In building a theory of the world, it helps if one's vision is a little bleary.
We build our orders, but only at the expense of creating randomness elsewhere.
George Johnson, Fire in the Mind
Perrow's quadrants, devised by the sociologist Charles Perrow as a result of his investigation of the Three Mile Island nuclear plant accident, abandons many details. His quadrants are a stripped, minimalist model. That is, they are "tacit," or low-level models, while high-level models, which contain a lot of detail, are "overt." More tacit models are responsive to the call for "coarse-graining;" for a "crude look at the whole," essential for dealing with nonlinear environments, and are a hallmark of Aids to Learning. The following summarizes Perrow's framework. (1)
- Charles Perrow's book (2) takes a unique approach to accident prevention and risk management. He focuses upon organizational causes of accidents rather than limiting his study to human error and equipment failure. He argues some accidents are inevitable and are in fact, normal. To understand Perrow's approach, one must first distinguish between what he calls incidents and accidents. Then, one needs to have a firm grasp upon two key concepts, interaction and coupling, which form the foundation of his thesis....
- Incidents and Accidents
- Perrow differentiates between simple incidents like backing a car into a telephone pole and nuclear accidents like Three Mile Island. For analysis, he organizes all "systems" (major end items) into four levels. Level one is the part, the smallest component of any system. Examples include valves, filters, gauges, etc. Level two is the unit, which is made up of parts. Examples are motors, pumps, wiring panels, etc. Level three is the subsystem, which is an array of units. Examples include propulsion systems on naval destroyers and navigation sets in aircraft. Level four is the system itself, which is the summation of its subsystems. Examples are aircraft carriers and space shuttles.
- An incident involves damage to parts (level one) and/or a unit (level two). An accident "is a failure in a subsystem, or the system as a whole, that damages more than one unit (a subsystem, or level three) and in doing so disrupts the ongoing or future output of the system (level four)." Incidents and accidents both begin with equipment failure or human error, but accidents continue on out of control with multiple, unanticipated failures in units and subsystems.
- Perrow's definitions can be confusing. A truck tire could blow, cause a wreck and kill a soldier. An extremely serious situation, but one he would categorize as an incident. Perrow takes an impersonal, rational, systemic approach....to analyze potential catastrophes. Integral to his analysis are the concepts of interaction and coupling.
- Linear and Complex Interactions
- The concept of interaction helps Perrow identify which systems are most prone to accidents. Linear interactions describe highly structured systems which are logical, sequential and planned. They function as a series of expected events in a predictable sequence. If damage to a part occurs, the problem can be identified and corrected with little disturbance to the overall system. Linear interactions are also characterized by minimal feedback loops which make it easier to understand and monitor the entire system.
- Complex interactions on the other hand, are less predictable. Breakdowns within one or more units and/or subsystems can occur because of unplanned and unforeseen interactions. Unexpected events may occur, regardless of intended system designs. Problems are not easily identifiable in complex systems, especially during the confusion that ensues from an accident.
- Advanced technology could make systems more complex and more difficult to understand and predict. Or, innovation could result in increased simplicity. One decreases the chance of accidents by increasing linearity in complex interactive systems. (The big attractions of complex systems lies in production efficiencies, not in safety considerations.) Holding everything else constant, linear designs are inherently more safe.
- Tight and Loose Coupling
- Perrow's second major concept is coupling, or the amount of "slack, buffer or give between two items." Loosely coupled systems are characterized by decentralized operations, mission orders, ambiguous performance standards and flexible control mechanisms. Change has little effect upon loose organizations. These types of systems allow a wide variety of responses during emergency situations. If something goes wrong, there is time to correct the problem without catastrophic consequence. Processes do not flow in rigid sequence. Field expedient solutions to problems and substitute equipment are readily accommodated.
- Tightly coupled systems are highly centralized and rigid. Output is closely monitored within specified tolerances. Subsystems are interdependent. Change causes massive ramifications throughout the system. Tightly controlled time schedules with little slack are sensitive to delays. Production sequences must be strictly followed. Substitutions are not easily accomplished and equipment breakdowns can bring the entire system to a halt. Safety features must be designed into the system because human intervention is not easily accommodated. Emergency override features may be built-in, but systems design makes on-the-spot, field expedient solutions difficult.
- Interaction/Coupling Chart
- Figure 7.1 shows relationships between interaction and coupling. A tight-linear organization falls into quadrant 1, tight-complex into quadrant 2, loose-linear in quadrant 3 and loose-complex in quadrant 4. Arguably, examples within the quadrants are used to illustrate various incident/accident potentials.
- A railroad company (quadrant 1) is a tight-linear organization. Tight coupling tendencies: Trains run on time. Management has limited flexibility in the use of tracks. Trains must be staggered and time buffers rigidly followed. Experienced personnel and specialized equipment are required; substitutes have to meet standards which limit options during emergencies. Linear interaction tendencies: Rail cars are spread around the country to meet customer demand. Failures within the system are relatively easy to locate. Direct, on-line information sources exist. Operations are sequential and procedures are usually conducted "by the numbers."
- NASA (quadrant 2) is a tight-complex organization. Tight coupling tendencies: Time schedules are rigidly followed (which partially explains the Challenger disaster). Once a spacecraft is launched, NASA is committed. Specific actions and sequence of events must occur. Safety features are designed into the system and few substitutions of equipment, supplies and personnel are possible. The inflexible nature of this tight system is illustrated by the tremendous ingenuity and luck required for the safe return of Apollo 13. Complex interaction tendencies: Highly specialized personnel work in the US space program. Equipment is tightly packed in small spaces and interdependencies of functions are great.
