Edgeware - Primer


Description of complex adaptive systems

CAS have a number of linked attributes or properties. Because the attributes are all linked, it is impossible to identify the starting point for the list of attributes. Each attribute can be seen to be both a cause and effect of the other attributes. The attributes listed are all in stark contrast to the implicit assumptions underlying traditional management and Newtonian science.

CAS are embedded or nested in other CAS. Each individual agent in a CAS is itself a CAS. In an ecosystem, a tree in a forest is a CAS and is also an agent in the CAS of the forest which is an agent in the larger ecosystem of the island and so forth. In health care, a doctor is a CAS and also an agent in the department which is a CAS and an agent in the hospital which is a CAS and an agent in health care which is a CAS and an agent in society. The agents co-evolve with the CAS of which they are a part. The cause and effect is mutual rather than one-way. In the health care system, we see how the system is co-evolving with the health care organizations and practitioners which make up the whole. The entire system is emerging from a dense pattern of interactions.

What we could be doing
World wide complexity
Make it or let it

Hidden Order



Diversity is necessary for the sustainability of a CAS. Diversity is a source of information or novelty. As John Holland argues, the diversity of a CAS is the result of progressive adaptations. Diversity which is the result of adaptation also becomes the source of future adaptations. A decrease in diversity reduces the potential for future adaptations. It is for this reason that biologist E.O. Wilson argues that the rain forest is so critical to our planet. It has significantly more diversity - more potential for adaptation - than any other part of the planet. The planet needs this source of information and potential for long-term survival. In organizations, diversity is becoming seen as a key source of sustainability. Psychological profiles which identify individuals' dominant thinking styles have become popular management tools to ensure there is a sufficient level of diversity, at least in terms of thinking approaches, within teams in organizations. Diversity is seen as a key to innovation and long term viability.

Many of us were taught that biological innovation was due in large part to genetic random mutations. When these random mutations fit the environment better than their predecessor they had a higher chance of being retained in the gene pool. Adaptation or innovation by random mutation of genes explains only a small fraction of the biological diversity we experience today. Crossover of genetic material is a million times more common than mutation in nature according to John Holland. In essence, crossover suggests a mixing together of the same building blocks or genetic material into different combinations. Understanding this can lead to profound insights about CAS. The concept of genetic algorithms is paradoxical in that building blocks, genes or other raw elements which are recombined in a wide variety of ways are the key to sustainability. Yet the process of manipulating these blocks only occurs when they are in relationship to each other. In genetic terms, this means the whole string on a chromosome. Holland argues that "evolution remembers combinations of building blocks that increase fitness." It is the relationship between the building blocks which is significant rather than the building blocks themselves. The focus is on the inter-relationships.


In organizational terms, this suggests that it is not the individual that is most critical but the relationships between individuals. We see this frequently in team sports. The team with the best individual players can lose to a team of poorer players. The second team cannot rely on one or two stars but instead has to focus on creating outcomes which are beyond the talents of any one individual. They create outcomes based on the interrelationships between the players. This is not to dismiss individual excellence. It does suggest that individual abilities is not a complete explanation of success or failure. In management terms, it shifts the attention to focus on the patterns of interrelationships and on the context of the issue, individual or group.

Min specs
Generative relationships

CAS have distributed control rather than centralized control. Rather than having a command center which directs all of the agents, control is distributed throughout the system. In a school of fish, there is no 'boss' which directs the other fishes' behavior. The independent agents (or fish) have the capacity to learn new strategies and adaptive techniques. The coherence of a CAS' behavior relates to the interrelationships between the agents. You cannot explain the outcomes or behavior of a CAS from a thorough understanding of all of the individual parts or agents. The school of fish reacts to a stimulus, for example the threat of a predator, faster than any individual fish can react. The school has capacities and attributes which are not explainable by the capacities and attributes of the individual agents. There is not one fish which is smarter than the others who is directing the school. If there was a smart 'boss' fish, this form of centralized control would result in a school of fish reacting at least as slow as the fastest fish could respond. Centralized control would slow down the school's capacity to react and adapt.

"Some people really want to stop controlling, but are afraid. Everywhere things are changing, creating high degrees of uncertainty and anxiety. And the more anxious you are, the more in control you need to be. Making all this even worse, we've bought into the myth that leaders have all the answers. Managers who accept this myth have their levels of anxiety ratcheted up again. ...If complexity theory can begin freeing managers from this myth of control, I think you'll see people a whole lot more comfortable."
Linda Rusch
Vice President of Patient Care
Hunterdon Medical Center
New Jersey

Distributed control means that the outcomes of a complex adaptive system emerge from a process of self-organization rather than being designed and controlled externally or by a centralized body. The emergence is a result of the patterns of interrelationships between the agents. Emergence suggests unpredictability - an inability to state precisely how a system will evolve.

