This is chapter 2 of the "The Macroscope" by Joël de Rosnay



The fundamental concepts that recur most often in the biological, ecological, and economic models of the preceding chapter can easily be grouped into several major categories: energy and its use; flows, cycles, and stocks, communication networks; catalysts and transforming agents; the readjustment of equilibriums; stability, growth, and evolution. And, above all, the concept of the system-living system, economic system, ecosystem-that binds together all the others.

Each of these concepts applies to the cell as it does to the economy, to an industrial company as it does to ecology. Beyond the vocabulary, the analogies, and the metaphors there appears to exist a common approach that makes it possible to understand better and describe better the organized complexity.

The Systemic Approach

This unifying approach does indeed exist. It was born in the course of the last thirty years from the cross-fertilization of several disciplines- biology, information theory, cybernetics, and systems theory. It is not a new concept, what is new is the integration of the disciplines that has come about around it. This transdisciplinary approach is called the systemic approach, and this is the approach that I present here in the concept of the macroscope. It is not to be considered a "science," a "theory," or a "discipline," but a new methodology that makes possible the collection and organization of accumulated knowledge in order to increase the efficiency of our actions.

The systemic approach, as opposed to the analytical approach, includes the totality of the elements in the system under study, as well as their interaction and interdependence.

The systemic approach rests on the conception of system. While often vague and ambiguous, this conception is nevertheless being used today in an increased number of disciplines because of its ability to unify and integrate.

According to the most widely used definition, "a system is a set of interacting elements that form an integrated whole" ( see notes ). A city, a cell, and a body, then, are systems. And so are an automobile, a computer, and a washing machine! Such a definition can be too general. Yet no definition of the word system can be entirely satisfying; it is the conception of system that is fertile-if one measures its extent and its limits.

The limits are well known. Applied too easily, the systems concept is often used wildly in the most diverse areas: education, management data processing, politics. For numerous specialists it is only an empty notion: trying to say everything, it evokes nothing in the end.

Yet its reach cannot be held to the precision of definitions; the concept of system is not readily confined. It reveals and enriches itself only in the indirect illumination of the many clusters of analogous, modeled and metaphoric expression. The concept of system is the crossroads of the metaphors; ideas from all the disciplines travel there. Reaching beyond single analogies, this circulation makes possible the discovery of what is common among the most varied systems. It is no longer a matter of reducing one system to another, better-known one (economics to biology for example); nor does it mean transposing knowledge from a lower level of complexity to another level. It is a question of identifying nonvariants- that is, the general, structural, and functional principles- and being able to apply them to one system as well as another. With these principles it becomes possible to organize knowledge in models that are easily transferred and then to use some of these models in thought and action. Thus the concept of system appears in two complementary aspects: it enables the organization of knowledge and it renders action more efficient.

In concluding this introduction to the concept of system, we need to locate the systemic approach with respect to other approaches with which it is often confused.

  • The systemic approach embraces and goes beyond the cybernetics approach (N. Wiener, 1948), whose main objective is the study of control in living organisms and machines ( see notes ).
  • It must be distinguished from General Systems Theory (L. von Bertalanffy, 1954), whose purpose is to describe in mathematical language the totality of systems found in nature.
  • It turns away from systems analysis, a method that represents only one tool of the systemic approach. Taken alone, it leads to the reduction of a system to its components and its elementary interactions.
  • The systemic approach has nothing to do with a systematic approach that confronts a problem or sets up a series of actions in sequential manner, in a detaile
  • d way, forgetting no element and leaving nothing to chance.

    Perhaps one of the best ways of seeing the strength and the impact of the systemic approach is to follow its birth and development in the lives of men and institutions.

    The Search for New Tools

    The process of thought is at once analytic and synthetic, detailed and holistic. It rests on the reality of facts and the perfection of detail. At the same time, it seeks factors of integration, catalytic elements for invention and imagination. At the very moment that man discovered the simplest elements of matter and life, he tried, with the help of the famous metaphors of the "clock," the "machine," the "living organism," to understand better the interactions between these elements.

    Despite the strengths of these analogical models, thought is dispersed in a maze of disciplines each secluded one from another by communication-tight enclosures. The only way to master these numbers, to understand and predict the behavior of the multitudes made up of atoms, molecules, or individuals, is to reduce them to statistics and to derive from them the laws of unorganized complexity.

    The theory of probability the kinetic theory of gases, thermodynamics, and population statistics all rely on unreal, "ghostly" phenomena, on useful but ideal simplifications that are almost never found in nature. Theirs is the universe of the homogeneous, the isotope, the additive, and the linear; it is the world of "perfect" gases, of "reversible" reactions, of "perfect" competition.

    In biology and in sociology, phenomena integrate duration and irreversibility. Interactions between elements count as much as the elements themselves. Thus we need new tools with which to approach organized complexity, interdependence, and regulation.

    The tools emerged in the United States in the 1940s from the crossfertilization of ideas that is common in the melting pot of the large universities.

    In illustrating a new current of thought, it is often useful to follow a thread. Our thread will be the Massachusetts Institute of Technology (MIT). In three steps, each of about ten years, MIT was to go from the birth of cybernetics to the most critical issue, the debate on limits to growth. Each of these advances was marked by many travels back and forth-typical of the systemic approach-between machine, man, and society. In the course of this circulation of ideas there occurred transfers of method and terminology that later fertilized unexplored territory.

    In the forties the first step forward led from the machine to the living organism, transferring from one to the other the ideas of feedback and finality and opening the way for automation and computers (Fig. 39).


    In the fifties it was the return from the living organism to the machine, with the emergence of the important concepts of memory and pattern recognition, of adaptive phenomena and learning, and new advances in bionics: artificial intelligence and industrial robots.[1] There was also a return from the machine to the living organism, which accelerated progress In neurology, perception, the mechanisms of vision (Fig. 40).


    In the sixties MIT saw the extension of cybernetics and system theory

    to industry, society, and ecology (Fig. 41).


    Three men can be regarded as the pioneers of these great breakthroughs: the mathematician Norbert Wiener, who died in 1964, the neurophysiologist Warren McCulloch, who died in 1969, and Jay Forrester, professor at the Sloan School of Management at MIT.

    There are of course other men, other research teams, other universities-in the United States as well as in the rest of the world-that have contributed to the advance of cybernetics and system theory. I will mention them whenever their course of research blends with that of the MIT teams.

    "Intelligent" Machines

    Norbert Wiener had been teaching mathematics at MIT since 1919. Soon after his arrival there he had become acquainted with the neurophysiologist Arturo Rosenblueth, onetime collaborater of Walter B. Cannon (who gave homeostasis its name) ( see page 43 ) and then at Harvard Medical School. Out of this new friendship would be born, twenty years later, cybernetics. With Wiener's help Rosenblueth set up small interdisciplinary teams to explore the no man's land between the established sciences.

    In 1940 Wiener worked with a young engineer, Julian H. Bigelow, to develop automatic range finders for antiaircraft guns. Such servomechanisms are able to predict the trajectory of an airplane by taking into account the elements of past trajectories. During the course of their work Wiener and Bigelow were struck by two astonishing facts: the seemingly "intelligent" behavior of these machines and the "diseases" that could affect them. Theirs appeared to be "intelligent" behavior because they dealt with "experience" (the recording of past events) and predictions of the future. There was also a strange defect in performance: if one tried to reduce the friction, the system entered into a series of uncontrollable oscillations.

    Impressed by this disease of the machine, Wiener asked Rosenblueth whether such behavior was found in man. The response was affirmative: in the event of certain injuries to the cerebellum, the patient cannot lift a glass of water to his mouth; the movements are amplified until the contents of the glass spill on the ground. From this Wiener inferred that in order to control a finalized action (an action with a purpose) the circulation of information needed for control must form "a closed loop allowing the evaluation of the effects of one's actions and the adaptation of future conduct based on past performances." This is typical of the guidance system of the antiaircraft gun, and it is equally characteristic

    of the nervous system when it orders the muscles to make a movement whose effects are then detected by the senses and fed back to the brain (Fig. 42).

