1980, p 69). To the system, information is just part of the pattern of organization or structure.

Autopoietic systems benefit from boundaries and organizational closure by gaining the ability to build up complex information through recurring successful interactions between structure and environment. The resulting information tends to be carried centrally and transferred as a 'package.' Genes provide the most obvious example. Since available information is circumscribed by self-containment, however, autopoietic systems have restricted adaptive potential and a limited capacity for coping with uncertain and changing circumstances. The example of a house plant noted above illustrates this point. Recognition that trees have provenance - geographically specific genetic histories (e.g. Smith 1986, Kimmins 1987) - is another indication.

Lacking self-defined boundaries, sympoietic systems consequently lack the same degree of control and are open to a continual flux of information. I refer to the systems as organizationally ajar, since they are not totally open. Sympoietic systems regulate information input through internal structural coupling: information must be contained in a suitable structure in order to be integrated into the system. As an example, consider the incorporation of information into an ecosystem through the introduction of exotic species. Only those species with structures suited to the ecosystem will survive. For example, a species suited to a dry environment, such as a cactus, will not survive in a wet environment since it lacks the essential structural adaptations. In consequence, the information carried in the cactus will not be incorporated into the wet ecosystem. This dynamic, though restricted, flux of information allows the systems to evolve continuously by adapting to changing conditions and by generating new ones. Historical and successional change in ecosystems are examples.

These factors indicate one of the critical differences between autopoietic and sympoietic systems. The latter systems carry different bits of information distributed among their components and subsequently have no centralized control. This factor realizes their distinctive character as amorphous, cooperative, self-organizing entities. An example illustrating the importance of information storage in an ecosystem is the influence of seeds stored in the substrate as a determining factor in the shifting mosaic of a mixed-wood forest (e.g. Mladenoff et al. 1993) and in the vegetation composition of fresh water marshes (Keddy and Reznicek 1985, Parks Canada 1991).

This means sympoietic systems depend on the information contained in their components, which are typically autopoietic systems. Sympoietic systems can build their complexity by incorporating complex components. Although sympoietic systems are also restricted by their information content, they have the advantage of also being open to new organizational information.

Reproduction The differences regarding boundaries and information content illustrate another significant difference between the two systems types - reproduction. In this case, the process is different enough that the same term should not be applied to both. I suggest for sympoietic systems the term is effectively irrelevant since they are characterized by continuing poiesis: recursive production of structure and pattern of organization. Rather than reproducing - they are just continually-producing. This is not necessarily equivalent to growth of the system, but refers to continual replacement, and possible alteration, of components and relations.
Poiesis also occurs in autopoietic systems, yet these systems produce in a different sense as well. They produce offspring, something other than themselves. This emphasizes a point made by Krippendorf (1987), that reproduction is actually an allopoietic (other-producing) process. Although contrary to Maturana and Varela's comment that reproduction is a "moment in autopoiesis" (1980, p 101), I find this notion resolved a difficulty. Describing change in poietic systems, I have previously noted that for autopoietic systems change occurs between generations (Dempster 1996, 1997). At the same time, I have struggled with understanding attainment of highly complex information given the reiterative and restricted self-production as described by Maturana and Varela. Recognizing the allopoietic nature of reproduction provides the necessary loop-hole. Although I refrain from entering discussion of the implications here, I note three other factors relevant to understanding the build-up of information. First is the important combinatory power of sexual reproduction. Second, is the distinction Gunther and Folke (1993) make between latent (stored) and working (expressed) information. Third, is what Kelly (1994, p 39) terms a "law of god: When something works, don't mess with it; build on top of it. In natural systems, improvements are 'pasted' over an existing debugged system." This parallels Simon's (e.g. 1996) analogy of the watchmaker. Consider the implications of an interruption for a watchmaker who must construct a watch in 100 separate successive steps vs. a watchmaker who can do 10 sets of 10 steps. Through organizational closure, autopoietic systems are able to combine and store effective bits of information, building up a complex pattern of organization.

History Understanding the role of system history is crucial (e.g. Prigogine and Stengers 1984, Allen and Hoekstra 1992). This is particularly true in systems carrying complex information since the latter is a result of both developmental and evolutionary system history. Autopoietic systems, for example, develop predictable responses to predictable environmental conditions, passing on the accumulated information to subsequent generations. Due to structural coupling, the resulting information provides the potential for producing a system structure that is relevant, yet restricted to, particular environments. As noted below, these potentials and restrictions are important for considering system predictability.
Although sympoietic systems rely on a flux of new information and are continually evolving, this does not cancel the importance of system history. The acceptance of new information depends on a system's structure. This will be a result of past interactions among component autopoietic systems, consequently incorporating their history as well. In addition, historical consideration of the self-organizing factors involved in system generation will often be critical.

Trajectories

The differences in the preceding characteristics mean the two systems types have different types of temporal trajectories. Autopoietic systems have a growth/developmental focus; sympoietic systems, an evolutionary focus. By this I mean that autopoietic systems follow some sort of path from a less to more developed stage, whereas sympoietic systems are continually, though not necessarily consistently, changing. There is no particular sense that the latter systems will reach a 'higher' or 'more mature' level of development or organization. (See Box 2.4 regarding definitional issues.)

In autopoietic systems, birth, youth, maturity, and death are much more clearly defined concepts. The key distinctions are two-fold. First - regarding the development notion - autopoietic systems are limited to a single trajectory: caterpillars turn into butterflies, not trees or elephants or even similar butterflies. There is even a reasonable degree of certainty as to when such changes will occur.

In sympoietic systems, there is uncertainty regarding both when and into what a system will change. Their trajectories have the possibility for dramatic and surprising change. The shifting mosaics of mixed-wood forests can manifest quite different species distributions in the same area over the long-term. The "dust-bowl" of the southeastern US, "weed species" takeover preventing reinstatement of the "forest" after a clearcut, and a flip between benthic and pelagic species dominance (Regier and Kay 1996) are other examples of unpredictable trajectories in sympoietic systems.

Second, autopoietic systems have finite temporal trajectories. Among those discussing autopoiesis, Zeleny (1977) and Bednarz (1988) are rare, by pointing out that death is also a fundamental characteristic. Many authors ignore this key point. The systems have defined temporal boundaries. The significantly different characteristics of sympoietic systems give them potentially infinite trajectories, making definition of 'beginnings' and 'ends' - their temporal boundaries - problematic.

