International Association of Facilitators
1999 Annual Meeting
Williamsburg, Virginia, USA

January 14-17 1999

Thread #1: Systems Thinking

Conditions That Support Self-Organization in A Complex Adaptive System

Glenda Eoyang and Doris Jane Conway
Management Analysis Division
203 Administration Building
50 Sherburne Avenue,
St. Paul, MN 55155
Fax: 651-297-1117
TTY: 800-627-3529

New models in mathematics and the physical sciences are opening a conversation about system-wide behaviors that are at the same time unpredictable but patterned, emergent but self-referential, complex but shaped by simple rules. Descriptive models of these strange systems come from a wide variety of disciplines, including ecology, economics, topology, computer and cognitive sciences, biology, and physics. The emerging fields that investigate these strange systems are called by many different names: new science, chaos theory, complexity theory, complex adaptive systems research, nonlinear dynamics, dynamical systems theory, and synergetics. This presentation will focus on the phenomena of self-organization, the conditions that seem to set the stage for self-organization to occur, and what complex adaptive system theory tells us about good facilitation. We will use some new language and provide some new insights into the facilitator's role.

Applications of complexity theory to human systems fall into two fairly clear-cut categories. The first is metaphorical and descriptive in nature. It derives language and concepts from the new science and applies them to explain what happens in a complex human environment. These applications focus on finding and pointing out examples of self-organization, emergence, butterfly effects, and attractors in human systems. The works of Margaret Wheatley, Myron Kellner-Rogers, and Laurie Fitzgerald fall into this category. They use the models and metaphors from the new science to describe outcomes of processes rather than how the processes work. They do not focus on why or how complex patterns emerge. In some ways, they treat the complex system as if it were a black box.

The second approach to the new sciences is more mechanical (not mechanistic). It also derives models and language from the science and mathematics of complex systems, but its goal is to investigate how things happen in human environments. The purpose of the work is to open the black box and come to some understanding of how complex systems generate their ever-surprising behaviors. The works of Ralph Stacey (to some extent), Jeffrey Goldstein, Stephen Guastello, and Kevin Dooley define this second category of work in the field.

This paper will pursue the more mechanical of the two approaches. It will articulate a model of mechanisms for how self-organization emerges in a complex system and what professional facilitators can learn from complexity to improve their understanding and practice of the art of facilitation. The paper includes three parts: (1) definition of terms, (2) conditions for self-organization and (3) facilitation in self-organizing systems.

Definition of Terms

Diverse vocabularies and inconsistent definitions of terms are perpetual barriers to the study and application of complexity to human systems. Two factors contribute to this lack of clarity within the field. First, the science is new, and its application to organizational systems is even newer. No single model or set of terms has emerged as a standard. Second, the field takes its roots from a variety of disciplines. Each discipline has its own history and method of discourse, so many different symbol sets have found their way into the field. A coherent set of models and metaphors may emerge as the science develops. In the meantime, each researcher must define his or her own terms. The following terms will appear in this paper.

A complex adaptive system is a collection of interdependent rule-following agents with complex interactions resulting in system-wide patterns across the group. Computer simulation models called "cellular automata" (e.g., Game of Life and Boids) demonstrate how separate agents can independently follow rules of behavior and collectively generate emergent patterns at a group level. An example of a simple rule set from a California school system is one where the school threw out the detailed rule book and operated with the following three "rules": (1) take care of yourself, (2) take care of others, and (3) take care of this place. Table I summarizes the relative characteristics of simple, complex, and random human systems.

Table 1. Simple, complex, and random human systems.





Central metaphor


Living organism

Hot gas


Finite number

Clear categories Interchangeable

Infinite in number

Each is unique

Highly connected

Infinite in number

Each is unique



Same as sum of parts Imposed design Known by reduction

Different from parts Self-organized design Emergent patterns

Statistical averages

Collection of parts

No patterns

Decision basis

Fact and data



Useful when

Familiar problem

Little freedom Whole is sum of parts

Problem is new

High freedom

Many connections

Unfamiliar problem

High freedom

No connections



Manufacturing process

Quality control

Performing teams



Customer contacts

Political changes

Media effects

Leadership roles

Predict the future

Control other's behavior

Maintain order

Manage edges

Manage patterns

Manage paradox

Manage transformation

Manage action

Protect self and others

Search for patterns














Indifference to change

Reactive, not proactive







Container is the external boundary that defines the "self" that develops during a self-organizing process. In physical systems, the container is a physical barrier, such as the wall of a beaker or the edge of an isolated ecosystem. In human systems, the container is less obvious, but it is always the essential attribute that is perceived to separate the "self" from its environment.

Coupling is a pattern of relationship in a complex system. A feedback loop sets an immediate relationship (a couple). Habitual feedback loops generate patterns of relationship across the system. There are three kinds of coupling:

Tight - Both sides respond immediately to the other. The control is close and little freedom is experienced.

This would be exemplified by a facilitator who keeps tight rein on what is said and when.

Loose - The sides are aware of and adapt to each other, but the response can be slower and less exact. The relationship is mutually adaptive. This would be exemplified by a facilitator who begins with a plan, but is willing and able to adjust the plan as the day goes along.

