Systems Language PDF

The Systems Language 1 The more we study the major problems of our time, the more we come to realise that they cannot be understood in isolation. They are systemic problems, which means that they are interconnected and interdepende...

The Systems Language 1 The more we study the major problems of our time, the more we come to realise that they cannot be understood in isolation. They are systemic problems, which means that they are interconnected and interdependent. Capra (1996) 1.1 INTRODUCTION Simply de¢ned, a system is a complex whole the functioning of which depends on its parts and the interactions between those parts. Stated like this, it is clear that we can identify systems of very di¡erent types:. physical, such as river systems;. biological, such as living organisms;. designed, such as automobiles;. abstract, such as philosophical systems;. social, such as families;. human activity, such as systems to ensure the quality of products. The traditional, scienti¢c method for studying such systems is known as reductionism. Reductionism sees the parts as paramount and seeks to identify the parts, understand the parts and work up from an understanding of the parts to an understanding of the whole. The problem with this is that the whole often seems to take on a form that is not recognizable from the parts. The whole emerges from the interactions between the parts, which a¡ect each other through complex networks of relationships. Once it has emerged, it is the whole that seems to give meaning to the parts and their 4 The systems language interactions. A living organism gives meaning to the heart, liver and lungs; a family to the roles of husband, wife, son, daughter. It is not surprising therefore that there exists an alternative to reduction- ism for studying systems. This alternative is known as holism. Holism con- siders systems to be more than the sum of their parts. It is of course interested in the parts and particularly the networks of relationships between the parts, but primarily in terms of how they give rise to and sustain in existence the new entity that is the whole ^ whether it be a river system, an automobile, a philosophical system or a quality system. It is the whole that is seen as important and gives purpose to the study. Holism gained a foothold in many di¡erent academic disciplines, bene¢t- ing from the failure of reductionism to cope with problems of complexity, diversity and change in complex systems. In what follows we look at the encounter of holism with philosophy, biology, control engineering, organ- ization and management theory, and the physical sciences. We see how the systems language associated with holism was developed and enriched in each case. Particularly fruitful were the encounters with biology and control engineering, which gave birth to systems thinking as a transdisci- pline, studying systems in their own right, in the 1940s and 1950s. This pro- duced a language that describes the characteristics that systems have in common, whether they are mechanical, biological or social. In a conclusion to the chapter I seek to explain why this language is particularly powerful for the purposes of managers. More detailed accounts of the development of holistic thinking can be found in Checkland (1981) and Jackson (2000). 1.2 PHILOSOPHY The classical Greek philosophers, Aristotle and Plato, established some important systems ideas. Aristotle reasoned that the parts of the body only make sense in terms of the way they function to support the whole organism and used this biological analogy to consider how individuals need to be related to the State. Plato was interested in how the notion of control, or the art of steersmanship (kybernetes), could be applied both to vessels and the State. Ships had to be steered safely toward harbour by a helmsman. A similar role needed to be ful¢lled in societies if they were to prosper. Holism was pushed to the margins of philosophical debate for many cen- turies, but the golden age of European philosophy, during the 18th and 19th centuries, saw a renewed interest in what it had to o¡er. Kant and Biology 5 Hegel were particularly in£uential in this respect. Kant was an ‘idealist’ who argued that we could never really know reality or whether it was systemic. However, he believed it was helpful for humans to think in terms of wholes emerging from and sustained by the self-organization of their parts. Hegel introduced process into systems thinking. An understanding of the whole, or the truth, could be approached through a systemic unfolding of thesis, antithesis and synthesis. Each movement through this cycle, with the synthesis becoming the new thesis, gradually enriched our grasp of the whole. It was these philosophical ideas that impacted on the scienti¢c disciplines, where they were given a more rigorous formulation. 