Rather than trying to predict the specific outcome of emergence, Stuart Kauffman suggests we think about fitness landscapes for CAS. A CAS or population of CAS are seen to be higher on the fitness landscape when they have learned better strategies to adapt and co-evolve with their environment. Being on a peak in a fitness landscape indicates greater success. However, the fitness landscape itself is not fixed - it is shifting and evolving. Hence a CAS needs to be continuously learning new strategies. The pattern one is trying to master is the adaptive walk or capacity of a CAS to move on fitness landscapes towards higher, more secure positions.

A Complex Way
Emerges from the fabric

At Home

The co-evolution of a CAS and its environment is difficult to map because it is non-linear. Linearity implies that the size of the change is correlated with the magnitude of the input to the system. A small input will have a small effect and a large input will have a large effect in a linear system. A CAS is a non-linear system. The size of the outcome may not be correlated to the size of the input. A large push to the system may not move it at all. In many non-linear systems, you cannot accurately predict the effect of the change by the size of the input to the system.

Weather systems are often cited as examples of this phenomenon of nonlinearity. The butterfly effect, a term coined by meteorologist Edward Lorenz, is created, in part, by the huge number of non-linear interactions in weather. The butterfly effect suggests that sometimes a seemingly insignificant difference can make a huge impact. Lorenz found that in simulated weather forecasting, two almost identical simulations could result in radically different weather patterns. A very tiny change to the initial variables, metaphorically something as small as a butterfly flapping its wings, can radically alter the outcome. The weather system is very sensitive to the initial conditions or to its history.

An example in an organizational setting of non-linearity is the huge effort put into a staff retreat or strategic planning exercise where everything stays the same after the 'big push'. In contrast, there are many examples of one small whisper of gossip - one small push - which creates a radical and rapid change in organizations.

Non-linearity, distributed control and independent agents create conditions for perpetual novelty and innovation. CAS learn new strategies from experience. Their unique history helps shape the path they take. Newtonian science is ahistorical - the resting point or attractor of the system is independent of its history. This is the basis of neo-classical economics and is the antithesis of complexity.

Min specs

Increasing Returns



Complex adaptive systems are history dependent. They are shaped and influenced by where they have been. This may seem obvious and trivial. But much of our traditional science and management theory ignore this point. What is good in one context, makes sense in all contexts. Marketers talk about rolling out programs that were effective in one place and hence should be effective in all. In traditional neo-classical economics, there is an assumption of equifinality - it does not matter where the system has come from, it will head towards the equilibrium point. Outliers or minor differences in the starting point or history of the system are ignored. The outlier or difference from the normal pattern is assumed to be dampened and hence a 'blip' is not important. Brian Arthur's work in economics has radically altered this viewpoint. For example, he cites evidence of small differences fundamentally altering the shape of an industry. The differences are not always dampened but may indeed grow to reshape the whole. Lorenz referred to this in meteorology as sensitive dependence to initial conditions which was discussed earlier as the butterfly effect. In economics, in nature, in weather and in human organizations, we see many examples where understanding history is key to understanding the current position and potential movement of a CAS.

Wicked questions

CAS are naturally drawn to attractors. In Newtonian science, an attractor can be the resting point for a pendulum. Unlike traditional attractors in Newtonian science which are a fixed point or repeated rhythm, the attractors for a CAS may be strange because they may have an overall shape and boundaries but one cannot predict exactly how or where the shape will form. They are formed in part by non-linear interactions. The attractor is a pattern or area that draws the energy of the system to it. It is a boundary of behavior for the system. The system will operate within this boundary, but at a local level - we cannot predict where the system will be within this overall attractor.

A complex way
Emerges from the fabric

Tune to edge

A dominant theme in the change management literature is how to overcome resistance to change. Using the concept of attractors, the idea of change is flipped to look at sources of attraction. In other words, to use the natural energy of the system rather than to fight against it. The non-linearity property of a CAS means that attractors may not be the biggest most obvious issues. Looking for the subtle attractors becomes a new challenge for managers.

"In the past, when managers have tried to implement change, they'd find themselves wasting energy fighting off resistors who felt threatened.  Complexity science suggests that we can create small, non-threatening changes that attract people, instead of implementing large-scale change that excites resistence.  We work with the attractors"

Mary Anne Keyes, R.N.
Vice President, Patient Care
Muhlenberg Regional Mediacal Center
Plainfield, NJ

CAS thrive in an area of bounded instability on the border or edge of chaos. In this region, there is not enough stability to have repetition or prediction, but not enough instability to create anarchy or to disperse the system. Life for a CAS is a dance on the border between death by equilibrium or death by dissipation. In organ- izational settings, this is a region of highly creative energy.


Next | Previous | Return to Contents List

All Components of Edgeware Primer Copyright 2000, Brenda J. Zimmerman.
Schulich School of Business, York University, Toronto, Canada.
Permission to copy for educational purposes only. All other rights reserved.