May 18-19, 1995
Gus A. Koehler, PhD.
L. Douglas Kiel, Ph.D.
University of Texas at Dallas121
Chaos prevents a stable strategy of problem solving. Klaus Mainzer, 1994.
To the uninitiated, chaos theory often lurks as a mathematical and scientific hinterland, of value only to a small cadre of theorists. Yet, a vast number of applied disciplines are now exploring chaos theory as a means for understanding and building systems that utilize the potentialities of this new approach. From neuroscience (Bower, 1988), to cardiology (Garfinkel, et. al., 1992) to business (Priesmeyer, 1992; Goldstein, 1989) and public management (Kiel, 1993; 1994) researchers are developing a new intellectual paradigm that offers the insights of chaos theory as a new vision for understanding various aspects of our world. In short, a new paradigm of social and human dynamics is emerging.
This new appreciation for chaos has led to an understanding of both the nonlinearity of the world in which we live and of the functional aspects of instability as a means for adapting to new situations. Chaos is one possible result of the dynamics of nonlinear systems. Nonlinearity refers to behavior in which the relationships between variables in a system are dynamic and disproportionate. In nonlinear systems small changes or small errors can have big effects. And, in nonlinear systems outcomes are subject to high levels of uncertainty and unpredictability. In nonlinear systems behavior is erratic and filled with surprises. Our world is filled with nonlinearity.
Disaster and emergency situations epitomize the nonlinearity of human events. These are events in which the relationships between relevant variables is churning. Even in our desire to create order and control the situation, events often seem to churn one step ahead of our best efforts. Heinz Pagels (1988, p.56) noted that, "life is.. nonlinear. And so is everything else of interest." Clearly, what makes disaster situations particularly interesting and challenging is the inherent nonlinearity in such events.
New thinking in response to the recognition of nonlinearity in human and organizational systems has focused on the functionality of disorder and instability (Kiel, 1994). Management scholar Ikijiru Nonaka (1988, p.59) offers a view of the functionality of disorder and instability in organizations "Chaos widens the spectrum of options and forces the organization to seek new points of view. For an organization to renew itself, it must keep itself in a non-equilibrium state at all times."
Most importantly, during times of high instability such as disasters and occasions when emergency services reach peak levels of activity it is essential to recognize that stability can only be regained by developing strategies that are themselves unstable. In short, we must match the instability of these environments with management practices and organizational strategies that are dynamic and fluid. This paper focuses on developing such instability in administrative systems.
If Disasters Were Simple Events: Linear Versus Nonlinear Behavior
The best way to understand how disaster and emergency events are nonlinear systems is to compare the behavior of such systems with that of linear or simple systems. In linear systems the relationships between relevant variables is stable. In linear systems the relationship between cause and effect is smooth and proportionate. In short, linear systems respond to big changes in a big and proportionate manner and linear systems respond to small changes in an equally small and proportionate way.
If disaster and emergency response processes were linear system we could predict the number of fatalities or the amount of resources and personnel required to bring order to chaos. We could predict how long reconstitution of the previous environment would take. We could make statements such as an 8.5 earthquake centered on Ventura avenue would kill exactly X number of people. We often make these linear estimates because we are limited by linear tools for prediction and response.
But when we look at real disasters do we see such prediction, such simplicity, or such linearity? The potential for nonlinearity and erratic behavior to occur in complex human environments emphasizes the overly simplistic assumptions we often make about system behavior and real outcomes. In 1992, the city of Chicago suffered a devastating downtown flood due to a failure in the city's tunnel wall. Only later was it discovered that a private contractor tried to report the failure but no city authority responded to the report. A crack that could have been remedied for $10,000 eventually cost taxpayers, the city, and business an estimated 1.7 billion dollars (Roeser, 1992). The nonlinear and explosive effect of a seemingly small crack led to real disaster. In disaster and emergency services management the outcomes of our errors, oversights, and even our best intentions may only, much later, result in real and unexpected surprises.
Consider how as a nonlinear system evolves over time we cannot predict all of the consequences of what seem, initially, to be totally reasonable management decisions. An example of our limited ability to predict all of the interactions in a complex and nonlinear world were the severe floods along the Mississippi river in the summer of 1993. For several decades prior to this flooding, the Army Corps of Engineers built a series of levees to protect many river front communities. These levees were seen as solutions to local flooding problems. These levees, however, served to change the course of the Mississippi in many areas. Some analysts now believe that these levees, and the decision decades earlier to build them, actually exacerbated downstream flooding that adversely affected many riverfront communities (Burton and Gibson, 1993). The seemingly simple decisions to aid individual communities with levees led to a tangled web of cause and effect that, over time, had disastrous results many years later for other communities. Clearly, the world of public management is a world in which managers cannot predict how future observers will judge the quality of their decisions.
Nonlinear Change: Equilibrium to Rhythms to Chaos
To fully understand the paradigm of nonlinear dynamics we need to examine the types of behavior that nonlinear systems generate. For our purposes, behavior refers to how change occurs in organizations and how organizational data evolve over time. One tenet of nonlinear dynamics is that complex systems defy simple formulation and thus may preclude the development of precise mathematical algorithms. As a response, students of nonlinear dynamics show a preference for graphic representations of data, behavior, and systems. The following pages of graphs typify this approach. Such graphic representation is consistent with the trend in public management toward the visual display of information. Graphs and pictures can tell very important stories for managers.
Nonlinear systems exhibit three distinct types of behavior over time. These behaviors are labeled as (1) convergence to stability or equilibrium; (2) stable oscillation; and (3) chaotic. Each behavior can appear over the long term behavior of a nonlinear system. Disasters, emergency response, and changes in quality all occur over time. Thus, in real work processes, each behavioral type does not reflect permanent commitment only to that behavior, because the real world generates many different patterns in the data organizations create. Equilibrium in Time
We can examine the different kinds of time series nonlinear dynamical systems generate by using a simple nonlinear equation and the help of line graphs. The logistic equation is an oft-used algorithm for generating nonlinear time series. The logistic map takes the form xt+1 = (wxt) * (1-xt). The variable w represents a control parameter in the range 0 < w < 4. The variable xt is the value of x at the current time and xt+1 represents the next time period following xt.