1.3 BIOLOGY The fruitfulness of the relationship between holism and biology can be accounted for by the complexity of the problems encountered by biologists in trying to understand whole organisms. Whole organisms seemed to resist the attempts of scienti¢c reductionists to reduce them to the sum of their parts. In the 1920s and 1930s, as a response to this, more holistically inclined biologists began to argue that organisms were more than the sum of their parts. They conceived that a hierarchy existed in nature ^ molecules, organelles, cells, organs, organisms ^ and, at certain points in the hierarchy, stable levels of organized complexity arose that demonstrated emergent properties, which did not exist at levels below. An organism was one such level. It was argued that an organism (e.g., an animal) had a clear boundary separating it from its environment and was capable, as its main emergent property, of a degree of autonomy. An organism sustained itself in a steady state by carrying out transactions across this boundary with its environment. It had to be capable of making internal transformations to ensure that it was adapted to its environment. The processes that maintained the steady state were referred to as homeostatic, an example being the self-regulating mechanism controlling body temperature. The behaviour of an organism could not, it seemed, be explained by the properties of its parts in isolation. It arose from the particular interdependence of the parts, which gave rise to a new level of organized complexity. Biology was seen exactly as the science appropriate to this level and could not therefore be reduced to physics or chemistry. Ludwig von Bertalan¡y has become the best known of the biologists who argued that organisms should be studied as complex wholes. In 1950 6 The systems language he published an article in which be made the well-known distinction between closed systems and open systems. A closed system engages in no exchanges with its environment. An open system, such as an organism, has to interact with its environment to maintain itself in existence. Open systems take inputs from their environments, transform them and then return them as some sort of product back to the environment. They depend on the environ- ment for their existence and adapt in reaction to changes in the environment. Von Bertalan¡y’s lasting fame and in£uence has derived from his sugges- tion that the sorts of behaviour he witnessed in open systems in biology could be seen demonstrated by open systems in other domains. Thus, he initiated and named ‘general system theory’ (see von Bertalan¡y, 1968) ^ a kind of transdiscipline in which systems were studied in their own right and which allowed insights from one discipline to be transferred to others. General system theory was soon embraced by management thinkers who transferred the open system model to their study of organizations. The biological system model is represented in Figure 1.1. It shows a system separated from its environment by a distinct boundary. The system has a complex structure, being di¡erentiated into subsystems that themselves have parts (systems arranged in a hierarchy of systems). The close interrelationships of mutual in£uence between the subsystems must ensure homeostasis ^ the maintenance of a steady state. One subsystem is acting in a kind of ‘management’ capacity, trying to ensure integration and co-ordination. The system takes inputs of material, energy and information Figure 1.1 The biological system model. Control engineering 7 from its environment, uses some to sustain itself and transforms the rest into outputs. These outputs may themselves allow the system to secure, through a cycle of events, more of the useful inputs it needs to survive. The open systems perspective propounded by von Bertalan¡y, and so in- £uential in the 1970s and 1980s, has more recently been challenged by the biologists Maturana and Varela (1980). They emphasize instead the closed system of interactions that occurs in living entities. These interactions ensure the self-production of the system and its autonomy. Such self- producing, or autopoietic (from the ancient Greek for self-production), systems respond to environmental disturbances, but not directly or simply; the nature of the response depends on their own internal organizational ar- rangements. This does not mean that autopoietic systems cannot change their structure, but it does mean that they do this only with a view to keeping their fundamental organizational identity intact. The emphasis on the circular organization of living systems, and their resistance to change, o¡ers a useful corrective to those general system theorists who stress the overriding importance of organization^environment relations. 