    Thus Wiener and Bigelow discovered the closed loop of information necessary to correct any action-the negative feedback loop-and they generalized this discovery in terms of the human organism.

    During this period the multidisciplinary teams of Rosenblueth were being formed and organized. Their purpose was to approach the study of living organisms from the viewpoint of a servomechanisms engineer and, conversely, to consider servomechanisms with the experience of the physiologist. An early seminar at the Institute for Advanced Study at Princeton in 1942 brought together mathematicians, physiologists, and mechanical and electrical engineers. In light of its success, a series of ten seminars was arranged by the Josiah Macy Foundation. One man working with Rosenblueth in getting these seminars under way was the neurophysiologist Warren McCulloch, who was to play a considerable role in the new field of cybernetics. In 1948 two basic publications marked an epoch already fertile with new ideas: Norbert Wiener's Cybernetics, or Control and Communication in the Animal and the Machine; and The Mathematical Theory of Communication by Claude Shannon and Warren Weaver ( see notes ). The latter work founded information theory.

    The ideas of Wiener, Bigelow, and Rosenblueth caught fire like a trail of powder. Other groups were formed in the United States and around the world, notably the Society for General Systems Research, whose publications deal with disciplines far removed from engineering, such as sociology, political science, and psychiatry.

    The seminars of the Josiah Macy Foundation continued, opening to new disciplines: anthropology with Margaret Mead, economics with Oskar Morgenstern. Mead urged Wiener to extend his ideas to society as a whole. Above all, the period was marked by the profound influence of Warren McCulloch, director of the Neuropsychiatric Institute at the University of Illinois.

    At the conclusion of the work of his group on the organization of the cortex of the brain, and especially after his discussions with Walter Pitts, a brilliant, twenty-two-year-old mathematician, McCulloch understood that a beginning of the comprehension of cerebral mechanisms (and their simulation by machines) could come about only through the cooperation of many disciplines. McCulloch himself moved from neurophysiology to mathematics, from mathematics to engineering.

    Walter Pitts became one of Wiener's disciples and contributed to the exchange of ideas between Wiener and McCulloch; it was he who succeeded in convincing McCulloch to install himself at MIT in 1952 with his entire team of physiologists.

    From Cybernetics to System Dynamics

    In this famous melting pot, ideas boiled. From one research group to another the vocabularies of engineering and physiology were used interchangeably. Little by little the basics of a common language of cybernetics was created: learning, regulation, adaptation, self-organization, perception, memory. Influenced by the ideas of Bigelow, McCulloch developed an artificial retina in collaboration with Louis Sutro of the laboratory of instrumentation at MIT. The theoretical basis was provided by his research on the eye of the frog, performed in 1959 in collaboration with Lettvin, Maturana, and Pitts. The need to make machines imitate certain functions typical of living organisms contributed to the speeding up of progress in the understanding of cerebral mechanisms. This was the beginning of bionics and the research on artificial intelligence and robots.

    Paralleling the work of the teams of Wiener and McCulloch at MIT, another group tried to utilize cybernetics on a wider scope. This was the Society for General Systems Research, created in 1954 and led by the biologist Ludwig von Bertalanffy. Many researchers were to join him: the mathematician A. Rapoport, the biologist W. Ross Ashby, the biophysicist N. Rashevsky, the economist K. Boulding. In 1954 the General Systems Yearbooks began to appear; their influence was to be profound on all those who sought to expand the cybernetic approach to social systems and the industrial firm in particular.

    During the fifties a tool was developed and perfected that would permit organized complexity to be approached from a totally new angle-the computer. The first ones were ENIAC (1946) and EDVAC or EDSAC (1947). One of the fastest was Wirlwind II, constructed at MIT in 1951. It used-for the first time-a superfast magnetic memory invented by a young electronics engineer from the servomechanisms laboratory, Jay W. Forrester ( see notes ).[2]

    As head of the Lincoln Laboratory, Forrester was assigned by the Air Force in 1952 to coordinate the implementation of an alert and defense system, the SAGE system, using radar and computers for the first time.[3] Its mission was to detect and prevent possible attack on American territory by enemy rockets. Forrester realized the importance of the systemic approach in the conception and control of complex organizations involving men and machines in "real time": the machines had to be capable of making vital decisions as the information arrived.

    In 1961, having become a professor at the Sloan School of Management

    at MIT, Forrester created Industrial Dynamics. His object was to regard all industries as cybernetics systems in order to simulate and to try to predict their behavior.

    In 1964, confronted with the problems of the growth and decay of cities, he extended the industrial dynamics concept to urban systems (Urban Dynamics). Finally, in 1971, he generalized his earlier works by creating a new discipline, system dynamics, and published World Dynamics. This book was the basis of the work of Dennis H. Meadows and his team on the limits to growth. Financed by the Club of Rome, these works were to have worldwide impact under the name MIT Report.

    Figure 43 brings together the researchers and teams mentioned in the preceding pages and recalls the main lines of thought opened up by their work.


    The systemic approach depends on cybernetics and system theory. Perhaps it will be useful here to recall a few definitions. Cybernetics is the discipline that studies communication and control in living beings and in the machines built by man. A more philosophical definition, suggested by Louis Couffignal in 1958, considers cybernetics as "the art of assuring efficiency of action" ( see notes ). The word cybernetics was reinvented by Norbert Wiener in 1948 from the Greek kubernetes, pilot, or rudder.[4]

    One of the very first cybernetics mechanisms to control the speed of the steam engine, invented by James Watt and Matthew Boulton in 1788, was called a governor, or a ball regulator. Cybernetics has in fact the same root as government: the art of managing and directing highly complex systems.

    There are definitions of the word system other than that given at the beginning of this chapter. This is the most complete: "a system is a set of elements in dynamic interaction, organized for a goal. "

    The introduction of finality (the goal of the system) in this definition may be surprising. We understand that the purpose of a machine has been defined and specified by man; but how does one speak of the purpose of a system like the cell? There is nothing mysterious about the "goal" of the cell. It suggests no scheme; it declares itself a posteriori: to maintain its structure and replicate itself. The same applies to the ecosystem. Its purpose is to maintain its equilibrium and permit the development of life. No one has set the level of the concentration of oxygen in the air, the average temperature of the earth, the composition of the oceans. They are maintained, however, within very strict limits.

    The preceding definition is distinct from that of a certain structuralist tendency, for which a system is a closed structure. Such a structure cannot evolve but passes through phases of collapse due to an internal disequilibrium.

    In fact such definitions, as we said, are too general to be truly useful. They do not allow clarification of such ambiguities of expression as "a political system," "a computer system," and "a system of transportation." On the other hand, it seems to be much more profitable to enrich the concept of systems by describing in the most general way the principal characteristics and properties of systems, no matter what level of complexity they may belong to.[5]

    Open Systems and Complexity

    Each of the Russian dolls described in the first chapter is an open system of high complexity. These are important concepts that we must examine.

    An open system is in permanent relation with its environment, or, in general terms, with its ecosystem. It exchanges energy, matter, and information used in the maintenance of its organization to counter the ravages of time. It dumps into the environment entropy, or "used" energy. By virtue of the flow of energy through the system-and despite the accumulation of entropy in the environment-the entropy in an open

    system is maintained at a relatively low level. This is another way of saying that the organization of the system is maintained. Open systems can decrease entropy locally and can even evolve toward states of higher complexity.

    An open system, then, is a sort of reservoir that fills and empties at the same speed: water is maintained at the same level as long as the volume of water entering and leaving remain the same (Fig. 44).

    To emphasize the generality and importance of the concept of the open system, I have used the same kind of basic diagram for the industrial firm, the city, the living organism, and the cell. One must keep in mind that open system and ecosystem (or environment) are in constant interaction, each one modifying the other and being modified in return (Fig. 45).