Uncoupled - The sides are not related in any way. A change on one side doesn't affect the other. This can be a dysfunctional relationship or a functional, short-term response. Examples of uncoupling are a facilitator leaving the room for a group to think through its own issue; product design and marketing staff not talking; a manager hiding in her of office.

Difference is the internal variation in a complex system that defines its level of equilibrium. When there is little difference among the agents of the system, it is said to be in equilibrium. As the difference increases, the system moves away from equilibrium. When the difference is great, the system is said to be far from equilibrium. When a system is far from equilibrium, it has the capacity to self-organize.

Self-organization is the tendency of a complex adaptive system, when pushed far from equilibrium, to generate new organizational forms spontaneously. The works of Ilya Prigogine provide information about self-organization and what he refers to as the dissipative structures that result from self-organizing processes. He draws on models from thermodynamics to gain insight into the process of self-organization.

Transforming-feedback loops support the exchange of transforming information, energy, or material among the agents in the complex system. They allow various agents to engage with each other across their differences. They require (1) a two-way transfer and (2) mutual change, and (3) they permanently alter both sender and receiver. Either positive or negative feedback can be transforming. The feedback can be amplified or dampened. There is no single factor that determines the behavior of a system. Transforming feedback loops result in changes internal to agents, between and among agents, and ultimately in the system as a whole. This whole-system transformation is called self-organization.

Conditions For Self-Organization

Research in various areas, including product design, creativity, cognitive development, school reform, discourse analysis, and small-group dynamics, has described a developmental process of emergence in a complex human system. The steps in the process include:

1. Differentiation. During this step a difference is recognized in the system and is amplified by some internal dynamic or some environmental influence.

2. Transforming feedback or coupling. During this step, a transforming bridge is built across the differentiation identified in Step 1.

3. Self-organization. During this final step, the system moves to a new state in which the structural framework of the past is superseded by an emergent structure that responds to the feedback loops established in Step 2. However, there are no guarantees that self-organizing will occur or that it will occur in ways anticipated or even for the good.

As a result of step 3, new differences become significant in the system, and the self-organizing process moves into its next cycle of differentiation, coupling, and self-organization.

Three conditions must be met for the self-organizing process to generate coherent patterns:

First, the system must be surrounded by a containing boundary. This condition defines the "self" that will be developed during the self-organizing process. If the containing boundary is not clear or if it is too large, self-organization may not be recognized. The self-organizing process will take so long or the pattern will be so global that an observer will not be able to detect it. Without evidence of an emergent pattern, an observer can never know whether self-organization occurred in the system. Second, the system must incorporate some significant difference before the self-organizing process will begin. The number of significant differences must be quite small. A limited number of differences allow the self-organizing process to proceed on a relatively uninterrupted track. If the dynamics of the system are revolving around many focal differences, multiple and competing self-organizing paths will emerge. In such complicated conditions, it's possible that no one structure will receive sufficient resources from the system to be defined clearly in the process of self-organizing. Third, the system must include transforming feedback loops that allow organizing information and energy to flow across the system. If no system-wide communication mechanisms are present, no system-wide structures can emerge.

These three conditions establish the environment in which a complex adaptive system can self-organize to generate new relationships among agents and therefore new system-wide structures.

Facilitation in Self-Organizing Systems

As our awareness of the conditions supporting self-organization has developed, we can look at our own facilitation practices and understand in a new way the practices of facilitation that support self organization. These conditions we have intuitively been aware of and used, but we have not had the language from the science of complexity to describe what was happening.

In notes from a Margaret Wheatley 1996 Cape Cod Institute presentation, she talks about the domains of self-organization being the three categories of identity, relationships, and information. For a historically trained Institute of Cultural Affairs (ICA) Technology of Participation (ToP) methods person, these categories are represented in the cultural section of the social process triangles. In the '90s, Wheatley has given us permission to focus directly on the cultural triangle, rather than indirectly, as ICA historically has tended to do. (How exciting that was when I first heard her thinking!) In fact, her comments and my response serve as an example of articulating a "difference that makes a difference."

Within the domains we will consider the conditions supporting self-organization. This is a minority view, an acknowledgment of the need for structure to a greater degree than the Wheatley advocates will suggest. But it is a view that intuitively and experientially seems honest and true in our work.

The first condition is that the system must be surrounded by a containing boundary. The container is made up of environmental elements for the group to work within. It includes such practices as clarifying the purpose of the group's work; stating the givers, norms, operating guidelines, mission, values, and common vision; and providing creative questions. From ToP methodology, these practices establish rational and experiential objectives and a framework for the questions. Anything providing context for the situation so that everyone is on the same page is part of the container. Considering the individuals within the group, the container might be their own sense of their role in the group, their sense of self-esteem, or their ability to be self-referenced (i.e., realize and trust their own internal knowing).

If a container is not created and articulated, a group can flounder for a long time with unclear boundaries, differing expectations, and unclarified roles.