1.4 CONTROL ENGINEERING The other ¢gure who stands alongside von Bertalan¡y, as a founding father of systems thinking as a transdiscipline, is Norbert Wiener, a mathematician and control engineer. In 1948 Wiener published a book on what he called, borrowing from the Greek, cybernetics ^ the science of control and communication in the animal and the machine. Cybernetics, Wiener argued, was a new science that had application to many di¡erent disciplines because it dealt with general laws that governed control processes whatever the nature of the system under consideration. The two key concepts introduced by Wiener into the systems lexicon were control and communication. In understanding control, whether in the mech- anical, biological or political realm, the idea of negative feedback is crucial. This concept allows a proper, scienti¢c explanation to be given of purposive behaviour ^ behaviour directed to the attainment of a goal. It was Wiener’s insight that all such behaviour requires negative feedback. In this process, information is transmitted about any divergence of behaviour from a present goal and corrective action taken, on the basis of this information, to bring the behaviour back towards the goal. In a central heating system a thermostat monitors the heat of a room against some preset temperature and uses the information that the temperature is too low or high to switch 8 The systems language Figure 1.2 A negative feedback system. the system on or o¡. Communication is equally signi¢cant because if we wish to control the actions of a machine or another human being we must communicate information to that machine or individual. Figure 1.2 shows a simple, negative feedback system. It operates by sensing the current output of the process that is to be controlled. The output is compared with the desired goal and, if it diverges from this, an activator adjusts the input to bring the process back toward achieving the desired goal. In this way, systems regulate themselves and are controlled, in the face of environmental disturbances, through the e¡ective communication of information. It is of course very important that the sensor and comparator operate continuously and rapidly. This ensures that discrepancies are identi- ¢ed at the earliest possible opportunity and corrective action can immediately be initiated. It is also worth noting that it is not necessary to understand the nature of the process, which might be a complex system, in order to employ the negative feedback device. The controller can regard it as a ‘black box’ and adjust it simply by manipulating the inputs in order to achieve the desired outputs. Organization and management theory 9 Although it did not impinge much on the consciousness of Wiener, another form of feedback, positive feedback, has become signi¢cant for systems thinking. While negative feedback counteracts deviations from a goal, positive feedback ampli¢es them. For example, one mistimed tackle in a soccer match can lead to a series of deliberate fouls, escalating into uncon- trolled aggression from both sides. Identifying situations where the parts of a system are locked into a positive feedback loop, and its behaviour is spinning out of control, is of obvious signi¢cance to managers. A good referee can re-establish order with the astute use of a yellow card. A ¢nal systems concept that I need to introduce in this section is ‘variety’. Variety is a term ¢rst used by Ashby (1956) to refer to the number of possible states a system can exhibit. According to Ashby’s law of requisite variety, systems can only be controlled if the would-be controller can command the same degree of variety as the system. Today, systems are complex and change rapidly; they exhibit high variety. Managers need to pay attention to reducing the variety of the system they are seeking to control and/or to in- creasing their own variety. This process of ‘balancing varieties’ is known as variety engineering. We shall see how it is done in Chapter 6. 1.5 ORGANIZATION AND MANAGEMENT THEORY Early attempts to marry holism with organization and management theory took two main forms. In the ¢rst some basic systems concepts were incorpo- rated in the prevailing scienti¢c management tradition to yield optimizing approaches, such as systems engineering. In the second there was a wholesale transfer of the biological analogy, especially as re¢ned by von Bertalan¡y, to yield systems models of organization emphasizing the importance of subsystems to overall organizational e¡ectiveness and the signi¢cance of the organization^environment ¢t. Both these early attempts met with di⁄culties because they failed to recognize that systems containing human beings are, what we now call, purposeful. The systems of components that engineers are used to dealing with are purposive ^ designed to reach the goal speci¢ed by the engineers. Biological systems are adept at survival, but if this is their purpose it is obviously something ascribed to them from the outside and not something they think about themselves. The parts of social systems however ^ human beings ^ can generate their own purposes from inside the system, and these might not correspond at all to any purposes prescribed by managers or 10 The systems language outsiders. Social and organizational systems, therefore, have multiple pur- poses: they are purposeful. It was soon clear that a di¡erent kind of terminology would be useful for describing and working with purposeful systems. A number of roles had to be delimited relevant to purposeful systems and re£ecting some alternative sources of purposes. The term ‘stakeholder’ is used to refer to any group with an interest in what the system is doing. Decision-makers