    A closed system exchanges neither energy nor matter nor information with its environment; it is totally cut off from the outside world. The system uses its own internal reserve of potential energy. As its reactions take place, entropy advances irreversibly. When the thermodynamic equilibrium is reached, entropy is maximized: the system can no longer produce work. Classical thermodynamics considers only closed systems, but a closed system is an abstraction of physicists, a simplification that has made possible the fundamental laws of physical chemistry (Fig. 46).

    How to define complexity? Or, to avoid definitions, how to illustrate and enrich the significance of the concept? Two factors are important: the variety of elements and the interaction between elements.

    A gas, a simple system, is made up of similar elements (molecules of oxygen, for example) that are unorganized and display weak interactions. On the other hand, a cell-a complex system-includes a large variety of organized elements in tight interaction with one another. One could illustrate the concept of complexity with several points.

  • A complex system is made up of a large variety of components or elements that possess specialized functions.
  • These elements are organized in internal hierarchical levels (in the human body, for example, cells, organs, and systems of organs).
  • The different levels and individual elements are linked by a great variety of bonds. Consequently there is a high concentration of interconnections.[6]
  • The interactions between the elements of a complex system are of a particular type; they are nonlinear interactions.
  • The effects of simple linear interactions can be described by mathematical relationships in which the variables are increased or diminished by a constant quantity (as in the case of a car moving at the same average speed on a highway) (Fig. 47).

    However, in the case of nonlinear interactions the variables are multiplied or divided by coefficients which are themselves functions of other variables. This is the case of exponential growth (the quantity plotted on the vertical axis doubles by unit of time) or of an S curve (rapid growth following stabilization) (Fig. 48).

    Another example of a nonlinear relationship is the response of enzymes to different concentrations of substrate (molecules that they transform). In some cases, in the presence of inhibitors, the speed of transformation is slow. In others, in the presence of activators, the reaction is rapid up to the saturation of the active sites. In Figure 49 below this situation is expressed in curves that show the number of molecules transformed (I) in the presence of an inhibitor, (2) in the presence of an activator, and (3) according to the relative concentration of inhibitors and activators.

    Linked to the concept of complexity are those of the variety of elements and interactions, of the nonlinear aspect of interactions, and of organized totality. There follows a very special behavior of complex systems that is difficult to predict. It is characterized by the emergence of new properties and great resistance to change.

    Structural and Functional Aspects of Systems

    Two groups of characteristic features make it possible to describe in a very general way the systems that can be observed in nature. The first group relates to their structural aspect, the second to their functional aspect.

    The structural aspect concerns the organization in space of the components or elements of a system; this is spatial organization. The functional aspect concerns process, or the phenomena dependent on time (exchange, transfer, flow, growth, evolution); this is temporal organization.

    It is easy to connect the structural and functional elements by using a simple graphic illustration, a "symbolic meccano," which makes it possible to construct models of different systems and to understand better the role of interactions.[7]

    The principal structural characteristics of every system are:

    A limit that describes the boundaries of the system and separates it from the outside world. It is the membrane of a cell, the skin of a body, the walls of a city, the borders of a country.

    Elements or components that can be counted and assembled in categories, families, or populations. They are the molecules of a cell, the inhabitants of a city, the personnel of an industrial firm and its machines, institutions, money, goods.

    Reservoirs in which the elements can be gathered and in which energy, information, and materials are stored. In the first chapter numerous examples were given: reservoirs in the atmosphere and in the sediments, reservoirs of hydrocarbons; stores of capital and technology; memory banks, libraries, films, tape recordings; the fats of the body, glycogen of the liver. The symbolic representation of a reservoir is a simple rectangle.

    A communication network that permits the exchange of energy, matter, and information among the elements of the system and between different reservoirs. This network can assume the most diverse forms: pipes, wires, cables, nerves veins, arteries, roads, canals, pipelines, electric transmission lines. The network is represented in diagrams by lines and dotted lines that link the reservoirs or other variables of the model.

    The principal functional characteristics of every system are:

    Flows of energy, information, or elements that circulate between reservoirs. They are always expressed in quantities over periods of time. There are flows of money (salaries in dollars per month), finished products (number of cars coming off the assembly line by the day or the month), people (number of travelers per hour), information (so many bits of information per microsecond in a computer). Flows of energy and materials raise or lower the levels in the reservoirs. They circulate through the networks of communication and are represented symbolically by a heavy black arrow (flows of information are indicated by a dotted-line arrow). Information serves as a basis for making the decisions that move the flows which maintain reserves or raise and lower the levels of the reservoirs.

    Valves that control the volume of various flows. Each valve is a center of decision that receives information and transforms it into action: a manager of industry, an institution, a transforming agent, a catalyst such as an enzyme. Valves can increase or diminish the intensity of flows. Their symbolic representation is that of a valve or a faucet superimposed on a line of flow (Fig. 50).

    Delays that result from variations in the speed of circulation of the flows, in the time of storage in the reservoirs, or in the "friction" between elements of the system. Delays have an important role in the phenomena of amplification or inhibition that are typical of the behavior of complex systems.

    Feedback loops or information loops that play a decisive part in the behavior of a system through integrating the effects of reservoirs, delays, valves, and flows. Numerous examples of feedback were given in the first chapter: population control, price equilibriums, the level of calcium in the plasma (see pp. 10, 25, 43). There are two kinds of feedback loops. Positive feedback loops contain the dynamics for change in a system (growth and evolution, for example); negative feedback loops represent control and stability, the reestablishment of equilibriums and self-maintenance.

    The model in Figure 51 combines all the structural and fundamental symbols described above. And here it is possible to illustrate the difference between a positive and a negative feedback loop. If the information received at the level of the reservoir indicates that the level is rising, the decision to open the valves wider will allow overflow; if the level is falling, the decision to reduce the outflow will lead to a rapid drying up of the reservoir. This is the work of a positive feedback loop, working toward infinity or toward zero. In contrast, the decision to diminish the flow when the level increases (and the inverse) maintains the level at a constant depth. This is the work of a negative feedback loop.


    The basic functioning of systems depends on the interplay of feedback loops, flows, and reservoirs. They are three of the most general concepts of the systemic approach, and they are the keys to the juxtaposition of very different areas from biology to management, from engineering to ecology.

    Positive and Negative Feedback

    In a system where a transformation occurs, there are inputs and outputs. The inputs are the result of the environment's influence on the system, and the outputs are the influence of the system on the environment. Input and output are separated by a duration of time, as in before and after, or past and present (Fig. 52).

    In every feedback loop, as the name suggests, information about the result of a transformation or an action is sent back to the input of the system in the form of input data. lf these new data facilitate and accelerate the transformation in the same direction as the preceding results, they are positive feedback-their effects are cumulative. If the new data produce a result in the opposite direction to previous results, they are negative feedback-their effects stabilize the system. In the first case there is exponential growth or decline; in the second there is maintenance of the equilibrium (Fig. 53).

    Positive feedback leads to divergent behavior: indefinite expansion or explosion (a running away toward infinity) or total blocking of activities (a running away toward zero). Each plus involves another plus; there is a snowball effect. The examples are numerous: chain reaction, population explosion, industrial expansion, capital invested at compound interest, inflation, proliferation of cancer cells. However, when minus leads to another minus, events come to a standstill. Typical examples are bankruptcy and economic depression (Fig. 54).

    In either case a positive feedback loop left to itself can lead only to the destruction of the system, through explosion or through the blocking of all its functions. The wild behavior of positive loops-a veritable death wish-must be controlled by negative loops. This control is essential for a system to maintain itself in the course of time.

    Negative feedback leads to adaptive, or goal-seeking behavior: sustaining the same level, temperature, concentration, speed, direction. In some cases the goal is self-determined and is preserved in the face of evolution: the system has produced its own purpose (to maintain, for example, the composition of the air or the oceans in the ecosystem or the concentration of glucose in the blood). In other cases man has determined the goals of the machines (automats and servomechanisms).