The most simple type of time-based behavior generated by nonlinear systems is convergence to a stability or equilibrium. This behavior occurs when we start from an initial point that quickly reaches and maintains a mathematically stable point (see figure 1). This behavior represents the ultimate equilibrium, where change does not occur over an extended period. As one can see, once the mathematical point of stability is reached the system remains stationary, even if the time series is extended indefinitely.
In a nonlinear world, one must wonder how many work or organizational systems will show such extremely stable behavior over time? Even the most stable work outputs, such as fire station equipment inspections, show some variation in output from month to month. Yet, at an abstract level the reader will see that the equilibrium in figure 1 is the Weberian ideal. In short, the work output is perfectly stable. In such a case, management could predict output perfectly because management knows exactly what to expect on a consistent basis. If we view figure 1 as output from any organizational process we can see it represents the Weberian ideal of organizational stability represented by a time series! The volatility and dynamism of the real world is gone and the machine-like bureaucracy marches on. But, can we really expect such stability in the real world? Can we think of any organizational output that is this stable over time? Or, as Cavaleri and Obloj (1993, p.57), note, "The behavior of virtually all systems important to organizations varies over time and does not follow a straight-line pattern". As Drabek (1994, p. 30) has also noted, "Disasters do not constitute a simple straight line extension of an auto accident or house fire." Rhythms in Time
A second type of nonlinear time series that can occur in the real world of organizational data is rhythmic or oscillatory behavior. This type of behavior is generally labeled as stable oscillation because work output, such as, service responses to citizens calls shift fluidly up and down in a patterned and stable fashion (see figure 2). This type of smooth change is incremental change that moves up and down in a predictable manner.
The time series is figure 2 is called a two-period cycle. This is because the cycle repeats itself every two time periods or every two data points; the cycle stabilizes at about point 20. Such periodic, or cyclical, time series can have varying periods such as 4, 6, or 8 periods before the cycle repeats itself. So rhythmic data can have lots of short little cycles or big, longer cycles.
One can imagine many agency and organizational systems relevant to emergency management that operate in such a cyclical manner. In fact we use this kind of language in public management. For example, calls for local emergency medical services are generally cyclical and rhythmic: messy and noisy, but rhythmic and continuous.
Consider the cycle of local emergency calls in any large city. Emergency calls peak during weekend evenings as the multiplying interactions of alcohol and automobiles collide. During other hours of the week calls decrease. The up and down cycles of traffic also include another rhythm, the rhythm of the work week. Weekend traffic has a different pattern than does workday traffic. Of course these patterns create other patterns. The pattern of police response activity to traffic accidents is also determined by the cycles of the "rush" hour. Chaos Over Time
A third type of nonlinear time series data that can be expected in the world of management is chaotic behavior. Priesmeyer (1992) has shown how chaos appears in organizational systems ranging from financial management data to data from production processes. Chaos is typified by behavior that, over time, appears random and disorderly (see figure 3). Chaos does, however, occur within definable parameters or mathematical boundaries. Thus chaotic behavior remains within boundaries or within limits. It is not random behavior that can result in any outcome. Chaotic behavior looks wild and erratic, but does not jump out defined mathematical limits.
When chaos occurs a nonlinear system does not retrace prior identifiable sequences of behavior and does not evidence obvious patterns in its behavior. Chaotic behavior thus appears extremely disorderly since patterns over time, a symbol of orderliness, do not appear to exist. Chaotic behavior simply skips from one identifiable point to the next, yet never extends outside clear and distinct boundaries. The reader will note that the data points in figure 3 do not extend beyond 0 or 1. Chaos thus looks like random behavior but is really unstable behavior over time that stays within clear boundaries.
Although chaotic time paths may look random they are generated by deterministic and rather simple mathematics. Thus the kind of chaos we see in figure 3 is referred to as "deterministic chaos". It appears that such deterministic chaos can be created by organizational systems and processes that are intended to be very mechanical and simple. Researchers have discovered both such deterministic chaos in organizational data (Priesmeyer, 1992) and the potential for this chaos in organizations (Mosekilde, et.al., 1991; Richards, 1990). This has two very important consequences. First, this means that work systems or processes with few parts and simple interactions can generate very complex data, that look erratic and chaotic over time. For managers, the effort to simplify processes may result in unexpected complexity. Second, if simple systems can generate complex behavior then imagine what may result when considering the complex organizations and environments that disaster and emergency services managers attempt to handle.
A word of caution is necessary at this point. To verify the existence of real mathematical chaos in organizational data requires the use of some very sophisticated statistical methods. So analysts need to be careful not to call all "messy" looking time series data as chaotic. Time series data may be nonlinear but not chaotic. So we must be sure to note when we are discussing real verifiable chaos or chaos as a metaphor. Surely, both approaches can help us attain the vision of the paradigm of nonlinear dynamics.
We can begin to see that as the relationships between the parts of nonlinear systems change, these systems can create choppy, or even erratic behavior over time. Nonlinear systems bounce around and can be quite messy. Nonlinear systems create graphs with lots of breaks and changes and with lots of ups and downs. The messiness and ups and downs of nonlinear systems clearly reminds us of the data public managers examine over days, weeks, and months. Employee performance goes up and down over time. Budgets go up and down over time. The erratic time-based data that organizations generate is a result of our nonlinear world. Sensitivity to Initial Conditions
Systems functioning in chaotic regimes may show a tendency to be highly sensitive to their initial conditions. This means that small changes or errors can have amplified effects. The point is reinforced if we consider the concept of the "butterfly effect". The butterfly effect can be better understood by examining the two different time series in figure 4. These time series use the same nonlinear equation as that used to generate figures 1-3. The time series in figure 4 start at mathematical points that differ by only one-ten millionth (the last decimal place). Yet, look at how these lines quickly separate, diverge, and create very different results, that continue over time. We see that a small difference, a small error, or a small change can have very novel, unexpected and even explosive effects over time.
Think back to the scenario of the federal alcohol, tobacco, and firearms raid on the Branch Davidians. We can now see the unexpected "tip-off" of the cultists as the "butterfly" that sent the government raid off in a totally unexpected direction. Of course, a multitude of possible results could have occurred. Look to figure 4 and imagine how things may have turned out differently if the tip-off had not occurred. Without the tip-off another totally different set of results may have occurred. Attractors - The Order in Chaos
One of the interesting qualities of nonlinear dynamics as a paradigm for emergency management is that even when the data we examine look erratic and chaotic we can find a deeper order in the data. By looking at this deeper order in organizational data, managers can find both a new means for understanding how much change exists in organizational output and performance, but can also begin to see how much effort will be needed to change and improve the performance and results of work processes in organizations.