or owners have the power to make things happen in systems; actors carry out basic tasks; customers or clients bene¢t or su¡er from what a system does. Problem-owners worry about the performance of some aspect of a system. Witnesses are a¡ected by systems but unable to in£u- ence their behaviour. Problem-solvers or analysts take on board the task of trying to improve systems. Since purposes emanate from the human mind, attention also has to be given to the di¡erent mental models that people bring to their roles. These mental models are made up, in each case, of a mix of the understanding and values that individuals have gathered through their experiences and educa- tion. The facts and values that they use in interpreting the world can perhaps themselves be understood in systems terms. They are said to consti- tute the world view, Weltanschauung (a German word meaning ‘world image’), or appreciative system employed by an individual or group. For those who want to manage purposeful systems or intervene to change them the resistance, or otherwise, of Weltanschauungen or appreciative systems to change becomes critical. If the only change that can be contem- plated takes place in the context of an existing mental model, then you are limited to bringing about ¢rst-order learning. If, however, the mental model itself can be changed, and purposes radically altered, then second- order change is possible. The ways in which world views change became a primary focus of ‘soft systems thinking’ and, within this, Hegel’s notion of a ‘dialectical debate’ between thesis and antithesis was particularly in£uential. Finally, in considering purposeful systems, we need to note how signi¢- cant the concept of boundary becomes. With a machine or organism it is usually very apparent where the boundary of the system lies. For those concerned with purposeful systems, however, this is rarely the case. Where the boundary is seen to be will depend on the world view of the person observing the system. For example, whether the boundary of a business organization should expand to include its natural environment, its local community, unemployed people, etc. are all very much issues open to debate. Values and ethics play a part in such decisions. There is the further matter of who should participate in de¢ning purposes, taking decisions and The physical sciences 11 drawing boundaries. And because resources and interests will be at stake, as well as di¡erent philosophies, power and politics will have a signi¢cant impact on purposeful systems. The encounter of holism with management and organization theory has thrown up complications not found when the focus of attention for systems thinking was the natural realm. Part II reveals, however, that this has not been an unequal challenge; holism has stood up to the task well enough. 1.6 THE PHYSICAL SCIENCES Systems thinking emerged as a transdiscipline, in the 1940s and 1950s, in large part as a reaction to the reductionism of the traditional scienti¢c method and the failure of that reductionism to cope with the complexity inherent in the biological and social domains. It seemed for some time, there- fore, that systems thinking was the antithesis of the scienti¢c method. More recently, however, the physical sciences seem to have undergone their own systems revolution and holism, and the concepts associated with it have been welcomed in physics and chemistry as o¡ering new forms of explana- tion and new avenues of exploration. Quantum theory in physics and the study of dissipative structures in chemistry are examples of a more holistic orientation in the physical sciences. Because they have undergone their own systems revolution, the physical sciences are now able to make their own contributions to the language of systems thinking more generally. Quantum physics brought to the fore the notion of indeterminacy and gave new meaning to the concept of relation- ships. From chemistry comes a reinforcement of the process view of systems and the idea of self-organization. Perhaps most important of all, however, has been the birth of a new kind of general system theory in science under the banner of chaos and complexity theory (see Gleick, 1987). Complexity theory ^ the more general term and the one we shall use ^ complements the normal systems concern for order by being equally concerned with disorder. The fact that so many complex systems appear to exhibit disorder, irregularity and unpredictability had seemed to put them beyond the reach of scienti¢c understanding. Complexity theorists did not actually dispute this. Indeed, their early studies reinforced the notion by demonstrating that a small change in the initial conditions of a system can lead to large-scale consequences later on: famously, a butter£y £apping its wings in the Amazon jungle can conceivably lead to storms in the South 12 The systems language China Sea. However, what they also found was that underlying apparent chaos was a surprising degree of pattern. Complex systems seem to be governed in some way by ‘strange attractors’, which means that although they never repeat