    In a negative loop every variation toward a plus triggers a correction toward the minus, and vice versa. There is tight control; the system oscillates around an ideal equilibrium that it never attains. A thermostat or a water tank equipped with a float are simple examples of regulation by negative feedback (Fig. 55).[8]

    Flows and Reservoirs

    The dynamic behavior of every system, regardless of its complexity, depends in the last analysis on two kinds of variables: flow variables and state or level variables. ( see notes ). The first are symbolized by the valves that control the flows, the second (showing what is contained in the reservoirs) by rectangles. The flow variables are expressed only in terms of two instants, or in relation to a given period, and thus are basically functions of time. The state (level) variables indicate the accumulation of a given quantity in the course of time; they express the result of an integration. If time stops, the level remains constant (static level) while the flows disappear-for they are the result of actions, the activities of the system.

    Hydraulic examples are the easiest to understand. The flow variable is represented by the flow rate, that is, the average quantity running off between two instants. The state variable is the quantity of water accumulated in the reservoir at a given time. If you replace the flow of water by a flow of people (number of births per year), the state variable becomes the population at a given moment.

    The difference between flow variables and state variables is illustrated perfectly by the difference between the profit and loss statement and the balance sheet of a firm. The profit and loss statement is concerned with the period between two instants, say January I and December 31. It consists of an aggregation of flow variables: salaries paid, total purchases, transportation, interest costs, total sales. The balance sheet applies to one date only, say December 31. It is an instant picture of the situation of the company at that single moment in time. The balance sheet contains a variety of state variables: on the assets side, real estate and property, inventory, accounts receivable; on the liabilities side, capital, long-term debt, accounts payable.

    Three examples will serve to explain the relationships between flow variables and state variables and will clarify several of the ways in which they act at the different levels of a complex system.

    Balancing one's budget. A bank account (reservoir) fills or empties in accordance with deposits or withdrawals of money. The balance in the account at a given date is a state variable. Wages and other income of the holder of the account represent a flow variable expressed in a quantity of money for a period of time; expenses correspond to the flow variable of output. The valves that control these two flows are the decisions that are made based on the state of the account (Fig. 56).

    "To make ends meet" is to establish an equilibrium of the flows: income (input) equal to expenses (output). The bank account is kept at a stationary level. This is a case of dynamic equilibrium.[9]

    When the input flow is greater than the output flow, money accumulates in the account. The owner of the account is "saving." In saving he increases his total income by the amount of interest his savings earn (an example of a positive feedback loop).

    When the output flow is greater than the input flow, debts are accumulated. This situation can deteriorate further, for interest on debts increases output (a positive feedback loop toward zero). If the situation is not remedied, it can lead to the exhaustion of funds in a short time.

    The maintenance of equilibrium requires tight control. Control can be exercised more easily on the output flow valve (expenses) than on the input flow valve (income). This control imposes a choice of new constraints: the reduction or the better distribution of expenditures. In contrast, to make one's income increase rapidly one has to have reserves (savings) at his disposal-or benefit by a raise in salary.

    Managing a company. In the short term the manager uses internal indicators such as sales, inventory, orders placed, changes in production margins, productivity, delivery delays, money in reserve. For longer periods he consults his balance sheet, profit and loss statement, and such outside indicators as the prime rate of interest, manpower, growth of the economy. Using the differences between these indicators and the business forecasts, the manager takes what corrective measures are necessary. Consider two examples related to inventory and cash management.

    An inventory is a reservoir filled by production and emptied by sales. When the inventory is too high, the manager can influence the flow of sales either by lowering prices or by reinforcing marketing. He can also control the input flow in the short term by slowing down production (Fig. 57).

    The reverse situation is that of strong demand. The inventory level drops rapidly, and the manager then tries to increase production. If demand remains strong, the company-its inventory low-will require longer delays in delivery. Customers will not want to wait and will turn to a competitor. Demand then decreases and the inventory level climbs. A negative feedback loop helps the business leader-or works to his disadvantage if he has increased production too much without having foreseen the change in the market. This is why the manager must control flow and inventory while taking into account delays and different response times.

    One of the most common cash problems for small businesses results from the time lag between the booking of orders, billing, and the receipt of payment. Regular expenses (payroll, purchases, rent) and the irregular receipt of customers' payments together create cash fluctuations. These are eased somewhat by the overdraft privilege that banks grant to some companies. The overdraft exercises a regulatory role like that of inventories, a full backlog of orders, or other reserves: it is the buffering effect we have already encountered, notably in the case of the great reservoirs of the ecological cycle.

    Food and world population. Two major variables measure world growth: industrial capital and population. The reservoir of industrial capital (factories, machines, vehicles, equipment) is filled through investment and emptied through depreciation, obsolescence, and wear and tear on machines and equipment. The population reservoir is filled by births and emptied by deaths (Fig. 58).

    If the flow of investment is equal to the flow of depreciation, or if births equal deaths, a state of dynamic equilibrium is achieved-a stationary (not static) state called zero growth. What will happen then when several flow and state variables interact?

    Consider a simple model, the well-known Malthusian model described in classic form. World resources of food grow at a constant rate (a linear, arithmetic progression), while world population grows at a rate that is itself a function of population (a nonlinear, geometric progression) ( see notes ) (Fig. 59).

    The food reservoir fills at a constant rate, the population reservoir at an accelerated rate. The control element is represented by the quantity of food available to each individual (Fig. 60).

    A decrease in the food quota per person leads to famine and eventually an increase in mortality. The demographic curve stabilizes in an S curve, typical of growth limited by an outside factor (Fig. 61).

    Equations corresponding to various state and flow variables can be programmed on a computer in order to verify the validity of certain hypotheses: what would happen if the birth rate doubled? if it were reduced by half? if food production doubled or tripled? The present example is of only limited interest because it is such a rudimentary model; in the presence of several hundred variables, however, the simulation presents and achieves, as we shall see, valuable results.


    Certainly there has been a revolution in our way of thinking; what now are the practical uses to which we can put it? Beyond the simple description of the systems of nature it leads to new methods and rules of action-nothing less, as you will see, than the instruction manual for the macroscope.

    Analysis and Synthesis

    The analytic and the systemic approaches are more complementary than opposed, yet neither one is reducible to the other.

    The analytic approach seeks to reduce a system to its elementary elements in order to study in detail and understand the types of interaction that exist between them. By modifying one variable at a time, it tries to infer general laws that will enable one to predict the properties of a system under very different conditions. To make this prediction possible, the laws of the additivity of elementary properties must be invoked. This is the case in homogeneous systems, those composed of similar elements and having weak interactions among them. Here the laws of statistics readily apply, enabling one to understand the behavior of the multitude-of disorganized complexity.

    The laws of the additivity of elementary properties do not apply in highly complex systems composed of a large diversity of elements linked together by strong interactions. These systems must be approached by new methods such as those which the systemic approach groups together. The purpose of the new methods is to consider a system in its totality its complexity, and its own dynamics Through simulation one can "animate" a system and observe in real time the effects of the different kinds of interactions among its elements. The study of this behavior leads in time to the determination of rules that can modify the system or design other systems.

    The following table compares, one by one, the traits of the two approaches.

    Analytic Approach Systemic Approach
  • isolates, then concentrates on the elements
  • unifies and concentrates on the interaction between elements
  • studies the nature of interaction
  • studies the effects of interactions
  • emphasizes the precision of details
  • emphasizes global perception
  • modifies one variable at a time
  • modifies groups of variables simultaneously
  • remains independent of duration of time; the phenomena considered are reversible.
  • integrates duration of time and irreversibility
  • validates facts by means of experimental proof within the body of a theory
  • validates facts through comparison of the behavior of the model with reality
  • uses precise and detailed models that are less useful in actual operation (example: econometric models)
  • uses models that are insufficiently rigorous to be used as bases of knowledge but are useful in decision and action (example: models of the Club of Rome)
  • has an efficient approach when interactions are linear and weak
  • has an efficient approach when interactions are nonlinear and strong
  • leads to discipline-oriented (juxtadisciplinary) education
  • leads to multidisciplinary education
  • leads to action programmed in detail
  • leads to action through objectives
  • possesses knowledge of details poorly defined goals
  • possesses knowledge of goals, fuzzy details
  • This table, while useful in its simplicity, is nevertheless a caricature of reality. The presentation is excessively dualist; it confines thought to an alternative from which it seems difficult to escape. Numerous other points of comparison deserve to be mentioned. Yet without being exhaustive the table has the advantage of effectively opposing the two complementary approaches, one of which-the analytic approach-has been favored disproportionately in our educational system.

