Organizations as Complex Adaptive Systems: Insights for Managing Emergencies
Introduction
I started this post over a year ago but just got around to finishing it now. It remains an incomplete account of complex adaptive systems (CAS) but may suffice as an introduction to the topic. It fits in nicely with the posts that came before it on viewing organizations as systems.
The last post in the series identified the organizations we belong to as systems. Systems were described as wholes that are greater than the sum of their parts because through their interactions with one another the parts produce emergent properties that none of the parts could create by themselves or by adding them up. Some emergent properties belonging to organizations were discussed including behavior, outputs to external stakeholders, accomplishments, qualities, and the overall defining characteristics of an organization. This post continues the discussion of organizations as systems by defining the type of system they are as complex adaptive systems.
The Theory
In building towards a sketch of CAS, a variety of sources will be drawn from including the work of Cilliers, Holland, Waldrop, Stacey, Griffin, Shaw, and others. All of these authors have already done the work of developing and synthesizing the even more fundamental theory underlying CAS. Across the literature, a number of defining characteristics of CASs are provided. In what follows, a selection of these characteristics are presented from which a rough understanding of CASs can emerge (Cilliers, 2010). While not exhaustive, the emphasized characteristics are intended to provide a working literacy substantive enough to apply CAS theory to organizations.
Agents: Holland (2014a) defines CAS as systems "composed of elements, called agents, that learn or adapt in response to interactions with other agents" (p.24). Elsewhere, Holland (2014b) suggests CAS characteristically consist of a "multitude of agents" (p.57) that are "diverse rather than standardized" (Holland, 2014, p.57). In doing so, he departs with other CAS thinkers who suggest it is not the number of agents but the richness of the interaction among them that is characteristic of a CAS (Cilliers, 1998). The table below contains examples of systems and the agents that compose them.
Staying with the example of organizations, the system is the organization (e.g., ICS, emergency management department, fire department) and the agents within them are members (e.g., firefighters, EOC personnel, command staff and resources).
Multiple Diverse Agents As a CAS's diverse number of agents act in parallel with others, they find themselves in an "environment produced by [their] interactions with the other agents in the system" (Waldrop, 1992, p.145). Agents are "constantly acting and reacting to what other agents are doing. And because of that, essentially nothing in [their] environment is fixed" (Waldrop. 1992. p.145). To explore multiple diverse agents, Holland (2014b) uses the example of a rainforest and discusses how the various species "act much like a stack of catch basins with overflow spouts" (Holland, 2014, p.12). Meaning, each living thing in the rainforest uses resources that are passed on from others. For an example closer to organizations, Holland (2014b) connects the diversity of agents to assembly lines where multiple specialists act in sequence to complete a product. Each assembly line worker performs their specialized function as the product passes their location.
Self Organization: In a CAS, control is distributed rather than concentrated. With no central authority unilaterally directing the system, coherent behavior arises through processes of self-organization as agents compete or collaborate with one another (Holland, 2014a; Waldrop, 1992). Cilliers (1998) defines self-organization as an emergent property of complex adaptive systems which "enables them to develop or change internal structure spontaneously and adaptively in order to cope with, or manipulate, their environment" (p.90). Here, Cilliers defines structure as the "internal mechanism developed by the system to receive, encode, transform and store information on the one hand, and to react to such information by some form of output on the other" (p.89). He elaborates on structure further elsewhere and refers to it as a network that joins the nodes, or agents, together (Cilliers, 2006). Together, the previous quotes provide an image of the structure of CAS as a network that forms connections between the agents in a CAS and serves the purpose of processing and storing incoming information. As it is encountered in any instant, the structure of a CAS "is a result of the sedimented history of the system" (Cilliers, 2006, p.106).