exactly the same behaviour, what they do remains within certain limits. The weather in England is notoriously unpredictable in detail, but we never experience extreme cold or extreme heat and, only occasionally, very heavy rainfall and hurricanes. Furthermore, the patterns that govern complex systems seem to be repeated at di¡erent levels of the system. The parts of the whole are similar in shape to the whole. Snow£akes and cauli£owers have been used as everyday examples of ‘fractal wholes’ demonstrating such self-similarity. Pursuing their research into order and disorder in complex systems, com- plexity theorists discovered what became known as the ‘edge of chaos’. This is a narrow transition zone between order and chaos where systems become capable of taking on new forms of behaviour ^ of self-organization and particularly innovative activity. The potential of complexity theory for helping managers is perhaps becoming clear. The organizations they manage seem chaotic and unpredict- able. But maybe they too are governed by strange attractors that can, after all, be understood. The environments in which organizations operate are turbulent and ever changing, yet organizations seem slow to adapt. Maybe if they can be driven to the edge of chaos they will be much more creative in the way they behave. A new systems view of organizations has been constructed out of these ideas. 1.7 WHY IS THE SYSTEMS LANGUAGE SO POWERFUL? In this chapter we have started to become familiar with the systems language. Our undertstanding will be deepened as we start to see how the language can be used to address management problems in Part II. Obviously, it takes e¡ort to learn a new language and we will have to encounter still more new concepts in what follows. In asking you to make this e¡ort I can perhaps rely on the fact that managers are fed up with being o¡ered simple solutions to complex, diverse problems. They recognize that more sophisticated solutions are necessary and that this may demand a more di⁄cult language. I am keen, however, to close the chapter with just four arguments as to why you should bother with the systems language. First, as we have seen, the emphasis on holism o¡ers a useful corrective to the reductionism that still governs much management thinking. Organiza- References 13 tions are complex and the relationships between the parts are crucial. There is a need for joined-up thinking in addressing their problems. Second is the emphasis modern systems thinking puts on process as well as structure. This stems from systems philosophy, from von Bertalan¡y’s open systems concept and from complexity theory. It is not always the right approach to design systems according to some prede¢ned blueprint. Allowing a process to take place can lead to innovative behaviour and ways forward that could not have been foreseen before the process was embarked on. Third is the transdisciplinarity of systems thinking. It draws its ideas and concepts, as we have seen, from a variety of di¡erent disciplines and in so doing can draw on their di¡erent strengths. Even if analogies derived from physics and biology do not hold strictly when applied to organizations, managers have access to a rich storehouse of insights if they can use other disciplines to provide them with new metaphors for understanding their role. Finally, the systems language has proven itself more suitable for getting to grips with real-world management problems than that of any other single discipline. It has given rise to a range of powerful systems approaches to man- agement. The next chapter starts to look at the development of this applied systems thinking. In Part II you will get the chance to judge the truth of the claim I am making here for yourself. REFERENCES Ashby, W.R. (1956). An Introduction to Cybernetics. Methuen, London. Capra, F. (1996). The Web of Life: A New Synthesis of Mind and Matter. Flamingo, London. Checkland, P.B. (1981). Systems Thinking, Systems Practice. John Wiley & Sons, Chichester, UK. Gleick, J. (1987). Chaos: The Making of a New Science. Abacus, London. Jackson, M.C. (2000). Systems Approaches to Management. Kluwer/Plenum, New York. Maturana, H.R. and Varela, F.J. (1980). Autopoiesis and Cognition: The Realization of the Living. D. Reidel, Dordrecht, The Netherlands. von Bertalan¡y, L. (1950). The theory of open systems in physics and biology. In: F.E. Emery (ed.), Systems Thinking (pp. 70^85). Penguin, Harmondsworth, UK. von Bertalan¡y, L. (1968). General System Theory. Penguin, Harmondsworth, UK. Wiener, N. (1948). Cybernetics. John Wiley & Sons, New York. Applied Systems Thinking 2 OR [operational research] is regarded by many as being in crisis. If OR is taken to be ‘classical OR’, this is indisputable... If, however, the definition of OR is widened to embrace other systems-based methodol- ogies for problem solving, then a diversity of approaches may herald not crisis, but increased competence and effectiveness in a variety of different problem contexts. Jackson and Keys (1984) 2.1 INTRODUCTION As systems thinking evolved, and systems concepts developed in the way