    To the opposition of analytic and systemic we must add the opposition of static vision and dynamic vision.

    Our knowledge of nature and the major scientific laws rests on what I shall call "classic thought," which has three main characteristics.

    Its concepts have been shaped in the image of a "solid" (conservation of form preservation of volume, effects of force, spatial relations, hardness, solidity).

    Irreversible time, that of life's duration, of the nondetermined, of chance events is never taken into account. All that counts is physical time and reversible phenomena. T can be changed to-T without modifying the phenomena under study.

    The only form of explanation of phenomena is linear causality; that is, the method of explanation relies on a logical sequence of cause and effect that extends for its full dimension along the arrow of time.

    In present modes of thought influenced by the systemic approach, the concept of the fluid replaces that of the solid. Movement replaces permanence. Flexibility and adaptability replace rigidity and stability. The concepts of flow and flow equilibrium are added to those of force and force equilibrium. Duration and irreversibility enter as basic dimensions in the nature of phenomena. Causality becomes circular and opens up to finality.[10]

    The dynamics of systems shatters the static vision of organizations and structures; by integrating time it makes manifest relatedness and development.

    Another table may help to enlighten and enrich the most important concepts related to classic thought and systemic thought (Fig. 62).

    Models and Simulation

    The construction of models and simulations are among the most widely used methods of the systemic approach, to the extent that they are often confused with the systemic approach itself.

    Confronted with complexity and interdependence, we all use simple analogical models. These models, established as part of an earlier analytical approach, seek to unite the main elements of a system in order to permit hypotheses concerning the behavior of the system as a whole- by taking into account as much as possible the interdependence of the factors.

    When the number of variables is small, we constantly use such analogical models to understand a system of which we have little information or to try to anticipate the responses or reactions of someone with a different model of the situation. Our vision of the world is a model. Every mental image is a fuzzy, incomplete model that serves as a basis for decision.

    The construction of simple analogical models rapidly becomes impracticable when large numbers of variables are involved. This is the case with highly complex systems. The limitations of our brain make it impossible for us to make a system "live" without the help of computers and simulation systems, so we turn to these mechanical and electronic means.

    Simulation tries to make a system live by simultaneously involving all its variables. It relies on a model established on the basis of previous analysis. Systems analysis, model building, and simulation are the three fundamental stages in the study of the dynamic behavior of complex systems.

    Systems analysis defines the limits of the system to be modeled, identifies the important elements and the types of interactions between these elements, and determines the connections that integrate the elements into an organized whole. Elements and types of connections are classified and placed in hierarchical order. One may then extract and identify the flow variables, the state variables, positive and negative feedback loops, delays, sources, and sinks. Each loop is studied separately, and its influence on the behavior of the different component units of the system is evaluated.

    Model building involves the construction of a model from data provided by systems analysis. One establishes first a complete diagram of the causal relations between the elements of the subsystem. (In the Malthusian model on ( page 77 ) these include the influences of birth rate on population and food rationing on mortality.) Then, in the appropriate computer language, one prepares the equations describing the interactions and connections between the different elements of the system.

    Simulation considers the dynamic behavior of a complex system. Instead of modifying one variable at a time it uses a computer to set in motion simultaneously groups of variables in order to produce a real- life situation. A simulator, which is an interactive physical model, can also be used to give in "real time" the answers to different decisions and reactions of its user. One such simulator is the flight simulator used by student pilots. Simulation is used today in many areas, thanks to the development of more powerful yet simpler simulation language and new interactive means of communication with the computer (graphic output on cathode ray tubes, high-speed plotters, input light pens, computer-generated animated films).

    Examples of the applications of simulation are to be found in many fields: economy and politics-technological forecasting, simulation of conflicts, "world models"; industrial management-marketing policy, market penetration, launching a new product: ecology-effects of atmospheric pollutants, concentration of pollutants in the food chain; city planning growth of cities, appearance of slums, automobile traffic; astrophysics- birth and evolution of the galaxies, "experiments" produced in the atmosphere of a distant planet; physics-the flow of electrons in a semiconductor, resistance of materials, shock waves, flow of liquids, formation of waves; public works-silting-in of ports, effects of wind on high-rise buildings; chemistry-simulation of chemical reactions, studies of molecular structure; biology-circulation in the capillaries, competitive growth between bacterial populations, effects of drugs, population genetics, data processing-simulation of the function of a computer before its construction; operational research-problems of waiting lines, optimization, resource allocation, manufacturing control; engineering-process control, calculations of energy costs, calculations of construction costs; education -simulated pedagogical practices, business games.

    Despite the number and diversity of these applications, too much cannot be expected of simulation. It is only one approach among many, a complementary method of studying a complex system. Simulation never gives the optimum or the exact solution to a given problem. It only sets forth the general tendencies of the behavior of a system -its probable directions of evolution - while suggesting new hypotheses. ( see notes )

    One of the serious dangers of simulation results from too much freedom in the choice of variables. The user can change the initial conditions "just to see what will happen." There is the risk of becoming lost in the infinity of variables and the incoherent performances associated with chance modifications. The results of simulation must not be confused with reality (as is often the case) but, compared with what one knows of reality, should be used as the basis for the possible modification of the initial model. When one continues to use such a process in successive approximations, the usefulness of simulation will become apparent.

    Simulation appears to be one of the most resourceful tools of the systemic approach. It enables us to verify the effects of a large number of variables on the overall functioning of a system; it ranks the role of each variable in order of importance; it detects the points of amplification or inhibition through which we can influence the behavior of the system. The user can test different hypotheses without running the risk of destroying the system under study-a particularly important advantage in the case of living systems or those that are fragile or very costly.

    Knowing that one can experiment on a model of reality rather than on reality itself, one can influence the time variable by accelerating very slow phenomena (social phenomena, for example) or slowing down ultrafast phenomena (the impact of a projectile on a surface). One can influence equally well the space variable by simulating the interactions that occur in very confined volumes or over great distances.

    Simulation does not bring out of the computer, as if by magic, more than what was put into the program. The contribution of the computer rests at a qualitative level. Handling millions of bits of information in a tiny fraction of time, it reveals structures, modes, and tendencies heretofore unobservable and which result from the dynamic behavior of the system.

    Interaction between user and model develops a feeling of the effect of interdependencies and makes it possible to anticipate better the reactions of the models. Evidently this feeling exists for all those who have had long experience in the management of complex organizations. One of the advantages of simulation is that it allows the more rapid acquisition of these fundamental mechanisms.

    Finally, simulation is a new aid to decision making. It enables one to make choices among "possible futures." Applied to social systems, it is not directly predictive. (How does one take into account such impossible-to-quantify data as well-being, fear, desire, or affective reactions?) Yet simulation does constitute a sort of portable sociological laboratory with which experiments can be made without involving the future of millions of men and without using up important resources in programs that often lead to failure.

    Certainly models are still imperfect. As Dennis Meadows observed, however, the only alternatives are "mental models" made from fragments of elements and intuitive thinking ( see notes ). Major political decisions usually rest on such mental models.

    The Dynamics of Maintenance and Change

    The properties and the behavior of a complex system are determined by its internal organization and its relations with its environment. To understand better these properties and to anticipate better its behavior,

    it is necessary to act on the system by transforming it or by orienting its evolution.