The second condition is that the system must incorporate some significant cant difference before the self-organizing process will begin. We know from science that diversity produces a healthy system; we constantly struggle with acknowledging and incorporating the diversity present in every situation. What are the differences that make a difference? How do we become aware of them? How do we use them? As facilitators, we strive to enable people to state their own perspectives, and we encourage the diverse thinking (divergent thinking) before we move to convergent thinking for group consensus, as in Sam Kaner's excellent model. Historically our ToP methods have focused on seeking the common ground. Sometimes the differences are ignored; high differentiation is needed as a condition for self-organizing. Both differentiation and communication are necessary for self-organization to take place.

A public sector management team's exercise last spring - taken from Glenda Eoyang's book (see bibliography) -looked at the differences that make a difference. The team listed 15 differences its members could think of - places where boundaries existed. The obvious ones were age, gender, ethnic background, and education levels. Then the team looked at areas that were really differences that made a difference in the workplace. The question was worded as, "Which of these boundaries make a difference in the group's behavior?" The team noted six areas of difference: (1) perspective (big picture versus focused), (2) learning types, (3) peak operating times (morning person - non-morning person), (4) Myers-Briggs types, (5) focal points - business administration, products, or consulting, and (6) staff positions. The team chose Myers-Briggs types difference and talked about the boundaries among these type differences. Questions were: "What happens on each side of these boundaries in the best and worst circumstances?" and "What would happen if we could move the boundaries so that some of us show up on different sides of the boundaries?" This enabled the team to appreciate the differences and how they enrich the team's work and advance its purposes.

Examining boundaries is one way to identify differences that make a difference. Boundaries are places in the system where the different converge. As facilitators, we have the privilege to address boundaries not commonly addressed, like the connection between a person's spiritual and professional existence. We can pay attention to boundaries that others are not aware of or choose not to recognize. We can address people's fears and frustrations. Part of a facilitator's role is to notice the boundaries and decide which need to be constant and which need to change. What difference might it make if a facilitator, for example, focused on different learning styles rather than the hierarchical boundaries of a team?

The third condition is that the system must include transforming feedback loops. "Transforming" means having a profound influence on a group. Both sender and receiver are responsive and are permanently altered. "Feedback loops," means communication created by different types of couples within an organization. We can establish couples across boundaries to connect parts of the system that are different from each other. As facilitators, we can share our own experiences to create the coupling between the group members and us. Within a facilitated process, transforming-feedback loops can occur through a number of facilitation techniques. Some of the many possibilities: employing the four levels of the focused conversation method to create feedback loops that allow different preferences of knowing to emerge and be reflected in the group's thinking; being respectful of and attentive to the situation and the people involved; applying a design that is thought through and customized to the group; being flexible in adjusting the process to newly emerging circumstances; following a pattern of working in small groups and later a large group; recognizing the importance of the meeting site and its setup; and

Feedback also includes such skills and techniques as listening, articulation of shared beliefs, metaphorical thinking, post-meeting written evaluation by participants, prepared agendas, meeting notes, analysis of the process, and communication of work results to a group of people beyond those who worked on it.

The keys to success in using this model are (1) an expectation that the facilitator initially analyze the group or organization to determine the real situation; the created container and the feedback need to come from the organization's history, rather than be detached from its history; (2) the facilitator's awareness of the feedback loops working within the group and decision whether new ones should be created; (3) the facilitator's creation of an appropriate and accurate container; (4) the facilitator being alert to the differences that are making a difference and articulating those to the group when it's appropriate; and (5) The facilitator must have only a few simple rules for the group's process. The group must be clear about the three conditions within which they are working.

When these conditions - a crafted container, an awareness of the differences that make a difference, and transforming feedback loops - are present in a situation, total freedom and creativity are available for self organization to happen.


Briggs, J., and Peat, F. D. (1989). Turbulent Mirror. New York: Harper Row.

Broomfield, J. (1997). Other Ways of Knowing - Recharting our future with ageless wisdom. Rochester, Vermont: Inner Traditions.

Eoyang, G. (1997). Coping with Chaos: Seven Simple Tools. Cheyenne, Wyo.: Lagumo Publishing.

Farrell, J. D. & Weaver, R. G., (1997). Managers as Facilitators a practical guide to getting work done in a changing workplace. San Francisco: Berrett-Koehler.

Goldstein, J. (1994). The Unshackled Organization. New York: Productivity Press.

Guastello, S. (1995). Chaos, Catastrophe, and Human Affairs: Applications of Nonlinear Dynamic to Work, Organizations, and Social Evolution. Mahwah, N.J.: Lawrence Erlbaum Associates, Inc.

Prigogine, I., and Stengers, I. (1988). Order Out of Chaos. New York: Bantam New Age Books.

Stacey, R. (1993). Managing the Unknowable. New York: Jossey-Bass.

----------, (1996). Creativity and Complexity in Organizations. San Francisco: Berrett-Koehler Publishers.

Wheatley, M. J. (1992). Leadership and the New Science. San Francisco: Berrett-Koehler Publishers.

----------, and Rogers, M. K.. (1997). A Simpler Way. San Francisco: Berrett-Koehler Publishers.