To examine the "order within chaos" in organizational data we can examine the "attractor" of time series data. An attractor is a graphical method chaos researchers use to determine how much change is occurring in a set of data overtime. The attractor presents an image of all of the change in the data that work process data or employee performance data generate. By viewing this attractor we can begin to see the unusual and unique forms of order that the data in organizations create.
To look at time-based data managers usually examine line graphs. An attractor is a different way of viewing time series data. An attractor is a mapping of data that allows us to see how all the data, be it work group output or individual employee output, relate to each other. These figures are called attractors because the data seem to be "attracted" to certain regions on the graph. While line graphs show us how each element of data changes relative to the data point behind it and in front of it, an attractor mapping shows us how all of the data we are examining change relative to each other. The attractor can show managers how much variation and change is occurring in organizational data over time.
The attractor is graphed in a t/t+1 phase plane. In this case, t (time) represents the data at one point in time, while t+1 represents the same data at the next point in time. The t is plotted on the vertical axis and t+1 is plotted on the horizontal axis as shown in figure 5. The attractor in figure 5 is derived from a chaotic time series generated by the logistic equation. For another method for graphing attractors see Priesmeyer (1992). The Functions of Chaos
Conventional systems thinking has focused to an extreme extent on actions necessary to stabilize systems. While instability was recognized, previous systems thinkers saw a return to stability as the only real alternative. This focus on stability thus minimized efforts to examine the positive aspects of periods of instability or even chaos. Scientific investigation of chaotic behavior in a variety of natural and human systems over the past two decades now reveals that chaotic episodes are actually highly functional aspects of system evolution. Chaos serves at least two evolutionary functions.
First, chaos serves to avoid entrainment or mode lock-in. Entrainment refers to system behavior that maintains itself even when environmental change or internal demands suggest new behavioral changes for survival and adaptation. Examples of mode lock-in are seen in corporations that refuse to alter products or services when markets and customer demands have changed. Mode lock-in is seen when employees fight and resist potential labor saving improvements in an effort to maintain existing methods and processes. Chaotic episodes offer the potential to break entrainment and offer new forms of behavior and functioning. Chaos helps to break existing molds.
Second, chaos allows relevant systems to explore the entire range of behaviors available to them. This is because systems in chaotic phases bounce around the phase plane exploring their every possibility for new and alternative behavior. The range of available options is, of course, bounded by the parameters of the system. It is this erratic behavior though that creates the uncertainty and unpredictability typical of chaotic periods. It is however, uncertainty and unpredictability that are essential aspects in the mechanisms of learning. Learning comes from the instability of uncertainty and is evidenced by new forms of behavior and response.
This second function of chaos emphasizes that chaos is fundamentally a learning mechanism. This is a learning mechanism that allows systems to test their evolutionary potential. The unfortunate reality that emanates from the chaos of a disaster situation is that weal ways learn much. We learn about the capabilities of existing disaster response system, the capacity of humans to survive exceedingly traumatic events, and the capacity of renewal in both natural and human systems. Controlling Chaos
Understanding the functions of chaos reveals that chaos represents both risk and opportunity. The risk of chaos is that a system may not reach another point of stability and thus be overwhelmed by constant uncertainty and instability. The opportunity of chaos is that new ways of behaving and responding to environmental challenges may be developed and become essential elements of emergent ways of responding to an uncertain world.
Since continuous chaos, particularly during natural disasters, would overwhelm our capacity to bring any level of even minimal sustainable livability to our populations methods for controlling chaos and bringing some order to this disorder is necessary. Natural scientists have for the last several years examined methods for controlling chaos (Ott, Grebogi and Yorke, 1990). These efforts have resulted in three fundamental methods for controlling chaos.
One method for controlling chaos is to alter the parameters of the system. This means limiting the degrees of freedom or the extent of the behavior available to a system. In short, by clamping down on the parameters of behavior, the hope is to alter behavior and create greater stability and predictability. In a disaster situation these degrees of freedom are often beyond the control of human actors. As an earthquake travels up the logarithmic Richter scale, the degrees of freedom in potential damage expand making control an increasingly difficult endeavor. The parameters of destruction may outstrip response system capacity.
A second method for controlling chaos uses "perturbations" or disturbances during chaotic episodes to change behavior back to more predictable and smoother functioning. This refers to the sensitivity of chaos to small changes. The intent with such interventions is to use small change that create nonlinear effects, that create phase shifts from erratic behavior to more fluid behavior.
A third and most recent method for controlling chaos is aimed at altering the "orbit" of a chaotic to system to a more desirable orbit on its attractor (Ditto and Pecora, 1993). This approach uses continuous tracking and seeks to identify changes in system behavior that occur over time. By tracking such changes alteration of the parameters are expedited. This approach is similar to the first method noted above, but this third approach represents a more adaptive or cybernetic approach to controlling chaos.
Each of the above methods for controlling chaos reveals analogs for management practice. Efforts to control system parameters best defines the dominant scientific management and Newtonian control model advanced by linear Western science. This model argues that placing strict controls on behavior of systems and people management is most likely to get levels of certainty and prediction that may result in management goal attainment. The most striking example of this management style is the paramilitary structure that still dominates many bureaucracies. The negative externalities of this approach to management are evidenced by inadequate response to changing circumstances due to mode lock-in, excessive and debilitating bureaucratic oversight, and demoralized employees.
The second approach to controlling chaos based on identifying pressure points to alter system behavior is synchronous with emerging views of managing organizations and inter-organizational response. This new view argues that recognition of the nonlinearity of human systems demands that we examine methods that use minimal pushes to develop maximum results. This approach focuses more on open lines of communication (Comfort, 1994) rather than controlling and dominating hierarchies.