The organism metaphor - an autopoietic metaphor - has, however, enabled and reinforced an interpretation of ecosystems as progressing through various stages of development. Admittedly, such an interpretation is more useful in some ecosystems, or at some scales, than others. For example, the metaphor is somewhat applicable regarding succession in ecosystems dominated by a disturbance regime such as pests in forests or fire in grasslands. In both of these cases conceptualizing some sort of life/death cycle or reproduction seems relevant for understanding the systems as a whole. The same metaphor is less useful, however, for conceptualizing tropical rainforests or pelagic communities. These systems seem to correlate more strongly with sympoietic characteristics. These examples illustrate the need to emphasize these descriptions as ideals - the 'ends' of the continuum. 'Real' systems exist at various middling positions.

Equilibrium-Balance Equilibrium is another term which has many uses and implications that must be clarified. In this case, I choose to apply the term balance instead, to refrain from generating confusion. See Box 2.5 regarding definitional issues.
Continued autopoiesis requires successful structural coupling to particular environments, with continued predictable structural inputs. In consequence, autopoietic systems are geared toward self maintenance; toward maintaining a structure that is suited to that particular environment. These systems are centered on maintaining or returning to a particular identity or internal condition. Trees, for example, need to maintain a particular leaf area in order to satisfy a balance between loss of water and receipt of solar radiation. The systems are homeostatic and are primarily controlled by negative feedback. Warm-blooded animals provide an obvious example. Such systems have a specific range of viable internal body temperatures. Since environmental conditions change they have developed mechanisms for maintaining internal temperature: sweat or pant if it is too hot, shiver or put on a sweater if it is too cold. A key point, however, is that these responses have developed to cope within a relatively predictable range of environmental changes such as seasonal or diurnal changes. System responses may be inadequate if change beyond the 'normal' range is experienced. Even ectotherms must maintain body temperature within a particular range and will stay in, or move to, habitats that allow them to satisfy these requirements.
Sympoietic systems carry a variant of 'stability,' but it must be distinguished from that of autopoietic systems. Lacking centralized control, their balance is maintained by dynamic tension. The mechanism for holding the position is not homeostasis, but self-organization, which generates an evolutionary dynamic balancing. I will return to discuss this balancing after explaining the process of self-organization for it is this process that provides the degree of stability and predictability that these systems have.

Figure 2.5 illustrates the difference between the balance exhibited by the two types of system in contrast to the standard representation for equilibrium in systems. Box 2.6 provides further discussion on this point.

Sustainability

Considering the different elements of sustainability noted in Section 1.3.3, the system characteristics just described can usefully be discussed regarding their contribution to the balance among these elements.

Maintainability - Adaptability Adaptation requires a stable environment (J. Theberge 1996, pers. comm.), but changing environments require adaptability. Since environments are continually changing and uncertain, autopoietic systems, restricted by their information content, may eventually reach a point at which adaptation is no longer possible. This is true in both a developmental and an evolutionary sense. Rather than acting as an impediment for sympoietic systems, however, uncertainty provides a critical advantage: it is the very source of opportunities for adapting to change.*13 The characteristics described suggest that autopoietic systems are weighted toward maintainability, sympoietic systems toward adaptability.

Predictability Since autopoietic systems are restricted in the manners described, they are relatively predictable through observation of ancestral autopoietic systems. Their characteristics emphasize the degree to which future trajectories are entrenched, especially by organizational closure.
Prediction is also often assumed to be possible for sympoietic systems, yet there is need for caution in making such an assumption. Discussion below will indicate that part of our current difficulty regarding the achievement of sustainable systems has been a result of inaccurate interpretation: mis-taking sympoietic systems as the more predictable autopoietic systems. Since sympoietic systems are organizationally ajar, the interactions generating them can change as a result of new organizational information. It is still important, however, to consider the degree to which the trajectory of these systems is entrenched. There will be some restrictions on possible future systems states, due to restrictions on information input and due to restrictions on their component autopoietic systems. Although the systems are inherently unpredictable, they are not necessarily completely unpredictable.
Since sympoietic systems do demonstrate some degree of stability or continuity, there is a strong desire to believe in the possibility of forecasting the changes that will occur, or at least knowing which ones will be predictable. To consider at what level this may or may not be possible and to describe how sympoietic systems may be identified without recourse to arbitrarily defined boundaries, I turn to a discussion of self-organization.

2.3 Self-Organization and Complexity

Self-organizing processes have generated systems of increasing complexity throughout the history of the universe, including the emergence of stars, planets, complex molecules, life, consciousness, and culture (Kay 1984, Eriksson 1986, Iberall and Soodak 1987, Swenson 1989) (Figure 2.6). Each new level of system has significantly different properties - termed emergent properties - than the component systems they are created from (Eriksson 1986, Swenson 1989, Robb 1992). The new systems are "wholes that are greater than the sum of their parts."

In each case, random elements (i.e. components) are constrained into coherent patterns or arrangements (i.e. systems). For example, rather than atoms being scattered uniformly throughout the universe, they coalesce into stars and galaxies; raindrops are constrained into river systems; complex molecules are constrained into cell components. As emphasized by the opening comment of this chapter stressing the importance of relations when defining systems, I emphasize that the self-organizing factors place constraints on relations, sometimes by creating new ones such as illustrated in Figure 2.2 between pendulum and triangle. Since each new level is generated from successively more complex systems, each incorporates new degrees of complexity. The question of interest, then, is: What are the factors acting in these systems that make them appear to organize themselves?

The approach taken here is to try to understand these constitutive factors, extending concepts learned from observing simpler systems, to help understand more complex systems. Due to the significant increases in complexity, considerable caution must be applied when using an understanding of lower level systems to explain higher level systems (Boulding 1957). As argued in the introduction, however, the primary intent is to generate new understanding by using a different perspective than that which is commonly used. It is not intended as a replacement for either the reductionist or holistic approach, nor to other integrative approaches. It is presented as a complement which may provide some new insight.

2.3.1 Existing Literature

In the following chapter, I use Kuhn's (1970) interpretation of science as an example of social systems. His description of pre-paradigm periods that characterize immature sciences reflects the current state of confusion regarding self-organizing systems theory. As noted in the introduction to this chapter, there is a considerable variety of definition and explanation regarding the concept and process of self-organization. My intent here is not to sort out these discrepancies, nor to provide a definitive review or critique of the concepts and explanations. My intent is to draw out those factors that seem to be commonly expressed (although occasionally with different terms) that seem to be relevant for understanding sympoietic systems. In consequence, the following discussion is not purported to be comprehensive. As noted in the methods section, I have relied on secondary sources regarding many aspects relevant to this thesis including the systems concepts that are a subject of this section. I consequently point out that the citations refer to the sources from which I obtained the information, and do not necessarily represent the originators of the ideas. Although these are often cited in the secondary sources, in order to prevent mistaken attributions I do not include such references unless such attributions are clear or common. Future research will involve more comprehensive literature review. Box 2.7 lists key authors who have contributed to development of the concepts discussed in this section. I note classic works, excluding some of the more technical ones, and also note some key books written for the lay reader.