While it involves the structure of the system, self-organization begins with agents taking action based on the information available to them about the system they are in and the surrounding environment (Cilliers, 1998). To illustrate this point, one of the examples used by Cilliers is a school of fish. Cilliers writes that the condition of fish in the school, or system, depends on factors including the time of year and the availability of food, oxygen, and light. He explains that as these conditions change, the size of the school of fish adjusts itself as needed to "ensure the best match between the system and its environment" (Cilliers, 1998, 89). Cilliers explains there is no agent, or fish, in the system who decides what action the school of fish should take, "nor does each individual fish understand the complexity of the situation" (p.90). This is self-organization in action: Agents taking action that produces global coherent patterns (Stacey, Griffin & Shaw, 2000; Waldrop, 1992).
Consider the role of structure in the school of fish example of self-organization. At any moment, the size and type of the school of fish and their configuration represents structure. This structure is a product of the past experience of the school that has led to the school's present size and arrangement. Past events triggered adaptations by the agents that in turn changed structure as a means of coping with a fluid environment leading to the arrival of structure as it is encountered in the present. At any given instant, the structure of the school of fish determines how information is received, acted upon, and internalized by the system. Wheatley and Kellner-Rogers (1996) write:
All living systems have the capacity to self-organize, to sustain themselves and move toward greater complexity and order as needed. They can respond intelligently to the need for change. They organize (and then reorganize) themselves into adaptive patterns and structures without any externally imposed plan or direction (p.18-19).
Adaptive: As their name indicates, CASs are adaptive, a characteristic Axelrod and Cohen (2000) suggest does not belong to all complex systems. They instead reserve the term for systems where "the strategies used by agents or a population change over time as the agents or population works for improved performance" (p.18.). To Axelrod and Cohen (2000), adaptation does not necessarily need to be effective to count as such. In their understanding of adaptation, any "actions that may lead to improvement" (Axelrod & Cohen, 2000, p.9) are considered. Strategy, a concept used in CAS theory, is the "way an agent responds to its surroundings and pursues its goals" (p.4). Strategies generally take the form of: "If you encounter circumstances X, then do Y" (Axelrod & Cohen, 2000, p.141) where Y is believed to be worthwhile and likely to produce desirable results (Waldrop, 1992). Understood as the courses of action agents pursue, strategies are changed based not only upon the feedback received from carrying them out but the anticipated results of carrying out a certain strategy and conditions in the system and environment (Holland, 1992). Agents "keep the strategies that pay off well, and let others die out" (Waldrop, 1992, p.165). This is the core of adaptation in CAS.
"This ability of the parts to adapt or learn is the pivotal characteristic of complex adaptive systems. Some adaptive systems are quite simple: a thermostat adapts by turning the furnace on or off in an attempt to keep its surroundings at a constant temperature. However, the adaptive processes of interest here are complex because they involve many parts and widely varying individual criteria (analogous to the constant temperature sought by the thermostat) for what a "good outcome" would be" (Holland, 1992, p.19).
It is important to note that agents adapting to one another does not necessarily indicate adaptation at the level of the whole. For example, two coworkers adapting their strategies to one another does not in all cases lead to changed performance or properties at the level of the whole. However, the accumulation of adaptations by agents can lead to system-level adaptation (Axelrod and Cohen, 2000; Stacey, Griffin & Shaw, 2000).
Openness: CAS are open systems. To Cilliers (1998), the openness of CAS manifests itself as a difficulty in determining what is inside the system and what it is in its environment as the system is constantly interacting with its surroundings. In addition to the difficulty of being able to discern where a boundary that constitutes a CAS should be drawn, openness also refers to the capacity of the system to receive signals, resources, and information from its environment and adapt and self-organize as needed. In other words, CAS are not "closed off" from their environments. As a part of this openness, CAS exist in a state "characterized by continual flow and change" (Capra & Luisi, 2014, p.86). CAS are constantly on the move.