The third approach for controlling chaos is consistent with cybernetic approaches to management. These approaches rely on constant feedback to ensure that work and administrative systems are continuously adjusting to environmental and organizational demands and changes. Again, we see the importance of communication and feedback in all efforts to control chaos. Comfort's work (1994) reveals the importance of modern information technologies to expedite this approach. A Dynamical View of Disasters
One of the advantages of viewing the world from the lens of chaos theory or nonlinear dynamics is that we begin to see that the world is actually filled with flux and change. The world is, from this perspective, an infinite array of time series of system all marching to their own unique drummer. While pattern and similarity occur the unique elements of each system's organization and environment represent that nuance that lead to unique evolutionary outcomes. Fields of Action
To better understand the relevance of nonlinear dynamics to the challenges of disaster response and management, and of improving the quality of outcomes, a brief introduction to the elements of dynamical systems theory is necessary. Dynamical systems are comprised of two elements. These elements are (a) the area or field on which the "action" or "motion" takes place (the formal label for this region is the "manifold of states") and (b) the set of rules that determine the motion or action in the field of action that lead to results. achieved in the field of action, (these rules are called "vector fields") (Casti, 1990, p. 54-55).
The field of action, or the "workplace", is determined by the nature of the work and the technology used to perform the work. Consider, for example, how the field of action for a classroom teacher is quite different than that of the more traditional office environment of an employee of the federal Bureau of Labor Statistics. What distinguishes emergency and disaster response professionals from other workers is that their "field of action" is both in the office (preparedness, waiting and watching) and in an external environment that may be changing as the disaster continues or as the fallout of the disaster continues over time. Disasters may also occur in unexpected areas. Thus even the field of action itself may be highly unpredictable for disaster response. Rules for Guiding Action
Workplace rules are elements such as policies, work processes, work behaviors, and employee attitudes that result in the actual outputs of the government work unit. Naturally, some guiding principles are intended to at least define the rules for the manager. Most observers assume the government manager simply seeks to optimize the use of resources to maximize the benefit to the taxpayer and the client of the agency.
But the rules can sometimes be vague and even work against each other. For example, in our efforts to be efficient, in time and resources for example, management may lower the effectiveness, or the ability, of the organization to reach its goals. By saving money we may decrease our ability to serve the citizenry. On the other hand, managers may find new ways to meet desired goals, only to find that such goal attainment will require more money and time to achieve.
The set of rules available to the emergency services manager is also not a completely open-ended set of possibilities. Public managers work within an environment of considerable constraints (Lerner and Wanat, 1992). Budget constraints dictate levels of agency service and response. Civil service regulations limit managements' ability to hire and fire. Requirements to save lives first impose certain (and essential) restrictions on how emergency services should be organized and rendered. And the ordinary limitations imposed by statutorily mandated policy and the following inevitable red-tape generate considerable administrative constraints.
We can see that these rules represent policies, processes, and the work behavior and attitudes of management and staff. It is the interaction of the rules of motion with field of action that determines the direction and result of the motion in the workplace. The dynamics created by the interaction of the "rules" and the "field of action" lead to agency response, outputs and performance.
From the manager's perspective, the task is to move to the desired position on the region of action. The manager must utilize some set of rules (or heuristics, formal or informal) to drive the agency. If a new space on the region of action is desired a new set of rules is generally required. Most importantly, the outputs in terms of performance and the outcomes in terms of policy results are the measures used to determine a manager's success in melding the "rules" to the "field of action". The Changing Nature of Workplace Rules
Public management thinkers have tried to identify the rules that drive agency actions. The classic effort in this area is Downs' (1967) comprehensive listing of laws and propositions intended to "help analysts forecast bureau behavior" (p. 261). Down's rules set out some general bureau and management behaviors that appear universal, such as managers will capture as much financial resources as possible and large bureaus are more resistant to change than small bureaus. These rules are intuitive and provide guidance as to how people will act in government agencies.
Yet, Downs also recognized that change is evident throughout government organizations. He also was aware of agency policy, in this sense policy for problem handling and decision-making, as sources of stability and change," a bureau can change its everyday actions without changing its rules; it can change its rules without shifting its rule-making structure; and it can alter its rule-making structure without adopting any different fundamental purposes" (1967, p.168). Yet, Downs does not note that the policy itself can serve as a source of change due to the dynamic environment in which they exist. In an administrative world in which in which chance can impede on seemingly fixed administrative and work process "rules", surprises are inevitable. Even when the "rules" appear clear and simple the nonlinear dynamics of the administrative and organizational world can generate real complexities and surprises.
It is clear that new rules are emerging in the workplace in public organizations. For example, the increasing calls for quality improvements in service provision demand new rules for achieving results and serving citizens. The expanding call for empowering employees demands new attitudinal and behavioral rules for both managers and employees. Management rules may require more "letting go" while employees must take more responsibility and initiative.
Most importantly, we begin to see that fixed policies may serve as inhibitions to positive change. If these fixed policies serve to entrain disaster and emergency response the necessary flexibility and instability for adaptive response is lost. Chaos, Variation, Learning, and Disaster Response
Chaos theory teaches us of the value of variation as a means for learning has obvious relevance to the management lens of nonlinear dynamics. In nonlinear systems it is the non-average behavior, the unusual event, the unexpected fluctuation that drives the processes of change. This suggests then, as with variation beyond the control limits in quality measurement systems, that variation should not be considered a problem in management systems but rather an opportunity to learn why the variation occurred. It is the peaks and valleys in quality measurement data that may provide the best source to improve administrative systems. Managers must learn to ask why output, problems, or performance peaked at one point and then reached a valley at another.
The ideal work system produces quality data that remains within the confines of upper and lower control limits. It is, however, not the normal variation that management should only be concerned with, but also those instances where "abnormal" variation occurs. Even if such abnormal variation cannot be controlled, management may learn much about work systems by investigating those external factors that alter work and outputs. Taking variation as an opportunity for learning means that managers do not get angry with such variance but instead see variation as a source of and for learning.
Understanding the value of variation for the public manager also emphasizes the value of instability in work and organizational systems. We know that unstable systems will generate more variation than stable systems. Chaos thus does not have to be seen as confusion. The functional aspect of chaos is learning, as systems and individuals are allowed to test their parameters of output, service, and quality.
Disaster and emergency service managers recognize the value of variation. This is evidenced in sessions to capture what was learned from each disaster. But variation can teach new lessons, not only from the unique nature of events, but also by trying new methods and techniques. This means that we need to examine new methods to ensure that we practice variation by perhaps using diverse teams to try diverse methods on essentially the same disaster or emergency problem. While patterns may exist in disasters, managers need to discover methods for breaking patterns of response in an effort to embed variation and the potential for learning. Work Teams, Instability, and Learning
The importance of work teams has long been recognized in management (Likert, 1961). This emphasis seems to currently be reaching a high pitch in the literature of public and business management. These expanding emphases are due both to an awareness that teams permit a larger number of potential inputs into decision-making and problem-solving and to the fact that teams may energize cohesive behavior that may enhance organizational effectiveness.