The concept of self-organization arose in systems discussions, shortly after the birth of general systems theory. It has been gaining ever increasing popularity as it becomes recognized as a pervasive phenomenon, explanatory of many systems and their behaviours. The term is used, however, in reference to many types of self-organization, from relatively simple physical and chemical systems to the complexities of social and cultural systems. As Andrew (1989, p 24) notes, "it is difficult, probably impossible, to find a precise definition of what is understood by a self-organizing system. Nevertheless it is an important and useful idea. One of the reasons definition is difficult is that a given system may seem to be self-organizing, or not, depending on how it is described." Dalenoort (1989a, p 300) adds that a person with knowledge of a system may be able to predict outcomes, yet the system may seem self-organizing to someone without such knowledge.

On an intuitive level, self-organization refers to exactly what is suggested: systems that appear to organize themselves without external direction, manipulation, or control (Jantsch 1980, Ho and Saunders 1986, Salthe 1989). As will be shown by explanation of the process below, this definition is problematic - Andrew and Dalenoort make key points. In essence, self-organization just refers to the need and/or possibility for a new description of a system due to the presence of a new set of relations defining the system. Since this need arises in many different situations, confusion emerges.

Several authors, who seek to distinguish between different types of self-organization, offer limited typologies. However, I have yet to find a comprehensive description integrating the various types - especially not one that is understandable! For reference, some of the distinctions and typologies made by others are listed briefly in Box 2.8. Although some of the distinctions can be recognized as nested concepts representing more specific cases of self-organization, others overlap. In a sense, different authors isolate a particular factor and then categorize all other systems as not-that-factor. Especially in more complex systems, then, the different typologies illustrate different perspectives on the same system. In complex systems such as ecosystems and social systems all of the factors are involved.

In this section I explain the basic process of self-organization, linking the explanations to the different types identified in Box 2.8. Specific ideas and examples are referenced; however, much of the explanation is general. References used for this section include Jantsch (1980), Prigogine and Stengers (1984), Cramer (1993), Schneider and Kay (1994), and a variety of papers in Kilmister (1986b), Dalenoort (1989b), and Yates (1987).

2.3.2 Process of Self-Organization

Several key factors that play a role in the process of self-organization are highlighted in this thesis. In particular, I place emphasis on:

interaction of global-local influences,
points of dynamic tension,
feedback and recursion,
and complexity.

Non-equilibrium thermodynamic conditions are generally taken to be a distinguishing characteristic of self-organizing systems (e.g. Kay 1984, Nicolis and Prigogine 1989). Since I will be transferring the concept of self-organization into the socio-cultural domain, however, I will not emphasize such conditions. By defining social system components and relations as socially constructed elements - social roles, policies, cultural norms - rather than their biophysical counterparts, thermodynamics is a less useful consideration. Although it is possible to find a metaphor for energy in the socio-cultural domain, I argue that self-organizing processes are actual constitutive factors in social systems, not just metaphors. Therefore, I refrain from emphasizing thermodynamics, although it is essential to recognize that thermodynamics is fundamental to all physical and biological systems.

Global-Local Influences

Macroscopic influences can have significant effects on microscopic entities. Consider the influence of gravity on a river, a skier, or a small speck of dust. Such an influence, although often 'far-away,' imposes a direction; a simple rule. Any entity under the influence of such a direction will eventually encounter some kind of blockage or constraint. When the direction is global and continuous in nature and the constraints are maintained, the resulting interaction will generate pattern or structure. For example, river systems are created by a simple rule, 'water flows downhill,' and the constraints provided by landscape features from soil particles to pre-Cambrian shields (Figure 2.7).

Global influences are typically field-like in nature (e.g. gravity, magnetism) and may be long-range and invisible. Gravity acting as a global influence on physical entities are the simplest examples to recognize: water drops spilled on a table top, avalanches, and just how frequently do you dust the underside of your bookshelves? Figure 2.8 is another illustration of pattern formation. Part (a) illustrates a random scattering of iron rods on a tray; (b) represents the same tray of rods after being placed in a magnetic field; and (c) represents a different tray after being placed in a magnetic field - only one of the rods was plastic. The last diagram illustrates an important aspect of these influences. By global, I mean that these influences impose a coherent direction on all components/entities. I do not mean that they occur at a distant, long-range, or planetary scale, although these may also be the case. Due to their nature, these global influences are often so pervasive as to be unrecognizable.

By local influence, I refer to the blockages or constraints that act as an impediment for the components/entities. They typically occur on a meso- to micro-scale (defined in reference to the components/entities as microscopic). In some cases local influences may reflect a property of the components/entities (individually or collectively) such as the role of surface tension in creating the pattern of water drops on a table. The structures created tend to be mesoscopic - or at least, smaller than the global influences, and larger than the local influences. Consider gravity (macro, directional influence), landscape (meso, blocking influence), and water molecules (micro component) which create rivers (meso, structures).

This is the most basic interaction from which ordered structures emerge. Jantsch (1980, p 81) refers to such a system - one that "reflects only the interplay of static attracting and repelling forces" - as a conservative self-organizing system. To denote such systems as "self" organizing seems a mis-nomer, however. The systems may appear to organize themselves, but this just represents a lack of understanding regarding the influences, especially the global influences, at play. These systems do, however, involve mechanisms that are also relevant in more complex systems, making their discussion useful. In particular, they carry two important, and ubiquitous, self-organizing system characteristics: their similar but unique patterns, and the presence of system attractors. As random elements are constrained by similar rule sets, small irregularities generate unique emergent structures - no two rivers are the same and water spilled on the table will create a new pattern each time. The properties or characteristics of the local influences, then, will affect the structures that emerge.

Figure 2.9 illustrates just two of the factors affecting river formation and the emergent structures that result. White water is found where the grade is steep, and the substrate does not easily erode. Slow-moving, meandering rivers and their companion bow-lakes occur with gentle slopes and soft substrates. The notion of attractors can be useful here - each rule set (global local interaction) creates a different attractor. The three types of river systems illustrated in Figure 2.9 each represent a different attractor.

This description must be recognized as a very simplistic description of river formation, excluding their dynamic nature. By incorporating rainfall, groundwater flow, bed load, and other aspects, Iberall (1987) explains the self-organizing process of river formation in more detail and in a manner that takes it beyond Jantsch's definition of conservative self-organization.