    Every system has two fundamental modes of existence and behavior: maintenance and change. The first, based on negative feedback loops, is characterized by stability. The second, based on positive feedback loops, is characterized by growth (or decline.) The coexistence of the two modes at the heart of an open system, constantly subject to random disturbances from its environment, creates a series of common behavior patterns. The principal patterns can be summarized in a series of simple graphs by taking as a variable any typical parameter of the system (size, output, total sales, number of elements) as a function of time (Fig. 63).[11]

    Dynamic stability: equilibrium in motion. Maintenance is duration. Negative controls, by regulating the divergences of positive loops, help to stabilize a system and enable it to last. Thus the system is capable of self-regulation.

    To bring together stability and dynamics might seem to be paradoxical. In fact the juxtaposition demonstrates that the structures or the functions of an open system remain identical to themselves in spite of the continuous turnover of the components of the system. This persistence of form is dynamic stability. It is found in the cell, in the living organism, in the flame of a candle.

    Dynamic stability results from the combination and readjustment of numerous equilibriums attained and maintained by the system-that of the "internal milieu" of the living organism, for example (see p. 42). We deal with dynamic equilibriums; this imposes a preliminary distinction between balance of force and balance of flow.

    A balance of force results from the neutralization at the same point of two or more equal and opposed forces. This might be illustrated by two hands immobilized in handcuffs or by a ball lying in the bottom of a basin (Fig. 64).

    When there are two forces present-two armies or two governments- we speak of the "balance of power." But a balance of force is a static equilibrium; it can be modified only as the result of a discontinuous change in the relationship of the forces. This discontinuity could lead to an escalation when one force overpowers the other.

    On the other hand, a balance of flow results from the adjustment of the speeds of two or more flows crossing a measuring device. Equilibrium exists when the speeds of the flows are equal and moving in opposite directions.[12] This is the case of a transaction at a sales counter, where merchandise is exchanged for money (Fig. 65).

    A balance of flow is a dynamic equilibrium. It can be adapted, modified, and modeled permanently by often imperceptible readjustments, depending on disturbances or circumstances. The balance of flow is the foundation of dynamic stability.

    When equilibrium is achieved, a given "level" is maintained over time (like the concentration of certain molecules in the plasma, or the state of a bank account ( see page 74 ). This particular state is called a steady state; it is very different from the static state represented by the level of water in a reservoir having no communication with the environment (Fig. 66).

    There are as many steady states as there are levels of equilibrium at different depths of a reservoir. This makes it possible for an open system to adapt and respond to the great variety of modifications in the environment.

    Homeostasis: resistance to change. Homeostasis is one of the most remarkable and most typical properties of highly complex open systems. The term was created by the American physiologist Walter Cannon in 1932 ( see page 43 ). A homeostatic system (an industrial firm, a large organization, a cell) is an open system that maintains its structure and functions by means of a multiplicity of dynamic equilibriums rigorously controlled by interdependent regulation mechanisms. Such a system reacts to every change in the environment, or to every random disturbance, through a series of modifications of equal size and opposite direction

    to those that created the disturbance. The goal of these modifications is to maintain the internal balances.

    Ecological, biological, and social systems are homeostatic. They oppose change with every means at their disposal. If the system does not succeed in reestablishing its equilibriums, it enters into another mode of behavior, one with constraints often more severe than the previous ones. This mode can lead to the destruction of the system if the disturbances persist.

    Complex systems must have homeostasis to maintain stability and to survive. At the same time it bestows on the systems very special properties. Homeostatic systems are ultrastable; everything in their internal, structural, and functional organization contributes to the maintenance of the same organization. Their behavior is unpredictable; "counterintuitive" according to Jay Forrester, or contravariant: when one expects a determined reaction as the result of a precise action, a completely unexpected and often contrary action occurs instead. These are the gambles of interdependence and homeostasis; statesmen, business leaders, and sociologists know the effects only too well.

    For a complex system, to endure is not enough; it must adapt itself to modifications of the environment and it must evolve. Otherwise outside forces will soon disorganize and destroy it. The paradoxical situation that confronts all those responsible for the maintenance and evolution of a complex system, whether the system be a state, a large organization, or an industry, can be expressed in the simple question, How can a stable organization whose goal is to maintain itself and endure be able to change and evolve?

    Growth and Variety. The growth of a complex system-growth in volume, size, number of elements-depends on positive feedback loops and the storage of energy. In effect a positive feedback loop always acting in the same direction leads to the accelerated growth of a given value ( see page 73 ). This value can be number (population growth), diversity (variety of elements and interactions between elements), or energy (energy surplus, accumulation of profits, growth of capital).

    The positive feedback loop is equivalent to a random variety generator. It amplifies the slightest variation; it increases the possibilities of choice, accentuates differentiation, and generates complexity by increasing the possibilities for interaction.

    Variety and complexity are closely allied. Variety, however, is one of the conditions for the stability of a system. In fact homeostasis can be established and maintained only when there is a large variety of controls. The more complex a system, the more complex its control system must be in order to provide a "response" to the multiple disturbances produced by the environment. This is the law of requisite variety proposed by Ross Ashby in 1956 ( see notes ). This very general law asserts in mathematical

    form that the regulation of a system is efficient only when it depends on a system of controls as complex as the system itself . In other words, control actions must have a variety equal to the variety of the system. In ecology, for example, it is the variety of species, the number of ecological niches, the abundance of interactions among species and between community and environment that guarantee the stability and continuance of the community. Variety permits a wider range of response to potential forms of aggression from the environment.

    The generation of variety can lead to adaptations through increase in complexity. But in its confrontation with the random disturbances of the environment, variety also produces the unexpected, which is the seed of change. Growth is then both a force for change and a means for adapting to the modifications of the environment. Here one begins to see the way in which a homeostatic system can evolve as a system constructed to resist change. It evolves through a complementary process of total or partial disorganization and reorganization. This process is produced either by the confrontation of the system with random disturbances from the environment (mutations, events, "noise") or in the course of readjustment of an imbalance (resulting, for example, from too rapid growth).

    Evolution and emergence. Living systems can adapt, within certain limits, to sudden modifications coming from the outside world. A system actually has detectors and comparators that enable it to detect signals from within or without and to compare the signals to equilibrium values. When there are discrepancies, the emission of error signals can help to correct them. If it cannot return to its former state of homeostatic equilibrium, the system, through the complementary play of positive and negative feedback loops, searches for new points of equilibrium and new stationary states.

    The evolution of an open system is the integration of these changes and adaptations, the accumulation in time of successive plans or "layers" of its history.[13] This evolution materializes through hierarchical levels of organization and the emergence of new properties. The prebiological evolution (the genesis of living systems) and the biological and social evolutions are examples of evolution toward levels of increasing complexily. At each level new properties "emerge" that cannot be explained by the sum of the properties of each of the parts which constitute the whole. There is a qualitative leap; the crossing of a threshold; life, reflective thought, and collective consciousness.

    Emergence is linked to complexity. The increase in the diversity of elements, in the number of connections between these elements, and in the play of nonlinear interactions leads to patterns of behavior that are difficult to predict - especially if they are founded solely on the properties of the elements. We know, for example, the properties of each of the amino acids that make up the protein chain. But because of the convolutions of this chain, certain amino acids that are far apart in the sequence find themselves together in space. This situation gives the protein emergent properties that enable it to recognize certain molecules and to catalyze their transformation. This would be impossible if the amino acids were present in the milieu but not arranged in the proper order-or if the chain were straightened out.

    The "Ten Commandments" of the Systemic Approach

    The systemic approach has little value if it does not lead to practical applications such as facilitating the acquisition of knowledge and improving the effectiveness of our actions. It should enable us to extract from the properties and the behavior of complex systems some general rules for understanding systems better and acting on them.

    Unlike the juridical, moral, or even physiological laws which one might still cheat, a misappreciation of some of the basic systemic laws could result in serious error and perhaps lead to the destruction of the system within which one is trying to act. Of course many people will have an intuitive knowledge of these laws, which are very much the result of experience or simple common sense. The following are the "ten commandments" of the systemic approach.