A work team with good "dynamics" thus can resolve many organizational issues. One of the expected lessons from this understanding is that finding a good work team is only slightly less important than keeping it together. This conventional wisdom asserts that a stable work team will promote learning as team members work together to develop new methods and adaptive innovations (Miller, 1993). The Value of Unstable Work Teams
One recent study in public management, however, emphasizes the value of instability in work teams (Miller, 1993). Gerald Miller (1993) conducted a study designed to determine levels of learning among work teams. The study simulated bidding on a government bond auction. Learning was determined by the number of errors the teams made in response to different incentives and different levels of initial funding. Several different work teams were involved in the study. Some teams were comprised of the same team members throughout the study. Other teams were unstable teams in which membership shifted as new members were installed.
Miller's results showed quite the opposite of the conventional wisdom. In all of Miller's experiments the stable teams showed either no learning or a rate of learning that was slower than the unstable groups. Unstable teams started with error rates that were initially higher than the stable groups, but quickly overtook the stable groups and produced fewer erroneous decisions. Thus, over time, unstable teams, learned significantly faster than stable teams. Miller concludes in his study that instability in work teams leads to "... a short-run performance decline and, yet, long-term effectiveness" (1993, p.57).
While Miller's study requires replication for solid generalization his efforts appear to promote the value of instability and variation in management systems. Apparently, the major inhibition to learning in stable teams is an initial acceptance and overly rigid view of a problem that minimizes possible solution sets. The new members of unstable teams appear to serve as "devil's advocates" that promote alternative decisions and thus divergent outcomes. This further suggests that unstable teams represent a more readily adaptive response to changing situations (Miller, 1993). Thus, instability in work team memberships promotes many qualities considered valuable in contemporary government organizations.
The value of instability, then, in work teams is that these "messy" teams serve the role of creating emerging structures consistent with the nonlinear notion of process structures. The team structure when thus viewed as a shifting and dynamic structure of human interaction generates an increased number of decision inputs and thus the likelihood of expanded alternative and solutions. At the same time this approach adds to the complexity of public management as managers may no longer rely on constancy in teams but rather must devote attention to shifting groups of teams. Freedom and Instability in Work Teams
Perhaps, most significantly, the notion of unstable teams is also consistent with the challenge of understanding the nonlinearities of work organizations themselves. Since so much of the interplay of work dynamics are beyond reasonable efforts at comprehension such shifting work teams provide management the opportunity to allow self-organizing teams to generate and explore their own parameters of learning. Such teams would be as chaos researcher Frank Ford notes, "...liberated to explore their every dynamical possibility" (Gleick, 1987, p.306). The usual constraints of public law and service will serve to minimize the total set of solutions available to a group.
Such an approach to team building also strikes at the heart of the quality management perspective. Quality management urges that employees be liberated to develop new methods and levels of client service. Such unstable teams provide a means for expanding the range of possible methods. This suggests that teams should be formed from employees across functional areas. This would be valuable as employees learn how other functional unites resolve their problems or provide new means for service delivery.
What work groups really represent is a potential source for generating new dynamics and creative fluctuations in organizations. Public managers must increasingly show a willingness to permit creative fluctuations to occur and infiltrate government organizations in hopes of generating nonlinear and explosive improvements in service quality. Furthermore, this approach emphasizes a new mode of management control. This mode of control is one of an evolving control that accepts instability as a necessary state for responding to the breaking of existing symmetries that hinder government performance and improvement.
On a practical level experiments may be conducted in emergency and disaster response that engage two activities. First, more cross-teaming preparedness between potential disaster groups is needed. This approach will allow more knowledge dissemination at both the organizational and individual level. Second, why not consider the use of unstable teams during disasters. What is learned by one team may be readily disseminated by simply altering the configuration of personnel rather than by using some perhaps overly specified organizational structure. Flattened Hierarchies: Disorder by Liberating Structures
Another prevailing theme among contemporary management thinkers for responding to rapid change and calls for continuous improvement is to alter the structure of organizations. This concern over organizational structure generates some rather predictable behavior on the part of public managers. When problems arise, often, the first culprit is the organization chart. This phenomenon seems particularly apparent among new senior managers whose first action often is to re-organize in hopes of improving the performance of an organization in decline or one that is considered underperforming. These administrative efforts, although sometimes without adequate cause, do create considerable turmoil and resistance. One can also argue that such efforts to change the organization chart are really an easy fix for real and deeper problems, such as inadequate attention to process or a lack of concern for quality, that are not easily "fixed" by changing organization structure.
At the core of these persistent calls for improving performance by changing traditional bureaucratic hierarchies, is a focus on decreasing the amount of hierarchy in organizations. This notion of "flattening" the hierarchy has both two meanings and dual purposes. First, flattening the hierarchy can mean cutting down on the number of administrative layers in an organization. By cutting down on the layers of management in an organization, information should flow more smoothly as there are fewer people involved who may slow down information flows or who for purposes of self-interest improperly manipulate information. More fluid and rapid flows of information should improve organizational efficiency and response.
The second reason for flattening hierarchies concerns who should make decisions in organizations. This purpose for flattening hierarchies is based on the notion that those employees closest to the action, or the clientele, are best suited to make decisions. From this perspective, hierarchy is flattened because employees are not expected to obtain permission for every action, but instead are empowered to provide quick responses to client needs.
An example of this type of flattening is community-based policing. This popular method of having police officers get to know, on an individual basis, the people "on their beats" and thus the real problems these people face. The increase in police "store fronts", small police offices in high crime areas also serves to get the government service provider closer to the citizen.
It is important to recognize that changing organizational structures does incorporate an essential aspect of what we know about nonlinear and dynamical systems. This important feature is that in such dynamical systems, structure creates behavior. In short, the way the parts fit together dictates much of the behavior generated by employees. Changing structure in organizations is likely to impact the level and quality of performance and service provision.