Points of Dynamic Tension

Self-organizing systems are dynamic. They result from interactions among many different global and local influences, including feedback. Bénard cells, hexagonal convection cells that occur at a large scale in the atmosphere, are useful for explaining some of the factors involved. These structures can also be created experimentally by heating fluid in a pan (e.g. Prigogine and Stengers 1984, Cramer 1993, Mainzer 1994).

At a low temperature, heat transfers slowly through the fluid as illustrated in Figure 2.10. At a high temperature, the fluid becomes chaotic. At a critical intermediate point, Bénard cells form. These structures, arising seemingly out of nowhere, are referred to as dissipative structures (e.g. Prigogine and Stengers 1984, Cramer 1993). Critical factors affecting the system include the counteracting influences of gravity, and viscosity (Prigogine and Stengers 1984, Mainzer 1994).*14 Figure 2.11 represents these control parameters, noting the gradation of influence, and illustrating the critical point of tension where no factor fully controls the system's behaviour. Systems are held in a constant state of flipping between one control factor and another. Such points of tension are important and ubiquitous to self-organizing systems.

As with the formation of a river pattern, these structures are generated by the interaction of global and local influences that limit the system's behaviour. Rather than a random distribution of water molecules, they are constrained into a particular pattern. The difference in this situation is that the heat (energy) input holds the system at a critical point. These points will vary according to the properties of the system such as fluid density.

A key factor in creating the dynamic nature of this system is energy input. Prigogine, deriving a name from the fact that the systems dissipate energy, named the emergent structures dissipative structures (e.g. Nicolis and Prigogine 1989). It is from attempting to understand such systems that the importance of non-equilibrium thermodynamic conditions became recognized as critical (e.g. Nicolis and Prigogine 1989, Kay 1984, Schneider and Kay 1994). As noted, however, I do not wish to emphasize thermodynamics. I suggest reinterpreting the second law of thermodynamics as a global influence - entropy production provides a direction to systems that exchange energy in a manner comparable to the direction gravity provides for systems with mass.*15 Figure 2.11 must be recognized as incomplete, then. Figure 2.12 includes some of the other influences affecting the emergence of Bénard cells.

Similar influences are constitutive factors in more complex systems where emergent structures arise at points of tension among many different influences. For example, consider the factors determining leaf area in a plant, illustrated in Figure 2.13.

The key point to understand regarding these descriptions is that the influences described are causal influences. Complex systems and structures emerge because of the interactions.

Complexity

I have been using the term complex in a somewhat general sense, although I have also alluded to some difficulty with its definition. Here I note that it is also used precisely to refer to a specific class of system. Although not all self-organizing systems are complex, since many of the ones of interest to this discussion are, it is worthy to note some of the relevant concepts. For this particular portion, key references include Weinberg (1975), Campbell (1985), Cramer (1993), and Mainzer (1994).

Middle number systems Systems can be divided in three categories based on the number and complexity of components and relations (and I would add of global-local influences): small, medium and large number systems. (See Weinberg 1975 for a clear description.) The division primarily arises from the techniques we use to describe them. Small number systems, characterized by "organized simplicity," can be understood through analysis of their individual components and connections. The systems are simple enough that such analysis is still relevant when the pieces are put back together. Large number systems are characterized by "unorganized complexity." They can be understood by measuring properties of the whole system (e.g. temperature) or through statistical analyses (e.g. sampling individuals and averaging for the whole). In middle number systems neither approach is effective. As Weinberg (1975, p19-20) notes: "for systems between the small and large number extremes, there is an essential failure of the two classical methods." He continues by stating a "Law of Medium Numbers."

For medium number systems, we can expect that large fluctuations, irregularities, and discrepancy with any theory will occur more or less regularly.
The importance of the Law of Medium Numbers lies not in its power of prediction, but in the scope of its application. Although good mechanical and statistical systems are actually quite rare, we are literally surrounded by medium number systems. Computers have medium numbers of components, cells have medium numbers of enzymes, organizations have medium numbers of members, people have medium numbers of vocabulary words, and forests have medium numbers of tress, or flowers, or birds.
As with most general systems laws, we find a form of the Law of Medium Numbers in folklore. Translated into our daily experience - combining our familiarity with such systems and our ineptitude in their face - the Law of Medium Numbers becomes Murphy's Law:
Anything that can happen, will happen. (Weinberg 1975, p 20)

The subjective aspect of complexity is perhaps clearest here - middle number systems are complex because our potential for understanding them is most limited. However, I believe this is actually just another recursive, causal twist - mechanical and statistical analyses were derived first because these were simpler. Irrespective of how capable we are at describing the three types, it is in middle number systems that things happen. The clearest examples come from computer simulation, especially cellular automata. (See Langton 1992, Kauffman 1993, and Kelly 1994 for descriptions.)

Redundancy Another key aspect of complex systems is the need for a particular degree of redundancy. Consider the various pathways, through many different species, for nutrient flow or capture of solar radiation in an ecosystem. Redundancy is another enough-but-not-too-much characteristic - it must be balanced by some degree of unpredictability to allow innovation. Lack of redundancy makes a system vulnerable; too much makes it inefficient.

Redundancy makes probabilities unequal, instead of smoothing them out evenly across the whole range of possibilities. It means that the parts of a system are not wholly independent of one another, but are linked statistically, in a pattern of possibilities. (Campbell 1982, p 74)

Emergence This concept refers to 'wholes that are greater than the sum of their parts.' Consider water for example. Given an understanding of the properties and behaviour of the gases hydrogen and oxygen, the peculiar behaviour of water could not be predicted. This points to one of the difficulties with the term. Having observed the behaviour of water, we have been able to propose explanations that are based on the properties/characteristics of the components and the relation between them. Debate about the definition of emergence questions whether it should represent only those cases where the emergent property/system is impossible to predict even with complete knowledge of components and their relationships or whether it should refer to properties/systems that are surprising, allowing the need for subjective interpretation/relative definition. (See Nagel 1961, and Gilbert 1996.) Since it would always be impossible to know, by the first definition, whether a property is emergent or whether we are just ignorant, the latter definition seems more useful. Acknowledging fundamental uncertainty noted above, the distinction is actually a moot point.
Emergent characteristics, properties or systems, then, are those that arise unexpectedly from interactions among components; characteristics, properties or systems which cannot be predicted by the characteristics or properties of the system components, but which require some knowledge of the behaviour of the whole system in order to be able to be understood. River systems and the unique properties of water no longer appear emergent, but Bénard cells, life, and cognition illustrate the range of emergent structures, properties and behaviours.