described in the previous chapter, increasing attention was given to whether it could be used to tackle practical real-world problems. In this chapter we start to consider the work of those involved in a more applied approach to systems thinking, especially those who wanted to apply systems ideas to managerial problem situations. To this end we ¢rst consider the nature and limitations of what has come to be called ‘hard systems think- ing’. We then look at how applied systems thinking developed, during the 1970s, 1980s and 1990s, to overcome some of the weaknesses of hard systems thinking, in the process making itself useful in a much wider range of problem situations. Section 2.4 introduces the main strands of applied systems thinking and picks out the characteristics of the particular approaches we shall be studying in more depth in Part II. A conclusion summarizes the state of applied systems thinking today and leads us to ask questions about how we can make the most creative use of the di¡erent modes of holistic intervention that now exist. Part III, on creative holism, will seek to provide the answers. 16 Applied systems thinking 2.2 HARD SYSTEMS THINKING When systems practitioners bring together various systems ideas and techniques in an organized way and employ them to try to improve a problem situation, they are said to be using a ‘systems methodology’ ^ another technical term to which we shall become accustomed. The attempt to devise such methodologies as a means of tackling real-world problems began around the time of the Second World War. It was during the Second World War, and its immediate aftermath, that the methodologies of Opera- tional Research (OR), Systems Analysis (SA) and Systems Engineering (SE) were born. OR was used extensively to assist the allied war e¡ort (e.g., in increasing the e⁄ciency of radar systems and in optimizing the results of bombing raids on German cities). After the war OR workers migrated into government departments and, especially in Britain, into OR groups estab- lished in the large nationalized industries. SA was promoted by the highly in£uential RAND (acronym for ‘Research ANd Development’) Corporation and used extensively to help the US military. Somewhat later, in the form of spin-o¡s, such as cost^bene¢t analysis, it found willing champions in central and local government departments. SE was an extension of the principles adopted by the engineering profession to large industrial engineer- ing projects (e.g., in the chemical and aerospace industries). Checkland (1981), recognizing similarities between the approaches of OR, SA and SE, labelled this kind of systems work ‘hard systems thinking’. We shall be exploring its nature, strengths and weaknesses more fully in Chapter 4. In essence, however, it o¡ered managers and management scientists a means of seeking to optimize the performance of a system in pursuit of clearly identi¢ed goals. Emphasis is placed on the application of a systematic methodology that, having established objectives, is able to identify problems that stand in the way of optimization and rectify them by employing scienti¢c modelling, rational testing, implementation and evalua- tion processes. Hard systems thinking was a breakthrough in terms of applying systems thinking to real-world problems. In many cases, as we shall see in Chapter 4, it o¡ers a methodology that remains the most appropriate way of proceed- ing to tackle such problems. A considerable amount of criticism has, however, been levelled at the limitations of hard systems thinking in the environment inhabited by managers. These criticisms relate to its inability to handle signi¢cant complexity, to cope with a plurality of di¡erent beliefs and values, and to deal with issues of politics and power. The extreme complexity and turbulence of problem situations, and of the The development of applied systems thinking 17 environments surrounding them, frustrate the aspirations of hard systems thinkers. Hard approaches require an objective account of the system of concern so that a mathematical model can be produced and an optimal solution to the problem recommended. The ‘reality’ facing today’s managers is so complex and subject to change that it is impossible to reduce problem situations to a form that would make them amenable to such modelling. How can we distinguish exactly which elements contribute to the problem situation, identify the relevant interactions between them and quantify their in£uence? Another limitation is that hard systems thinking is unable to deal satisfac- torily with multiple perceptions of reality. It demands that the goal of the system of concern be known or ascertained before analysis can proceed. OR, for example, requires ‘formulation of the problem’ on the basis of the objectives to be achieved. In managerial situations the establishment of agreed objectives will often lie at the very heart of the problem to be tackled. Di¡erent stakeholders will have diverse opinions about the nature of the system they are involved with and about its proper purposes. Consider, for example, a university ^ is it