Students of disasters have also recently identified the problems of bureaucratic models for contending with the complexity and instability of disasters. Drabek (1994, p.41) notes in a recent paper, "...some within the emerging profession (emergency management) will continue to apply organizational theories that have a high probability of failing them and their communities". This comment raises further questions about our bureaucratic and paramilitary approaches to even high instability situations such as disasters. Order Through Disorder
Interestingly, recent investigations, based on nonlinear dynamics, offer a strong argument for the value of flattening the number of administrative layers in organizations and for placing decisions closer to the "field of action". Hershey and his colleagues (1990) have examined the amount of disorder in information flows in four different types of organizational structures. This study is particularly important for government organizations because government is primarily an information producer. In government, the efficient and orderly production of information is essential.
Hershey et al. analyzed the information disorder created by the organizational forms of (1) the ideal horizontal (flattened) organization; (2) the traditional vertical hierarchy; (3) an intermediate structure combining both hierarchy and horizontal qualities and; (4) chaos. The use of the term chaos here refers to total chaos when a system is completely disordered, and not to the chaos of time-series behavior used throughout this book. In such a chaotic organization there are no direct linkages for information to flow. While both the horizontal and the chaos models represented flattened hierarchies, model 1 included "perfect" communication with a single leader, while in the chaos model organizational units were completely independent and devoid of any defined leader.
These experiments (Hershey, et al., 1990) showed that the ideal horizontal structure produced the least disorder in information flow. This horizontal model thus evidenced the highest degree of organizational efficiency. The reader will probably not be surprised to discover that the intermediate model ranked second in efficiency, the vertical model third and the chaotic independent model fourth. This study appears to confirm not only the value of flatter organizational structures but also the important link between organizational structure and behavior.
What is essential for managers to remember is that the more disorderly "flattened" hierarchy creates better information flow and organizational efficiency. This flattened hierarchy is more disorderly in structure because it is less rigid and less militaristic in the way information is handled and in the way employees are led. By giving up the traditional management need for hierarchical order, improved information processing and organizational efficiency can be enhanced. Police officers in store fronts or in community-based policing programs are given considerable discretion to gather information, to learn, and to help citizens. This freedom inevitably creates disorder as employees are provided more freedom to resolve issues and problems. Structure, Process and Self-Organization
The structure of the emerging views of organization focus on processes. The previous model of the flattened organization is a good first start for the disaster and emergency services manager. Our previous review of the "order" created by flatter structures lends credence to the notion that flatter is better. Yet, what we now know about processes in organization is that well designed work processes are the key to quality and productivity (Deming, 1986). Work re-engineering also informs managers that organizational structure becomes less important in the re-engineered organization. Management must rethink organizational structure and identify processes that are linked in the organization. This suggests less of a focus on traditional structuring by functional area and instead consideration of where functional units converge to create outputs and service (Hammer and Champy, 1993).
The first action on management's part is a complete process mapping throughout the organization. Process mapping requires a thorough analysis of how work is accomplished and how information flows within and outside the organization. While these efforts are just beginning to take hold in government, efforts to reinvent government will increasingly focus on such practices. Since structure creates behavior, the way processes are structured and organized will in large part determine whether improved quality and performance are attained.
Focusing on process also means a different view of operational control. Well defined and efficient processes themselves become the source of control. Quality experts (Deming, 1986; Carr and Littman, 1990; Cohen and Brand, 1993) note that if processes are well considered, individual variation becomes less important, because the process limits individual variation. In this sense the process creates its own self-organization. Control mechanisms are based on workers given the opportunities to base their work on outputs and outcomes, not just individual tasks.
Employees, in the new emergent government agency, will also be trained to measure their own performance using the techniques of statistical process control (Deming, 1986). Employees will be allowed to examine their own performance so they may take part in determining any adjustments necessary to improve their performance. The time-series of performance provided to employees allows self-adjustment and more personal knowledge of when and why variation occurs in work processes. This places employee "self-control" at the heart of quality processes and limits the need for excessive management oversight. The culture of service serves as the principal source for employee self-adjustment to improve performance.
Parameters are defined by management but variation within these parameters is accepted as the inevitable result of the nonlinearity inherent in human organizations. Furthermore variation is not seen as a threat but instead as an opportunity to learn via an understanding of the sources of variation. Control mechanisms thus represent features of bounded instability. The process and the constraints of government work serve as the "bounds" of work. But workers are genuinely engaged to use their full capacities within the confines of the constraints of government regulations and public law. When re-engineering plans break-up previous processes, the manager must find new forms of order and process. Process is More Important than Structure
The importance of organizational structure is deeply ingrained in the traditional vision of management. People often have their egos directly tied into their own place in an organization's structure. A focus on process, rather than on organization charts and structures is itself a paradigmatic shift for managers. Even getting managers to think of themselves as part of a process rather than as part of an overarching structure may enhance the level of participation in organizations so essential for fully utilizing the skills of employees.
Managers and employees often do not take the time to examine all of the linkages between their functional units and other in the organization. But one need not be an expert in organizations to understand how much actual work crosses the boundaries of functional units. There is a clear need to see organizations as multitudes of process. This view provide an improved way of seeing how organizations either fit or do not fit together. The increasing emphasis on cross-functional teams in organizations emphasizes this point.
But the notion of process structures from the nonlinear paradigm is important to remember. It is the internal processes that allow structures to reform after a transforming and qualitative change, such as a disaster. Only by producing viable and dynamically unstable processes in organizations can managers hope to develop organizations that can cope with rapid transformations and increased complexity.
The inevitable resistance to organizational and process change has also been examined by Goldstein (1989) from the perspective of "a far-from-equilibrium approach". Goldstein (1989, p. 23) argues that resistance to change in organizations may be mollified by "difference questioning". This approach focuses on identifying the differences that may be generated by new modes of work and organization. Change may not represent what may initially be perceived as a threat, if employees are aware of the nature of the new changes. It is such "difference questioning" that may generate the instability necessary to liberate employees and managers from strong commitments to forms of work that inhibit performance and productivity gains.
The changes presented by work re-engineering may represent the greatest threat to managers and employees who resist change. The possibility of completely new means for accomplishing organizational goals, suggests real organizational upheaval. Yet, the process of work re-engineering also offers some means for contending with these potential changes in work similar to Goldstein's (1989) notion of "difference questioning". These means involve the essential questions that drive recognition of both the problems and prospects for change in work processes. These two questions are Why? and What? Why do we do things this way? What will happen if we change the present work process? Such questions, combined with employee involvement may serve to mollify many employee and management fears over potential changes. The task for the public manager is to fully inform employees of the benefits of such change. The applied task is to develop work processes that ensure employees that the nature of work will be both more productive and more rewarding.