These complexity characteristics seem more descriptive than causal when considered in relation to components and linkages. I believe, however, that they should also be considered in relation to global-local influences - when there are a middle number of interacting influences, more complex things happen. Consider the origin of life, for example. Given the presence of molecules in some sort of organic soup various estimates have been made that indicate the high improbability of getting the 'right' combination together, suggesting that life is simply a fortuitous accident. Due to the interaction of global-local influences, however, molecules will not have been randomly distributed. For example, molecules suspended in water will gather in currents and eddies, shoreline pools, ocean bottoms, etc. With varying specific gravities, they will necessarily be sorting and congregating (just as sediment is sorted on a river bed or beach). This provides a limited degree of predictability regarding the presence of other molecules of particular types, allowing dependence on the environment for particular inputs - another local influence. Given the global influence of the second law of thermodynamics, molecules will necessarily be dissipating energy. These molecules, then, are responding to an attractor that represents the convergence of a number of global-local influences. The potential for development of an attractor corresponding to the limited recursion of a simplistic sympoietic system (e.g. hypercycles, Eigen and Schuster 1979) is much more likely than the spontaneous generation of an autopoietic entity out of some random organic soup. The origin of life was a process, not an event (Swenson 1989, Gunther and Folke 1993).

Feedback and Recursion

Although some (e.g. Jantsch 1980, Dalenoort 1989a) include non-cybernetic systems as self-organizing systems, others (e.g. Kay 1984, Csanyi 1986, Kilmister 1986a) identify cybernetic mechanisms - the inclusion of feedback in system formation - as a critical defining factor in self-organization. Although feedback is obvious and critical in complex systems such as organisms, it is also an important factor on a simpler level. For example, positive feedback can be observed as a factor contributing to the formation of river systems on both small and large scales. On a small scale, individual drops of water alter the landscape by moving soil particles. Such a change in the local influence will affect the next drop of water, encouraging more water to follow in the same direction, creating bigger streams of water, which alter the local landscape even more. Enhanced by such positive feedback, small irregularities can be magnified substantially, causing a small change to create big ones. Such irregularities are critical factors in the self-organizing process.

On a large scale, feedback plays a part in river formation through the hydrologic cycle. Water evaporation from large bodies of water, provides a supply of rain which, following the river-formation/self-organization 'rule,' generates the river system structure, funneling water into the large body of water. Although on a human time scale this may create rather stable structures, on a geological scale, river systems change.

Information The jump in complexity between some of the systems discussed above, for example, Bénard cells and plants, is significant. Although I have drawn parallels, the mechanism that explains the emergence of Bénard cells cannot fully explain the latter. In plants, information is an important addition. Such systems - information dependent self-organizing systems - are emphasized by Pattee (1987).
At its most basic, information is recognizable pattern with the emphasis on recognizable, since information is an extrinsically defined concept. It is a means for reducing uncertainty. A bit is defined as the amount of information required to reduce an alternative or choice by half. This points to the context dependent quality of information. Since a signal or message is not really information until it has been received, it is dependent on the system receiving it. Repeating structures increase expectations by making patterns - the very essence of information. Any pattern, from the ordered arrangement of atoms to letters forming words on a page, has the potential to reduce uncertainty. The very essence of self-organization, then, involves the generation of information. Patterns generated through the interaction of global-local influences provide some degree of predictability, hence reducing uncertainty and, consequently producing information. This aspect of self-organization is critical, especially for generation of more complex systems such as described regarding the origin of life.
As noted above, my use of information differs from Maturana and Varela; however, I find it useful for considering developmental and evolutionary change in poietic systems, and therefore include it as a key concept for consideration.

Recursion Global-local interactions generate rule sets, feedback reinforces them. The combinations create information and redundancy, which allow complexity. The tendency is to consider these processes with a focus on components. However, they are also relevant to the relations among components. The process of self-organization generates new sets of relations among components by restricting their behaviour. Feedback introduces a new, and critical, relation. Reiterative application of these rules and feedback create a 'pattern of possibilities.' This not only allows and enables, but eventuates and guarantees the generation of increasingly complex relations among components. Since these factors are critical to poietic systems, I consider them in more detail by returning to the discussion on poietic systems by incorporating ideas discussed in this section.

2.4 Revisiting Poietic Systems

The description of self-organization in the preceding section provides valuable concepts for understanding autopoietic and sympoietic systems. I describe the three basic system types to summarize the discussion to this point. In each case I note an "ideal" example that I will use throughout the remainder of this thesis. I then discuss the evolution of these system types, drawing attention to two different types of self-organization. To close the chapter I consider some implications regarding future causality.

2.4.1 System Types

Self-organizing systems are emergent systems generated by the interaction of global and local influences. They arise at critical points of tension that reflect a balancing among the various influences, including feedback.

"Ideal" example: Rivers Although rivers do not represent an example most illustrative of some self-organizing characteristics, they have significant heuristic quality. The interplay between gravity and landscape make it a useful example on a metaphorical level, so I will use this example to draw attention to self-organizing factors. It is critical, however, to remember the role of feedback on both small and large scales in such systems.

Sympoietic systems represent an intermediate category. They are not quite self-producing, but are more than self-organizing. Of particular importance regarding differences from the latter is the increased emphasis on recursion and information content. They are unbounded, evolutionary and depend on cooperative relations among components. Although the systems have pattern and demonstrate a dynamic balance, they are inherently unpredictable. The systems rely on the addition of new information as a source of adaptive potential and use creative self-organization.

"Ideal" example: Ecosystems These complex living systems were the original template from which the concept of sympoiesis was developed. In consequence, they provide an excellent example for understanding.

Autopoietic systems are self-producing. They contain all the organizational information necessary for their own development and continuation. This does not make them independent of their environment, however, since they depend on structural inputs. The systems are bounded and homeostatic, and, since they rely on transmitted self-organization, are relatively predictable.

"Ideal" example: Organisms Although Maturana and Varela (e.g. 1980) use the cell as the prime example of autopoietic systems, I find that characteristics of organisms are more useful, since they can be more easily visualized. This is especially relevant for application of these concepts outside of the biological realm.

These significantly different characteristics are critical for obtaining a full understanding of the complex systems that are relevant for sustainability. The systems types most relevant to the complex systems that are of concern are autopoietic and sympoietic systems. In the following chapters I will primarily refer to these systems. The point of understanding self-organization is to recognize it as a causal factor generating sympoietic systems. Since the latter systems lack self-defined boundaries, they must be defined by these factors.