    1. Preserve variety. To preserve stability one must preserve variety. Any simplification is dangerous because it introduces imbalance. Examples abound in ecology. The disappearance of some species as a consequence of the encroaching progress of "civilization" brings the degradation of the entire ecosystem. In some areas intensive agriculture destroys the equilibrium of the ecological pyramid and replaces it with an unstable equilibrium of only three stages (grain, cattle, and man) controlled by a single dominant species. This unbalanced ecosystem tries spontaneously to return to a state of higher complexity through the proliferation of insects and weeds-which farmers prevent by the widespread use of pesticides and herbicides.

    In economy and in management, excessive centralization produces a simplification of communication networks and the impoverishment of the interactions between individuals. There follow disorder, imbalance, and a failure to adapt to rapidly changing situations.

    2. Do not "open" regulatory loops. The isolation of one factor leads to prompt actions, the effects of which often disrupt the entire system. To obtain a short-term action, a stabilizing loop or an overlapping series of feedback loops is often "cut open"-in the belief that one is acting directly on the causes in order to control the effects. This is the cause of sometimes dramatic errors in medicine, economy, and ecology.

    Consider some examples of what happens in the rupture of natural cycles. The massive use of fossil fuels, chemical fertilizers, or nonrecyclable pesticides allows the agricultural yield to grow in the short term; in the long term this action may bring on irreversible disturbances. The fight against insects leads as well to the disappearance of the birds that feed on the insects; the result in the long term is that the insects return in full force-but there are no birds. The states of waking, sleeping, and dreaming are probably regulated by the delicate balance between chemical substances that exist in the brain; by regularly introducing, for short-term effect, an outside foreign molecule such as a sleeping pill, the natural long-term mechanisms are inhibited-worse, there is the danger of upsetting them almost irrevocably: persons accustomed to using barbiturates must undergo a veritable detoxification in order to return to a normal sleep pattern.

    3. Look for the points of amplification. Systems analysis and simulation bring out the sensitive points of a complex system. By acting at this level, one releases either amplifications or controlled inhibitions.

    A homeostatic system resists every measure, immediate or sequential (that is, waiting for the results of preceding measures in order to take on new ones). One of the methods that influence the system and cause it to evolve in a chosen direction is the use of a policy mix. These measures must be carefully proportioned in their relationships and applied simultaneously at different points of influence.

    One example is the problem of solid wastes. There are only three ways to reduce the flow of the generation of solid wastes by acting on the valve (the flow variable): reducing the number of products used (which would mean a drop in the standard of living), reducing the quantity of solid wastes in each product, or increasing the life expectancy of the products by making them more durable and easier to repair. The simulations performed by Jorgan Randers of MIT show that no one measure alone is enough ( see notes ). The best results came from a policy mix, a combination of measures used at the same time: a tax of 25 percent on the extraction of nonrenewable resources, a subsidy of 25 percent for recycling, a 50 percent increase in the life of the products, a doubling of the recyclable portion per product, and a reduction in primary raw material per product (Fig. 67).

    4. Reestablish equilibriums through decentralization. The rapid reestablishment of equilibriums requires the detection of variances where they occur and corrective action that is carried out in a decentralized manner.

    The correction of the body's equilibrium when we stand is accomplished by the contraction of certain muscles without our having to think about it even when the brain intervenes. Enzymatic regulation networks show that the entire hierarchy of levels of complexity intervene in the reestablishment of balance ( recall the example of the service station on page 51 ). Often corrective action has been taken even before one has been made conscious of taking it. The decentralization of the reestablishment of equilibriums is one application of the law of requisite variety. It is customary in the body, the cell, the ecosystem. But so far it appears that we have not succeeded in applying this law to the organizations that we have been assigned to manage.

    5. Know how to maintain constraints. A complex open system can function according to different modes of behavior. Some of them are desirable; others lead to the disorganization of the system. If we want to maintain a given behavior that we consider preferable to another, we must accept and maintain certain kinds of constraints in order to keep the system from turning toward a less desirable or a dangerous mode of behavior.

    In the management of the family budget one can choose a high style of living (living beyond one's means), with the constraints that it implies with respect to banks and creditors. Or one can choose to limit expenditures and do without goods one would like to possess-a different set of constraints.

    In the case of a nation's economy, those responsible for the economic policy choose and maintain the constraints that result from inflation with all their injustices and social inequalities-for they are judged a lesser evil than those brought about by unemployment.

    At the level of the world economy the growth race entails social inequalities, depletion of resources, and pollution. Theoretically, however, it allows a more rapid increase in the standard of living. The transition to a "stationary" economy would imply the choice of new constraints, founded on privation and a reduction in the standard of living and the imposition of more complex, more delicate, and more decentralized forms of control and regulation than in a growth economy. These means would call for increased responsibility on the part of each citizen.

    Liberty and autonomy are achieved only through the choice and application of constraints; to want to eliminate constraints at any price is to risk moving from an accepted and controlled state of constraint to an uncontrollable state that will lead rapidly to the destruction of the system.

    6. Differentiate to integrate better. Every real integration is founded on a previous differentiation. The individuality, the unique character of each element is revealed in the organized totality. This is the meaning of Teilhard de Chardin's famous phrase, "union differentiates." This law of the "personalizing" union is illustrated by the specialization of cells in the tissues or the organs of the body.

    There is no true union without antagonism, balance of power, conflict. Homogeneity, mixture, and syncretism are forms of entropy. Only union through diversity is creative; it increases complexity and leads to higher levels of organization. This systemic law and its allied constraints are well known by those whose purpose is to unite, to assemble, to federate. Antagonism and conflict are always born of the transition to a unified entity. Before regrouping diversities, we must decide to what limits we should push the process of personalization. Pushed too soon, it leads to an homogenizing and paralyzing mixture; pushed too late, it leads to the confrontation of individualism and personality-and perhaps a disassociation still greater than what had formerly existed.

    7. To evolve, allow aggression. A homeostatic (ultrastable) system can evolve only if it is assaulted by events from the world outside. An organization must then be in a position to capture the germs of change and use them in its evolution-which obliges it to adopt a mode of functioning characterized by the renewal of structures and the mobility of men and ideas. In effect all rigidity, sclerosis, and perpetuity of structures or hierarchy is clearly opposed to a system that allows evolution ( see notes ).

    An organization can maintain itself in the manner of a crystal or that of a living cell. The crystal preserves its structure by means of the balance of forces that cancel out each other in every node of the crystalline network-and by redundancy, or repetition of patterns. This static state, closed to the environment, allows no resistance to change within its milieu: if the temperature rises, the crystal becomes disorganized and melts. The cell, however, is in dynamic equilibrium with its environment. Its organization is founded not on repetition but on the variety of its elements. An open system, it maintains a constant turnover of its elements. Variety and mobility enable it to adapt to change.

    The crystal-like organization evolves slowly in the give and take of radical and traumatic reforms. The cell-like organization tries to make the most of events, variety, and the openings into the outside world. It is not afraid of a passing disorganization-the most efficient condition for readaptation. To accept this transitory risk is to accept and to want change. For there is no real change without risk.

    8. Prefer objectives to detailed programming. The setting of objectives and rigorous control-as opposed to detailed programming at every step-is what differentiates a servomechanism from a rigidly programmed automatic machine. The programming of the machine must foresee all disturbances likely to occur in the course of operation. The servomechanism, however, adapts to complexity; it needs only to have its goal set without ambiguity and to establish the means of control that will enable it to take corrective measures in the course of action.

    These basic principles of cybernetics apply to every human organization. The definition of objectives, the means of attaining them, and the determination of deadlines are more important than the detailed programming of daily activities. Minutely detailed programming runs the risk of being paralyzing; authoritarian programming leaves little room for imagination and involvement. Whatever roads are taken, the important thing is to arrive at the goal-provided that the well-defined limits (necessary resources and total time allotted to operations) are not exceeded.

    9. Know how to use operating energy. Data sent out by a command center can be amplified in significant proportions, especially when the data are relayed by the hierarchical structures of organizations or by diffusion networks.