Perpetual Novelty: As part of being constantly on the move, CAS are always evolving and innovating (Holland, 1992). In a CAS, "each agent finds itself in an environment produced by its interactions with other agents in the system. It is constantly acting and reacting to what the other agents are doing" (Waldrop, 1992, p.144). In other words, agents find themselves immersed in an environment of constant change meriting further adaptation and opportunity for interaction with new agents. As this is the case, the idea of defining an optimum for a system, agent, or group of agents becomes problematic. At the agent or agents level, an optimum may exist in a local area within a system but it is likely to be short-lived as the environment of these agents is in constant flux. Therefore, the environment that serves as the reference point for evaluating the strategies of a group of agents is everchanging so optimal is a moment-to-moment measurement.
"It is the process of becoming, rather than the never-reached end points, that we must study if we are to gain insight" (Holland, 1992, p.20).
Innovative: CAS are also innovative. As new agents appear in the system, new opportunities for interaction and adaptations become possible. Holland (2014a) writes that "a single new agent can be at once prey, a partner for exchange, and a parasite" (p.58). As they are adaptive, agents continuously explore this space of possible interactions and adaptations (Holland, 2014a).
Emergent Properties: Emergent properties were discussed in the last post as the outcome of diverse agents coming together to produce system-level properties none of the agents possess on their own or could create by simply adding them up. Rather, emergent properties are created through the interaction of the agents. This phenomenon is usually expressed in the phrase "the whole is more than the sum of their parts." Alluded to in the last post was the relationship of emergence and hierarchical organization. Holland (2014a; 2014b) indicates that multiple agents come together in a CAS to form an aggregate. Consider the example below.
"A group of individual workers will compose a department, a group of departments will compose a division, and so on through companies" (Waldrop, 1992, p.145-146).
The aggregates formed at one level display emergent properties that form building blocks for the next level in the hierarchy, contributing to emergent properties above the level where they first occurred.
Nonlinear: Interactions among agents as well as agents and the environment are nonlinear. Cilliers (2008) writes that "non-linearity...guarantees that small causes can have large results, and vice versa. It is a precondition for complexity" (p.4). Nonlinearity specifies interactions where causes create effects that may be distant in time and space and produce effects that are smaller or larger than the initial cause (Cilliers, 2008; Meadows, 2008). Nonlinear relationships of cause and effect make CAS full of surprises.
The Insights
If we see organizations as complex adaptive systems, adaptation is the rule and not the exception. Consistent with the earlier posts in this series, one of the key insights found here is the capacity for organizations to adapt. If conceptualized as CAS, organizations appear as collections of agents operating in continually fluctuating environments due to the results of their own adaptations as well as those of other agents and environmental changes. What is observed within the system emerges through the interactions of agents self-organizing and adapting to one another.
The capacity to adapt can be suppressed or enabled depending on the degree of constraint within the system. As Morin (1992) describes it, organizations can be highly constrained and as a result, rigidly ordered or loosely constrained to allow for disorder. Disorder in this sense refers to variability and deviation from a set plan. Adaptation can be seen as disorder that an organization can allow for through the constraints they design.
"Disorder refers to everything that is irregularity, deviation as regards to structure, random, unpredictability (Morin, 2008, p.63).
"Order refers to everything that is repetition, constant, invariant, everything that can be put under the aegis of highly probable relation, framed within the dependence of a law" (Morin, 2008, p.62).
One of the earlier posts discussed Morin's (2008) notions of programs and strategies. Strategies were introduced as an adaptive approach to planning that allows for the continual incorporation of new information and cancelation if found to be ineffective. Strategies are, then, a designed constraint that can allow for adaptation by agents. However, in order for this designed constraint to enable adaptation as intended, there will need to be a corresponding shift in management style that can deal effectively with the constructive disorder introduced by adaptation. This same insight can be applied to self-organization and perpetual novelty. To gain the advantages offered by both CAS phenomena, management must shift away from command and control and towards a systems-based method that enables organizations to act like CAS. Of course, there may be moments where command and control leadership are contextually appropriate (Snowden & Boone, 2007). For example, the initial operational period where unknowns are innumerable but expedient life-saving action is needed may require the tightening of constraints through command and control. However, and to borrow insight from Cynefin, the constraints can be loosened and a CAS can remerge when appropriate (Kurtz & Snowden, 2003).