Finally, the notion of the attractor may help in pushing work systems to new patterns of functioning. Think of the most stable and "tightly attracted" activities an agency performs. These highly attracted processes may be excessive sources of order that inhibit positive change. By improving quality in these areas it may enhance many of aspects of disaster response and emergency services. Planning, Strategy and Emergencies
Dallas, Texas resides on the southern edge of a geographic region subject to natural disasters called "tornado alley". People who reside in this part of north central Texas understand the destructive and deadly force of tornadoes. The focus on tornadoes in "tornado alley", however, revealed a recent example of mode lock-in and inflexible planning that has led to calls for a comprehensive reconsideration of Dallas' disaster management planning (Dallas Morning News, May 8, 1995).
Early on May 5, 1995 weather forecasts predicted heavy, but seasonally typical, spring thunderstorms for the north central Texas area. These storms often bring tornadoes. The city of Dallas emergency preparedness organization immediately devoted disaster resources to prepare for tornado damage and appropriate response. Rather than contending with the reality of tornado damage, however, torrential rainfall brought unexpected levels of flooding to Dallas. Emergency preparedness officials were simply unprepared for this level of flooding, at least in part, due to tornado damage directed resources. Sixteen citizens lost their lives due to flooding (Dallas Morning News, May 9, 1995).
This misguided planning in Dallas, although understandable, exemplifies the problems of mode lock-in during disaster planning. By locking into one mode in an early stage the City created initial conditions which inhibited flexible response to disaster. This emphasizes the importance of flexible planning and response to disaster. Initial mode lock-in may make alterations in mid-course quite difficult, particularly if the triggers to alter strategies are not known. Nonlinear Models for Planning
Scholars are increasingly critical of the types of models used for planning. For example, Jay Forrester (1987, p.110) has written, "There has been a reluctance to give up the linear mathematical procedures, with the result that models have been biased to fit the linear procedures at the expense of faithfulness in representing the real world". These linear models, moreover, have led traditional policy modelers to seek and identify variables and system behavior that lead to some clear image of the future (Kiel, 1992). As Anghel Rugina (1989) notes, such modeling efforts are based on mathematics that avoid the uncertainty inherent in real social systems. This has led many social scientists to believe that once the model produces a "stable" outcome to a problem then such a stability must represent the desired solution. This is obviously problematic since we know that social phenomena are inherently nonlinear and unstable. Traditional modeling thus seeks to generate stable solutions in an unstable world.
More recently, modelers have, however, begun to interject notions of instability, nonlinearity and uncertainty into computer models of social problems and policy options. One example of this emerging approach is Ann Stanley's (1989) model of the AIDS epidemic in the United States. The AIDS epidemic is clearly a major concern for public health officials across the nation.
Stanley (1989) examines the nonlinear nature of HIV infection and the concomitant nonlinear growth of actual AIDS disease. Her data are quite foreboding in their policy implications and for the pressures for public officials to respond to this nonlinear trend. Of further interest is that some of Stanley's models suggest the unexpected result of an eventual increase in AIDS within groups that were previously consider "low-risk". Interestingly, a recent study by medical researchers (Selik, Chu, and Buehler, 1993), supports Stanley's model, and, reveals an increasing rate of growth in AIDS among previously considered "low risk" groups. In particular, AIDS seems to be exploding among young women in many of America's urban areas.
While forecasting is perhaps an obvious example of where the information set available to the disaster response and emergency service manager is limited, recent studies show that actual processes of strategy making and goal seeking by managers are also subject to fluctuations that demand altering and adjusting plans and strategies. Strategy development for public managers clearly must include recognition of the many stakeholders involved in government (Nutt and Backoff, 1992). Government managers know that any effective strategy must consider all relevant actors impacted by any strategic plan. This means that successful strategies must incorporate the mutual interdependencies and mutual expectations of these actors. It is the dynamic nature of these interactions and expectations that over time can make the specific outcomes of strategies very erratic and unpredictable (Richards, 1990). At best it appears managers can only define the boundaries of their plans and strategies, the specifics may bounce around and change with little attention to the manager's initial intent.
Finally, in a nonlinear and uncertain world we need to think carefully about the language that some management analysts use. The term optimization is thrown around with abandon in some management circles. Who really believes that any human effort is optimal? And if we thought the effort was optimal how would we know? In our efforts to avoid mode lock-in, while doing our best we should eschew notions of optimal response. What seems more logical is ensuring that we have available a range of adequate responses across a range of potential disaster scenarios. This should ensure that we do not waste resources on unlikely events, but are still prepared for their arrival. Conclusion
Professionals involved in emergency response to disasters are what should be labeled "maximum uncertainty managers". Disasters reveal a level of uncertainty for public managers that is likely only equaled during battle in war. But how does one cope with such maximum uncertainty. The key here is continuous learning. Furthermore we learn, in an uncertain world where history does not necessarily repeat itself, that the rapid capacity to learn may be more important than experience.
We also learn that in the kind of uncertain and nonlinear world we live in that the tools available to managers are not like those available to the natural scientist. For as Jay W. Forrester (1987, p. 108) noted in reference to contending with our nonlinear world, "results are less generalizable, but more relevant. Sweeping theories are replaced by bounded classes of rules of thumb." These heuristics, or rules of thumb, thus are managements intellectual resources, resources that must be unstable as a means to deal with instability and inherent variation that exists in disaster situations.
Our understanding of how chaos theory, nonlinearity and instability and the sciences of complexity can help us better manage organizations is in its initial stages. There is much more that will be learned over the next several decades about complex systems and nonlinear dynamics as scholars and managers understand more about this new vision of reality. One of the more interesting aspects of this new paradigm is the timing of its appearance on the scene of management. Nonlinear dynamics as an emerging paradigm for emergency and disaster management in particular, and public management in general, has appeared at a critical juncture in our thinking about organizations and management. As Cavaleri and Obloj (1993, p. 387) write, "The discipline of management is itself at a bifurcation point in its evolution. Managers of today have more incentive than ever to explore new ways of managing and viewing the world." The goal of this conference is to explore these new ways of thinking about emergency and disaster response that may save lives and enhance the human condition.