    At the energy level the metabolism of the operator of a machine is negligible compared to the power that he can release and control. The same applies to a manager or to anyone in charge of a large organization. We must distinguish, then, between power energy and operating energy. Power energy is represented by the electric line or the current that heats a resistance; or it may be the water pipe that carries water pressure to a given point. Operating energy renders itself in the action of the thermostat or the water tap: it represents information.

    A servomechanism distributes its own operating energy through the distribution of information that commands its operational parts. In the same way the leader of an organization must help his own system to distribute its operating energy. To accomplish this he establishes feedback loops to the decision centers. In the management of an industry or in the structure of a government, these regulatory loops are called selfmanagement (autogestion), participation, or social feedback.[14]

    10. Respect response times. Complex systems integrate time into their organization. Each system has a response time characteristic of that system, by reason of the combined effects of feedback loops, delays at reservoirs, and the sluggishness of flows. In many cases, especially in industry, it is useless to look for speed of execution at any price, to exert pressure in order to obtain responses or results. It is better to try to understand the internal dynamics of the system and to anticipate delays in response. This type of training is often acquired in the actual running of large organizations. It gives rise to a sense of timing, the knowing when to begin an action, neither too soon nor too late, but at the precise moment the system is ready to move in one direction or the other. Sense of timing allows the best possible use of the internal energy of a complex system-rather than to have to impose instructions from outside against which the system will react.

    Avoiding the Dangers of the Systemic approach

    To be useful, the systemic approach must be demystified; what is useful in daily life must not be reserved for a small elite. The hierarchy of disciplines established in the nineteenth century, from the "most noble" sciences (mathematics and physics) to the "least noble" (the sciences of man and society), continues to weigh heavily on our approach to nature and our vision of the world. Skepticism or distrust of the systemic approach is found among those-mathematicians and physicists-who have received the most advanced theoretical training. At the same time, those who by nature of their research have been accustomed to think in terms of flow, transfer, exchange, and irreversibility-biologists, economists, and ecologists-assimilate more naturally the systemic concepts and communicate more easily among themselves.

    To demystify further the systemic approach and to enable it to remain a transdisciplinary attitude, a training in the mastery of complexity and interdependence, it may be necessary to get rid of the very terms systemic approach and systemic method. The global vision is not reserved for the few with wide responsibility-the philosophers and the scientists. Each one of us can see things in perspective. We must learn to look through the macroscope to apply systemic rules, to construct more rigorous mental models, and perhaps to master the play of interdependencies.

    And we must not hide the dangers of a too systematic use of the systemic approach. A purely descriptive approach-the "what is linked to what?" method-leads rapidly to a collection of useless models of the different systems of nature. The greatest generalization of the concept of system can also turn against itself, destroying its fecundity in sterilizing platitude. In the same way the uncontrolled use of analogies, homologies, and isomorphisms can result in interpretations that complicate rather than enlighten. Such interpretations are founded on superficial resemblances rather than on principles and fundamental laws that are common to all systems. According to Edgar Morin, "too much unification can become abusive simplification, then an idée fixe or a turn of phrase" ( see notes ).

    Once again we are encumbered with the danger of dogmatism. The systemic approach leads to an intransigent systematism or a reductionist biologism. There is danger of our being seduced by models that were conceived as ends of reflective thought, not as points of departure for research. We are tempted by the too simplistic transposition of models or biological laws to society.[15] The cybernetics of regulation at the molecular level offers general models, some aspects of which are transposable, with certain restrictions, to social systems. The greatest weakness of these models is that they apparently cannot take into account the relationship between force and the conflicts that arise between the elements of every socioeconomic system. The economist J. Attali remarked on this at a meeting of the Group of Ten devoted to the maintenance of biological and social equilibriums: "Unlike the sociologist, the biologist observes systems with well-established laws: they do not change as they are being studied. As for molecules, cells, or microbes, they will never complain of their condition!"

    One of the greatest dangers that menace the systemic approach is the temptation of the "unitary theory," the all-inclusive model with all the answers and the ability to predict everything. The use of mathematical language, which by nature and vocation generalizes, can lead to a formalism that isolates the systemic approach instead of opening it up to the practical. The General System Theory does not escape this danger. Sometimes it becomes locked into the language of graph theory, set theory, game theory, or information theory; sometimes it is nothing more than a collection of descriptive approaches that are often illuminating but have no practical application.

    The functional systemic approach offers one way of bypassing these alternatives. It avoids the dangerous stumbling blocks of paralyzing reductionism and total systematism; it clears the way for the communication of knowledge, for action, and for creation. For the communication of knowledge because the systemic approach has a conceptual framework

    of reference that helps to organize knowledge as it is acquired, reinforces its memorization, and facilitates its transmission. For action because the systemic approach provides rules for confronting complexity and because it assigns to their hierarchical order the elements that are the basis for decisions. And for creation because the systemic approach catalyzes imagination, creativity, and invention. It is the foundation of inventive thought (where the analytical approach is the foundation of knowledgeable thought). Tolerant and pragmatic, systemic thought is open to analogy, metaphor, and model-all formerly excluded from "the scientific method" and now rehabilitated. Everything that unlocks knowledge and frees imagination is welcomed by the systemic approach; it will remain open, like the systems it studies.

    The earth shelters the embryo of a body and the beginnings of a spirit. The life of this body is maintained by the great ecological and economic functions brought together in the ecosphere. Collective consciousness emerges from the simultaneous communication of men's brains; it constitutes the noosphere ( see notes ).

    Ecosphere and noosphere have energy and information as a base. Action is the synthesis of energy and information. But all action requires time. Thus time is the link between energy, information, and action. The following chapters will be devoted to such a global approach toward energy, information, and time-through trying to envisage old problems from a new perspective.

    [1] Bionics attempts to build electronic machines that imitate the functions of certain organs of living beings.

    [2] IBM subsequently used such memories in all its computers. This type of memory (for which Forrester still holds all major patents) is in the process of being replaced by semiconductor memories. (The former type is still found in most computers today.

    [3] SAGE: Semi-Automatic Ground Equipment.

    [4] The word was first used by Plato in the sense of "the art of steering" or "the art of government." In 1834 Ampere used the word cybernetics to denote "the study of ways of governing."

    [5] I do not consider here systems of concept or mechanical systems run by man, but instead systems of high complexity, such as living, social, or ecological systems.

    [6] The variety defined by W. Ross Ashby is "the number of different elements that make up a system or the number of different relationships between these elements or the number of different states of these relationships." The variety of a relatively simple system, made up of seven elements connected by two-way relationships and having two different sets of conditions, will be expressed by the enormous number of 242. What can be said of these interactions woven together in the heart of the cellular population ( see page 48 ) and in much greater number in the heart of society?

    [7] This symbolic representation was inspired by the one developed by Jay Forrester and his group at MIT in simulation models ( see notes ).

    [8] Other examples include the controls of population, prices, and the balance of payments ( see page 10 , page 25 and page 27 ).

    [9] This state of equilibrium is accomplished even though the account is emptied and refilled every month. (One could assume that in effect wages were being paid and deposited daily.)

    [10] Numerous points only mentioned here will be taken up again in the following chapters.

    [11] It must be remembered that the overall behavior of the system is the result of the individual behavior patterns of subsystems, patterns themselves determined by the interconnection of a large number of variables.

    [12] Or when flows have opposite effects even though they move in the same direction (as a reservoir filling and emptying at the same time).

    [13] Mechanisms of evolution will be introduced in the fifth chapter.

    [14] Social feedback will be discussed in the fourth chapter.

    [15] The danger of too direct transpositions from the biological to the social realms were clearly perceived by Friedrich Engels when he wrote to the Russian sociologist and journalism Piotr Lavrov in 1875: "The essential difference between human socially and animal socially is that animals, al best, collect while men produce This unique but major difference prohibits in its own right the transfer pure and simple of the laws of animal societies to the social systems of men '' ( see notes ). The work of A. J. Lotka in 1925 on the dynamics of population and the work of V. Volterra in 1931 on he mathematical theory of the life struggle have subsequently shown that we must be less dogmatic than Engels with respect to transfers from the biological to the social realms.