2.4.2 Evolution of Complexity

Complexity is a subjective notion that rests on our ability to comprehend (LoPresti 1996). Systems and behaviours are complex because we lack the ability to explain them, as noted in the discussion on middle number systems above. While I agree that this argument is relevant on a detailed or precise level, I maintain that a qualitative change in the degree of complexity exhibited by systems has occurred through time. A trend of increasing complexity from the origin of the universe to the current seems to be the most coherent explanation.*16 There is an observable and parallel increase from physical systems to living and cultural systems. Recognizing the different organizing processes, I perceive a progression from self-organizing to sympoietic to autopoietic systems. This evolution of complexity, or complexification (Swenson 1989) is another recursive process, since the generation of autopoietic systems provides new 'elements' which can be self-organized into new - more complex - self-organizing systems. These, in turn, can become sympoietic, followed by autopoietic systems at this new level. This recursive, evolutionary trend indicates the development of holarchies - nested and overlapping sets of systems. While I seldom explicitly draw attention to holarchies, their presence is implicit in discussion of complex systems composed of other complex systems. Evolution of multicellular colonies and organisms, and societies indicate these changes and the recognition of holarchies. Reconsidering the evolution of complex systems illustrated in Figure 2.6, the left-hand side represents development of increasingly complex self-organizing systems.*17 Of the systems listed at the beginning of the previous section (Box 2.8), these systems include the conservative and dissipative systems noted by Jantsch (which include dissipative structures), homogeneous cybernetic systems noted by Dalenoort, and the statistical self-organizing systems noted by Patee. Although developing a more comprehensive and detailed typology of this sort, by linking various typologies would be valuable, this is not my purpose here. My interest lies with the more complex systems - systems in which information and recursion are critical factors: living systems and beyond. It is important to recognize that the other self-organizing processes discussed are also constitutive factors in such systems. As noted earlier, each new level of complexity involves new rules, but systems must still adhere to the old ones as well. Humans can think, but cannot fly unless they overcome the dictates of gravity through such thinking and the subsequent development of technology. To understand the role of self-organization in autopoietic and sympoietic systems, I define two different types of self-organization which reflect similarities to those listed above, but which draw out different factors. I refer to the two types as creative and transmitted self-organization, making the distinction based primarily on the role of information and recursion.

Creative self-organization

Creative self-organization refers to the process described above in which structures emerge at critical points of tension among global and local influences. It includes Prigogine's description of dissipative structures (e.g. Prigogine and Stengers 1984) up to the emergent behaviour of complex systems recognized through computer simulations (e.g. Langton 1992). The resultant behaviours and structures are not necessarily sustainable or self-producing, relying on the system poising at this position of dynamic tension, held by counteracting influences. If the tension is lost, so is the emergent structure and behaviour.

Transmitted self-organization

In contrast, transmitted self-organization refers to the self-organizing ability of autopoietic systems which has been codified and passed on through generations. It refers to the self-organization potential of systems which initially began through creative self-organization but which have passed this ability on through subsequent generations. Many of these systems have reached a level of complexity that precludes spontaneous generation (i.e. creative self-organization) of another, similar, entity. The systems are consequently reliant on self-production and reproduction. These systems can organize themselves by arranging their components in the requisite manner, but they do this through preprogramming. The original creative self-organization process will not necessarily be apparent in continued poiesis. This somewhat static form of self-organization allows increased complexity but restricts flexibility and is a characteristic of autopoietic systems. The systems resulting from the two forms of self-organization will have very different properties. They correspond to the two system types at either end of the self-organizing portion of the continuum I drew above (Figure 2.2) from self-organizing (creative) to autopoietic (transmitted) systems. Sympoietic systems are, to some extent, a combination. They emerge from creative self-organization among systems with transmitted self-organizing ability. When comparing autopoietic and sympoietic system types, I will tend to refer to sympoietic as creative self-organizing; however, it is essential to recognize the importance of the autopoietic components involved.

Further Considerations

I make note of three final points. First, considering the evolution of complexity, I believe it would be somewhat ludicrous to believe that we might be the 'end of the chain,' the epitome of complexity - on two counts. One, this is a big universe. Given the notion of self-organization, it seems quite likely that there are other complex systems out there. Two, there is no logical, rational, observational reason to exclude the possibility of systems more complex than ours which are beyond our ken. Boulding (1957) comfortably named them "transcendent systems." Again, given the notion of self-organization, the presence of such systems moves from the possible-though-perhaps-unlikely-and-hard-to-believe to the not-only-quite-possible-but-even-quite-likely. Whether such systems exist now or will some time in the future I leave open for others to debate. I point out, however, that the notion of emergence also suggests that we can never reach the 'next level up' in order to observe the properties emerging and can only guess at their presence through inference, yet these properties and systems may have significant impacts on ours. The second point is that these systems, in a very real sense, are inevitable. There are causal influences here that ensure a certain degree of inevitability in the evolution of complexity. This must not, however, be confused with such evolution being predictable. There is an insurmountable degree of uncertainty and irregularity involved. The systems are generated by the organizing factors involved, the exact same factors would create the exact same systems over and over again. This is why chaotic behaviour is recognizable, and simulatable. Due to the impossibility of ever exactly knowing the factors involved in real systems, however, they cannot be predicted. Whether the universe and all systems in it are fundamentally driven by deterministic or stochastic factors is a moot point. It is also critical to recognize the importance of "historical accident." Small irregularities, reinforced through positive feedback, can create large changes in outcomes. Such outcomes may become codified. They may become locked in as a basic rule, generating new systems, which will consequently be dependent on the original irregularity. In the more complex, information dependent systems this is especially true. As Pezzey (1992, p 325) puts it: "once an advantageous innovation (whether a sharper tooth, a sharper sword, or a superior legal system) is made, it will eventually be spread by natural selection until it becomes a need." This ratchet effect*18 illustrates the combined evolutionary and recursive power of increasing complexity. It has a crucial influence in human systems and will return as a key factor through the remaining discussion. The third point refers back to the notion that we cannot expect to explain or understand more complex systems using the same tools and heuristics that we use to explain or understand 'simpler,' more fundamental systems. The new systems and behaviours emerging at each new level are 'special cases' of the more fundamental systems and behaviours at the preceding level. Each jump in complexity involves new restrictions on components. In consequence, their explanation requires new 'special' tools and heuristics. This recognition must provide a caution regarding extension of these concepts into the increasingly complex individual and social humans systems. Yet I believe this extension has significant explanatory potential, so, having noted the concern, I will continue with application of these systems concepts.