- A neighborhood gasoline station (quadrant 3) is a loose-linear organization. Loose coupling tendencies: Attendants have flexibility in servicing cars at the pumps and in the bays. Customers have choices between different grades of fuel, viscosities of oil and brand names on repair parts. Backlogs occur at different times of the day with few ramifications. Skills required are relatively few so employees can be readily replaced when problems occur. Linear interaction tendencies: Equipment is spread out-tools and diagnostic kits are scattered among the bays and cars. Turnover of personnel has little effect to include summertime when high school kids are hired temporarily. Customers are served on a first come basis, but exceptions can be made easily. Few surprises occur because information sources are direct and firsthand.
- A college is a loose-complex organization. (quadrant 4) Loose coupling tendencies: Class schedules are easily changed based on the availability of the instructor. If textbook orders are not filled before the semester begins, alternates may be selected or photocopies made of existing texts. Slack is present; a class falling on a holiday is easily slipped to another day. When an instructor is ill, another may substitute or students sit in on another class. Complex interaction tendencies: Feedback loops exist between the students, the dean and faculty...Indirect and inferential information sources complement the formal feedback loops that provided the impetus for change.
- There is less chance of accidents in loosely coupled organizations compared to tightly constructed ones. A catastrophe is far less likely at a gasoline station or a college compared to a mainline on Southern Railroad or during space shuttle flight. Once an incident or accident is about to occur, or is in progress, it is easier for a linear organization than a complex one to control the situation. One can fix a problem easier on a railroad than in a space shuttle.
Based upon his investigations, Perrow concludes that the inherent nature of effective "authority" [or command and managerial processes and styles] fundamentally differs in each individual quadrant, as follows: See Figure 7.2.
It is not only that linear and nonlinear techniques need to be meshed. All kinds of Tools of Analysis are routinely combined every day to solve problems in a linear way. So too can Aids to Learning be intertwined to provide insights. Meshing occurs both between and within these regimes.
In the following example, three nonlinear avenues will be drawn into intersection-Perrow's Quadrants, the Period-Doubling Cascade of the "playing field," and Van Creveld's Rules-providing a focus in which insights yield knowledge. The process is fundamentally different from linearity's focus on transforming data into information.
Suppose we were to dismantle the quadrants and lay them end-to-end in the following sequence: Tightly linear, loosely linear, loosely nonlinear, and tightly nonlinear, or Quadrants 1-3-4-2. See Figure 7.3.
What we would have is a continuum which "parallels" the place on which the game of nonlinearity is played. While it is parallel and consistent with the features of the period-doubling cascade, it is not demarcated. Nevertheless, one can conjecture that-
The composite picture that is derived, therefore, looks something like Figure 7.4.
To carry the matter of meshing Aids to Learning one step further: What if we were to relate Van Creveld's Rules to Perrow's Quadrants? I tried that in a article published in Parameters in 1996, which is included in the Appendix (3).
In this article I suggested that the Army's Force XXI "digitized battlefield" appeared to have the characteristics of a tightly coupled and complex system, and therefore might fall into Perrow's quadrant 2. If so, its future is problematic. Perrow explains that the problem with tightly coupled, complex systems is that
- the demands are inconsistent. Because of the complexity, they are best decentralized; because of the tight coupling, they are centralized. While some mix may be possible, and is sometimes tried (handle small duties on your own, but execute orders from on high for serious matters), this appears to be difficult for systems that are reasonably complex and tightly coupled, and perhaps impossible for those that are highly complex and tightly coupled. (4)
Perrow's quadrants, shorn of all distracting "noise" and embellishments, yields insights. After all, the obverse of a safety concern can be viewed as a military consideration. They are different sides of the same coin. Imagine the following discussion, somewhat along lines which have occurred in class:
Q: If we assume that democracies don't fight each other, in which quadrant(s) will future threats likely come from?
A: Probably Quadrant 1, which has the earmarks of an authoritarian regime.
Q:Where does Quadrant 1 lie on the Period-Doubling Cascade continuum?
A:Adjacent to the Edge of Equilibrium. It has minimal nonlinear attributes.
Q:What is its authority rule?
A: Highly centralized-a command and control economy, society, and military.
Q:In terms of "ends" and "means," what is the end?
A:Knock it over the edge into Equilibrium.
Q:What are the means?
A:Affect the small, but critical, elements of its nominal cas.
Q: What could be the actual center(s) of gravity? Think of the seven attributes of complex adaptive systems.
A:Usually we focus automatically on flows when it comes to centers of gravity. But, in this case, I think we ought to look more carefully at tagging...[Another student] Wait, if we can't get at that small component of nonlinearity directly, can't we just make the system's linearity even more intense, to the point where we are accomplishing the same thing?
Q: Our force projection lies in which Quadrant?
A: Probably Quadrant 1, too, assuming that everything but precision strike is ruled out.
Q: Are there exceptions, or is it a rule, that when both opposing forces emanate from the same Quadrant that winning can only come through attrition? Are there shades of gray?.and so on, and on.
| Coping with the Bounds Index | Foreword | Acknowledgments | Introduction | Part One Introduction | Chapter 1 | Chapter 2 | Chapter 3 | Chapter 4 | Part Two Introduction | Chapter 5 | Chapter 6 | Chapter 7 | Chapter 8 | Chapter 9 | Chapter 10 | Conclusion | Appendix 1 | Appendix 2 | Appendix 3 | Appendix 4 | Appendix 5 | Appendix 6 | Notes |