Another goal for the managers at this conference should be the development of preparedness systems and action plans that do not require excessive management control and oversight. The notions of self-organization that apply to natural systems can be applied to management. The best organizational systems are ones that can do without management. These are systems that have the response capable to solve problems with maximal learning and minimal top down direction.
For additional copies of this paper or correspondence, contact -
L. Douglas Kiel, Ph.D., Associate Professor of Government and Politics and Political Economy, School of Social Sciences, University of Texas at Dallas, P.O. Box 830688, Richardson, TX 75083. Phone - (214) 883-2019; FAX (214) 883-2735; E-MAIL: DKIEL@UTDALLAS.EDU
CONVERGENCE TO AN EQUILIBRIUM
Sensitivity to Initial Conditions
2 points starting to close together - rapidly diverge
A CHAOTIC ATTRACTOR
The Order in a Chaotic Time Series
Burton, T. and Gibson, R. "Repairing Mississippi Levees to Be Costly But System is Unlikely To Improve." Wall Street Journal, July 26, 1993, A3.
Carr, D. K. and Littman, I. D. Excellence in Government. Arlington, Va.: Coopers & Lybrand, 1990.
Casti, J. L. Searching for Certainty: What Scientists Can Know About the Future. William Morrow and Co.: New York, 1990.
Cavaleri, S. and Obloj, K. Management Systems: A Global Perspective. Wadsworth Publishing Company.: Belmont Ca, 1993.
Cohen, S. and Brand, R. Total Quality Management in Government. San Francisco: Jossey-Bass, 1993.
Comfort, L. K. "Self Organization in Complex Systems." Journal of Public Administration Research and Theory. 4(3), 1994, 393-410.
Dallas Morning News, "Dallas emergency-response plan got off to late start, offical says", Monday, May 8, 1995, p.1
Dallas Morning News, "Officials defend city response to storms", Tuesday, May 9, 1995, p. 21A.
Deming, W. Out of the Crisis. Cambridge, Mass.: Cambridge University Press, 1986.
Ditto, W. and Pecaro, L.,"Mastering Chaos", Scientific American. August 1993, 78-84.
Downs, A. Inside Bureaucracy. Boston: Little, Brown, 1967.
Drabek, T. "Disaster in Aisle 13 Revisited", in Disasters, Collective Behavior, and Social Organization. R. Dynes and K. Tierney (eds.), 1994, pp. 26-44, Newark: University of Delaware Press.
Forrester, J.W. Urban Dynamics. Cambridge, Mass.: The M.I.T. Press, 1968.
Forrester, J. W. "Nonlinearity in High-Order Models of Social Systems." European Journal of Operational Research, 1987, 30, 104-109.
Garfinkel, A., Spano, M., Ditto, W., and Weiss, J., "Controlling Cardiac Chaos" Science. 1992, 257:1230-1235.
Gleick, J. Chaos: Making a New Science. New York: Viking, 1987.
Goldstein, J. "A Far From Equilibrium Systems approach to Resistance to Change." Organizational Dynamics, 1989, 17(2), 16-26.
Hammer, M. and Champy, J. Reengineering the Corporation: A Manifesto for Business Revolution. New York: Harper Collins, 1993.
Hershey, D.; Patel, V.; and Hahn, J. "Speculation on the Relationship Between Organizational Structure, Entropy, and Organizational Function." Systems Research, 1990, 7(3), 207-208.
Kiel, L. D. "The Nonlinear Paradigm: Advancing Paradigmatic Progress in the Policy Sciences." Systems Research, 1992, 9(2), 27-42.
Kiel, L. D. "Nonlinear Dynamical Analysis: Assessing Systems Concepts in a Government Agency." Public Administration Review, March/April 1993b, 53(2), 143-153.
Kiel, L. D. Managing Chaos and Complexity in Government: A New Paradigm for Managing Change, Innovation and Organizational Renewal. San Francisco: Jossey-Bass. 1994
Lerner, A. and Wanat, J. Public Administration: A Realistic Reinterpretation of Contemporary Public Management. Englewood Cliffs, N. J.: Prentice-Hall, 1992.
Likert, R. New Patterns of Management. New York: McGraw - Hill, 1961.
Mainzer, K. Thinking in Complexity: The Complex Dynamics of Matter, Mind, and Mankind. New York:Springer-Verlag, 1994.
Miller, G. "Debt Management Networks." Public Administration Review, January/February 1993, 53(1), 50-58.
Mosekilde, E.; Aracil, J. and Allen, P. M. "Instabilities and Chaos in Nonlinear Dynamic Systems." System Dynamics Review, 1988, 4, 14-55.
Nonaka. I. "Creating Organizational Order Out of Chaos; Self-Renewal in Japanese Firms". California Management Review, Spring 1988, 38, 57-73.
Nutt, P C. and Backoff, R. W. Strategic Management of Public and Third Sector Organizations. San Francisco, CA: Jossey-Bass, 1992.
Ott, E. Grebogi, C. and Yorke, J.A., "Controlling Chaos", Physical Review Letters. 1990, 64(11):1190-1193.
Pagels, H. The Dreams of Reason. New York: Simon and Schuster, 1988.
Priesmeyer, H. R. Organizations and Chaos: Defining the Methods of Nonlinear Management. Westport, Conn.: Quorum Books, 1992.
Richards, D. "Is Strategic Decision-Making Chaotic?" Behavioral Science, July 1990, 35, 219-232.
Roeser, T. "Chicago Flood's Lessons-Privatize." Wall Street Journal, May 28, 1992, p.38.
Rugina, A. "Principia Methodologica 1: A Bridge from Economics to All Other Natural Sciences - Towards a Methodological Unification of All Sciences." International Journal of Social Economics, 1989, 16(4), 3-76.
Selik, R.M.; Chu, S.Y.; and Buehler, J.W. "HIV Infection as Leading Cause of Death Among Young Adults in U.S. Cities and States." Journal of the American Medical Association, June 16, 1993; 269(23), 2991-2994.
Stanley, E. A. "Mathematical Models of the AIDS Epidemic: An Historical Perspective." In D. L. Stein (Ed.), Lectures in the Sciences of Complexity. Reading, Mass.: Addison-Wesley, 1989.
Next Chapter: ATTACHMENT A
Return to table of contents