2.4.3 Comparison of Concepts

The most unique aspect of my thesis is the introduction of sympoiesis, a concept which incorporates ideas from self-organizing systems and autopoiesis. Compared with self-organization, its strength lies in emphasizing recursion and information as well as self-organizing factors. In contrast with autopoiesis, it emphasizes the inadequacy of organizational closure and self-defined boundaries for conceptualizing many complex systems. As such, it resolves some of the difficulties and debates in the literature regarding autopoiesis, in particular with respect to the problematic discussions regarding boundary issues. (This is especially true for social systems, a notion discussed in the next chapter.) There are a number of concepts in the literature that reflect characteristics similar to those I describe for sympoietic systems. Some of these are noted in Box 2.9. In comparison, sympoietic systems are somewhat different from SOHO (self-organizing holarchic open) systems (Koestler 1978, Regier and Kay 1996), CANL (complex adaptive non-linear) systems (Bella 1997), and complex adaptive systems (e.g. Holland 1992, Kauffman 1993). In essence, they are a special case of these more general complex systems conceptions. The key difference is inclusion of 'self'-production. In addition, by allowing contrast between bounded and unbounded systems, the concepts draw attention to particular characteristics and allow for a more detailed conceptualization. Of particular importance here is the distinction between systems that are organizationally open, closed, and ajar, emphasizing the information aspect of systems. I believe, then, that characterization of sympoiesis provides considerations that make its continued use valuable. Perhaps most useful is the contrast between autopoietic and sympoietic systems - a heuristic which I will demonstrate through the remainder of the thesis. Future research must involve integrating these concepts more fully, drawing comparisons and distinctions among them. In this thesis, however, my emphasis has been on developing the concept of sympoiesis and drawing out the distinctions between sympoietic and autopoietic systems. I have maintained this focus because I believe plurality provides an advantage. The complexities of the natural and social systems we must cope with in targeting toward sustainability cannot be fully understood from a single perspective - another view of reality is a benefit. It is this I have been striving for. Although the basic ideas set forward in this chapter parallel other system descriptions, I believe the distinctions and subtleties provide new and different opportunities for understanding. The remaining sections in this chapter indicate some of the concerns and implications.

2.4.4 Future Causality

I rely on an evolutionary perspective, emphasizing that the process of self-organization has generated increasingly complex systems which in turn, self-organize into more complex systems. I believe this evolution of complexity carries into cognitive and social systems and, quite possibly, beyond. As such, it has significant implications for sustainability of our species and for the activity and process of planning. A succinct argument indicating the potential seriousness of the implications is an argument regarding causality presented by Campbell (1985). Campbell's central thesis is that "when matter is appropriately organized, it becomes sensitive to causes arising from the future instead of just the past" (ibid., p 154). Table 2.2, taken from his paper, expresses the relation between system organization and causality exhibited by different types of systems. As he argues, the first three fit easily into the conventional scientific notion of causality which is based on belief that causes arise prior to their effects. At the next level, in cybernetic systems, effects and causes cannot be separated. Campbell uses the example of a cybernetic system created by placing microphone and speaker of a PA system close together. "All of the characteristics of the final screech - its frequency spectrum, intensity, stability, and so forth - are determined entirely by the organizational characteristics of the feedback circuitry" (ibid., p 158). This presents a different view of causality than the conventional notion since the initial trigger is not the cause. An investigator pursuing the initiator as the cause, would be "chasing after the noise in the system instead of the effective cause" (ibid., p 159). In such a system the organizational characteristics are a critical factor related to the effect. Changes in the outcome occur, not because of changes in the trigger, but because of changes to the set of relations that define the organization of the system.

Table 2.2 - Causality and Organization
Type of Organizations	Emergent Causal Property
None (elementary particles only)	Acausal
Mechanical objects	Deterministic cause and effect
Negentropy	Unidirectional cause and effect
Information
Information about self	Recursive causes and their effects
Information about future self	Future causality

In another example, emphasizing the importance of organization, Campbell notes that "it is now suspected that developmental changes [in organisms] are programmed in a very different manner from controlling the inducers of change. Tissues may be internally organized so that when a biochemical perturbation occurs it will cycle recursively and evolve itself and the tissues into a specified final form" (ibid., p 159). This claim parallels the work of Kauffman (e.g. 1993) who studies genetic algorithms and adaptive landscapes. He argues that cell differentiation in an organism results from responses to particular attractors. Both arguments cohere with the notion of morphogenic attractors (e.g. see Cramer 1993). As system organization becomes increasingly complex, so does causality. With another increase in complexity, we reach "mentality [which] is an enriched mode of causality that emerges from the special organizational state of the brain" (Campbell 1985, p 160). The special characteristics that arise include the potential for self-reference, and future self-reference. "Even the capacity of causality to operate from the future to the present loses its utter absurdity as an emergent property of the right sort of ultracomplex organization" (ibid. p 161). Campbell argues that understanding the complexity of human behaviours does not require "metaphysical silliness" but only questioning "what sorts of organizational features could sensitize the behaviour of matter to its future form" (ibid. p 161). The implications of such recognition, especially for planning are significant: A future self-referent system that also has mechanical ability to operate on the physical world around it is capable of extraordinary causal behaviour. It can modify the structure of itself and its surrounding under the direction of its internal future description so as to assume the form that it internally describes. This allows a representation of the future to direct activity in the present. In effect, future self-reference gives the future access to act causally on the present.
Future causality is the relationship of such attainable preconceived future organizational states to the events that bring them into being through preconception. (Campbell 1985, p 161-2) Along similar lines, and stating the view of a humanist psychologist, Smith (1993, p 18) notes: "We are our choices." Bolan (1996, p 499), a planner, suggests that "humans are both the creators and the products of their social worlds." The lesson arising from these comments is one of the central points of this thesis: unexpected consequences arise because of our actions, not in spite of them. Considering that the whole purpose of planning is to generate 'future descriptions' and then to 'modify structures and surroundings,' the implications are critical to recognize, acknowledge, and account for. I will return to these notions in the sections on planning.