CAS are in a constant state of becoming. Optimal conditions are ephemeral. As CAS are open to exchanges with their environment and the system's internal environment is continually evolving, CAS exist in a state of constant flux (Capra & Luisi, 2014). Because of this constant change, it is impossible to specify when a CAS has achieved optimal conditions. New agents may enter the system opening opportunities for new kinds of unanticipated exchanges that may lead to a new briefly sustained optimum set of conditions within the system. The same could be said for changing variables in the environment. There is no optimal rainforest, school of fish, or incident response. The field is always shifting, and as a result what it means to be "optimal" is always changing as well.
Holland (1992) explains "Though some parts of the system may settle down temporarily at a local optimum, they are usually 'dead' or uninteresting if they remain at that equilibrium for an extended period" (p.20). Understood as a CAS, an organization might be held at a determined local optimum through constraints (Morin, 1992). Holland (1992) points out this leads to stagnant uninteresting behavior. CAS need to be able to live constantly on the move. This does not mean that the management of organizations as CAS is rudderless or without foresight. As Dave Snowden often says, when it comes to CAS vector is managed, not velocity (Snowden, 2013). What is of concern to management and leadership within a CAS is the direction of travel of the system with less of a focus on how it gets there. Indelible images of how things should be done constrain the system to a fleeting optimum impeding the system's capacity to continue evolving through adaptation, innovation, and self-organization.
References
Axelrod, R., & Cohen, M. D. (2000). Harnessing complexity: Organizational implications of a scientific frontier. New York, NY: Basic Books.
Cilliers, P. (1998). Complexity & postmodernism: Understanding complex systems. London: Routledge.
Cilliers, P. (2006). On the importance of a certain slowness. Emergence: Complexity and Organization, 8(3), 105-112.
Cilliers, P. (2010). The boundaries of complexity, the limits of systems . International Council on Systems Engineering.
Holland, J. H. (1992). Complex adaptive systems. Daedalus, 121(1), 17-30.
Holland, J. H. (2014). Complexity: A very short introduction. Oxford, England: Oxford University Press.
Holland, J. H. (2014). Signals and boundaries: Building blocks for complex adaptive systems. Cambridge, MA: MIT Press.
Kurtz, C. F., & Snowden, D. J. (2003). The new dynamics of strategy: Sense-making in a complex and complicated. IBM Systems Journal, 42(3), 462-483.
Meadows, D. H. (2008). Thinking in systems: A primer. (D. Wright, Ed.) White River Junction, VT: Chelsea Green Publishing.
Morin , E. (2008). On complexity. (R. Postel, Trans.) Cresskill, NJ: Hampton Press.
Morin, E. (1992). Method: toward a study of humankind: the nature of nature (Vol. 1). (B. J.L Roland, Trans.) New York, NY: Peter Lang.
Snowden, D. (2013, February 26). An interview with Dave Snowden [Audio podcast]. The systemic insight podcast . (M. Jental, Interviewer) Retrieved from http://systemicinsight.libsyn.com/an-interview-with-dave-snowden
Snowden, D. J., & Boone, M. E. (2007). A leader's framework for decision making. Harvard Business Review, 85(11), 68-76.
Stacey, R. D., Griffin, D., & Shaw, P. (2000). Complexity and management: Fad or radical change to systems thinking. London, UK: Routledge.
Waldrop, M. M. (1992). The emerging science at the edge of order and chaos. New York, NY: Touchstone.
Wheatley, M. J., & Kellner-Rogers, M. (1996). Self-organization: The irresistible future of organizing. Strategy & Leadership, 24(4), 18-24.