2.4.5 Sustainability

Science - our pursuit of understanding - is a quest to uncover the patterns of organization in complex systems. This is especially the case with respect to the systems we depend upon, the biophysical sympoietic systems that are our 'environment' - the systems that provide us with the essential services necessary for existence. The key difficulty is that structure is the only aspect of these systems which is accessible to our senses. (In social systems, even this may not be tangible.) Plurality is, in some sense, created by differing interpretations of system pattern of organization. I believe that we have generally been assuming autopoietic (or bounded and non self-producing, not even self-organizing) pattern of organization. It is consequently time to recognize the presence of sympoietic systems and adjust our understanding. Such a shift is essential for attempting to maintain sustainable conditions. The differences between the systems types suggest that the characteristics required for system sustainability are quite different among the different types. Reconsider the three elements of sustainability identified in Chapter 1. The most basic and essential difference between autopoietic and sympoietic systems relates to the balance between maintainability and adaptability that each system type relies upon. Autopoietic systems must maintain organizational closure to be sustainable, emphasizing the maintainability element, whereas sympoietic systems use a continual flux of information, increasing their adaptability element. The standard perception of equilibrium described in Section 2.2.1 reasonably reflects autopoietic system requirements. While environmental uncertainty is anathema to autopoietic systems, it is an adaptive advantage for sympoietic systems. The behaviour suitable for autopoietic systems is not sustainable in sympoietic systems. These differences point to a fundamental contradiction that arises. As biophysical autopoietic entities, we require predictable inputs, however, these inputs must come from an environment that is continually changing. So there is a fundamental paradox here. Sympoietic systems arise and evolve by virtue of the complexity of their component autopoietic systems. Due to their complexity, autopoietic systems require specific predictable inputs from the sympoietic systems they are embedded within, yet the latter depend on uncertainty and continual change for their continued existence. This uncertainty and continual change, of course, results from the complex interactions among the increasingly complex autopoietic systems. Neither system is 'better,' or more 'independent,' than the other. I refer to this as the paradox of interdependence and return to its consideration in the next chapter. We humans play our part, then, by causing change in the sympoietic systems we are embedded within, yet by doing so we threaten the potential for sustaining suitable conditions for our own persistence. There is balancing here that is critical - balancing between change and non-change. From a concern for the human condition, however, I believe we currently have cause for concern. As noted in Chapter 1, the stresses we are placing on the sympoietic systems (both biophysical and socio-cultural) we are embedded within are of such a magnitude and direction that we are altering the balancing act in a manner which may be detrimental to our existence. These notions are relevant regarding our approaches to planning. As biophysical autopoietic entities, we rely on particular inputs from our environment. On a very basic level, then, our need to plan is based on a need to secure suitable environmental conditions for the future. Yet these are the very conditions which we are changing through our actions. To alter this negative influence it is imperative to understand the systems that guide our actions - our socio-cultural systems. Of particular importance, because of the significant intentionally prescriptive role they play, is an understanding of our planning systems. Throughout human history we have used planning as a means of influencing and governing our interaction with the natural environment. We have done so with much success, although also some failures. Over the past century planning has been based primarily on the conventional scientific approach, using the latter both for provision of knowledge and as a model for designing planning systems and approaches. The question, now, is whether these have the potential for achieving sustainability. I believe they do not. The different perspectives articulated here, however, can provide some new insight. This insight can be used for evaluating the approaches we use, and our consequent potential for attaining sustainability. In addition they provide clues regarding the approaches we should use and our ability to adopt them. Taking the same direction as the evolution of complexity, then, I progress from discussion of self-organizing, autopoietic, and sympoietic systems in the physical and biological domains to consider the influence of self-organizing factors on and in human systems. This discussion, which includes individual and collective human systems, on psychological, social, and cultural levels, is the subject of the next chapter.

Chapter Footnotes:

*1 Termed the three-body problem, the behaviour is an example of deterministic chaos.

*2 Pronunciation: auto as in automatic, poi as in oil, esis as in thesis or as in morphesis; syn as in synthetic, etc.
I often shorten poi to po as in go. The root word is poienen (Gr.), to produce, the same root word as in poem.

*3 Maturana and Varela have a rather thick writing style. For a more reader friendly explanation, see Mingers (1995, especially p 9-21).

*4 Maturana (1980) includes a useful glossary of terms.

*5 Maturana and Varela (e.g. 1980) simply use 'organization.' However, this can cause confusion when extending the discussion into social systems. I follow Capra (1996) using the phrase 'pattern of organization' in order to reduce this confusion. When my meaning is clear, I will occasionally use only 'organization.'

*6 There is an important definitional issue to note here. In 'systems' language, structure represents the relation among components, what is here being termed pattern of organization. Structure, as used here, more closely represents common usage, which typically refers to a physical entity.

*7 See Burghgrave (1992) for an interesting example using computers.

*8 Maturana and Varela discuss only open and closed organization.

*9 There is a certain degree of subjectivity here, which arises from categorization. The pattern of organization of any system defines it as a member of a particular class of systems. The role of the observer is to define the class. For example, one could separate people into two organizational classes: those who are and who are not allergic to peanuts. This does not negate the importance of structural coupling. In the end it is the system's structural interaction with the environment that determines continued autopoiesis. Nor does it negate recognition that the system, not the observer, defines its own boundaries, as argued more clearly below.

*10 Maturana and Varela use the term self-producing.

*11 Perhaps ironically I first came across the concept of autopoiesis in philosophical research on the moral considerability of ecosystems (Fox 1990). I returned to it later, upon finding the biological and ecological literature inadequate, when trying to grapple with the issue of sustainability in ecosystems (Dempster 1995). This indicates the lack of attention the concept has received in the biological literature. See Mingers (1995) for a relatively recent review of the use of the autopoietic concept in literature from various disciplines.

*12 See Maturana and Varela (1980, especially p. 90-92, 98-102), Beer (1980, especially p 68-9 ) for their arguments, and Pattee (1987) for a more general discussion on information and self-organization.

*13 Some interesting implications regarding adaptability theory (Conrad 1983) could be made here.

*14 Although generated by a relatively simple system, Bénard cells formation is not fully understood. The factors illustrated here do not provide a full explanation, however, they are instructive for explaining some of the critical processes that occur.

*15 This is a very generalized interpretation, which does not recognize some of the trickier issues, such as that entropy cannot be defined in an absolute sense, but only in relation to particular systems. See Schneider and Kay (1994, especially 629-633) for further explanations and an argument for recognizing thermodynamics as a critical theoretical basis for ecology. These authors note that their "re-examination of thermodynamics shows that the second law underlies and determines the direction of many processes observed in the development of living systems" (p 629).

*16 I do occasionally try to visualize alternate progressions, yet am perhaps too tightly tied to my cultural conditioning. For those who may think differently, I do not believe an alternate viewpoint negates the value of the concepts articulated in this thesis. I believe the dominant scientific perspective parallels my view, although it seldom admits to some of the relations, types of causality, and possibilities articulated here. I believe the discussion here provides a 'scientifically' based argument that at least points out that there is a lack of justifiable reasons for negating other possibilities. I, however, maintain the assumption articulated at the beginning establishing the origin of the universe as the effective beginning.

*17 While there maybe a tendency to include simple physical/mechanical systems as the most elementary in the evolutionary chain, followed by thermodynamic systems, such interpretation is inaccurate. Although useful for understanding, progression from mechanical to thermodynamic should not be considered as an evolutionary sequence since pure mechanical (i.e. non-thermodynamic) systems never existed.

*18 This is sometimes referred to as the 'treadmill effect' (e.g. Schnaiberg 1990, Pezzey 1992), which I find to be a less explanatory term.