primarily a research institution, a teaching factory, a servant of its local community, a supplier of trained labour to employers, a means of passing on the cultural norms of a society, a holiday camp that keeps kids o¡ the street, etc.? Hard methodologies, lacking mechanisms for generating accommodations around objectives, are unable even to get started when confronted with messy situations of this kind. In need of a clearly de¢ned goal, and an objective account of the situation, it is not surprising that hard systems thinkers should cleave to the point of view of the powerful to progress their analyses. This strategy also increases the chances of having some recommendations implemented. Obviously, however, it leaves hard approaches open to the charge of being unable to deal with politics and power, of serving only in£uential clients and of limit- ing their recommendations to those that defend the status quo. By the 1970s, because of the obvious failings of hard systems thinking, the systems community found itself in something akin to a crisis. In Section 2.3 we shall consider further the nature of this crisis and how applied systems thinking developed in order to overcome it. 2.3 THE DEVELOPMENT OF APPLIED SYSTEMS THINKING The history of applied systems thinking can be presented in terms of e¡orts to overcome the weaknesses of hard systems thinking as set out in the previous 18 Applied systems thinking section. Success in this endeavour has been hard-won, but over the last 30 years or so signi¢cant developments have taken place and the systems approach is now valued as making an important contribution to resolving a much wider range of complex problems than hard systems thinking was able to deal with. We can understand these developments best using a frame- work for classifying systems methodologies, developed by Jackson and Keys in 1984, called the System Of Systems Methodologies (SOSM). 2.3.1 Problem contexts The starting point in constructing the SOSM is an ‘ideal-type’ grid of problem situations or problem contexts. This grid has been described and presented in various ways (see Jackson and Keys, 1984; Jackson, 1993, 2000; Flood and Jackson, 1991), but an easily understandable version is shown as Figure 2.1. We argued earlier in the book that problem contexts become more di⁄- cult to manage as they exhibit greater complexity, change and diversity. In very general terms, systems thinkers see increasing complexity, change and diversity as stemming from two sources: the ‘systems’ managers have to deal with, as they become larger and subject to more turbulence; and the ‘participants’, those with an interest in the problem situation, as their Figure 2.1 Jackson’s extended version of Jackson and Keys’ ‘ideal-type’ grid of problem contexts. The development of applied systems thinking 19 values, beliefs and interests start to diverge. This gives rise to the ‘systems’ and ‘participants’ dimensions used to establish the grid. The vertical axis expresses a continuum of system types conceptualized at one extreme as relatively simple, at the other as extremely complex. Simple systems can be characterized as having a few subsystems that are involved in only a small number of highly structured interactions. They tend not to change much over time, being relatively una¡ected by the independent actions of their parts or by environmental in£uences. Extremely complex systems, at the other end of the spectrum, can be characterized as having a large number of subsystems that are involved in many more loosely structured interactions, the outcome of which is not predetermined. Such systems adapt and evolve over time as they are a¡ected by their own purpose- ful parts and by the turbulent environments in which they exist. The horizontal axis classi¢es the relationships that can exist between those concerned with the problem context ^ the participants ^ in three types: ‘unitary’, ‘pluralist’ and ‘coercive’. Participants de¢ned as being in a unitary relationship have similar values, beliefs and interests. They share common purposes and are all involved, in one way or another, in decision-making about how to realize their agreed objectives. Those de¢ned as being in a pluralist relationship di¡er in that, although their basic interests are compati- ble, they do not share the same values and beliefs. Space needs to be made available within which debate, disagreement, even con£ict, can take place. If this is done, and all feel they have been involved in decision-making, then accommodations and compromises can be found. Participants will come to agree, at least temporarily, on productive ways forward and will act accordingly. Those participants de¢ned as being in coercive relationships have few interests in common and, if free to express them, would hold con£icting values and beliefs. Compromise is not possible and so no agreed objectives direct action. Decisions are taken on the basis of who has most power and various forms of coercion employed to ensure adherence to commands. Combining the ‘systems’ and ‘participants’ dimensions, divided as suggested above, yields six ideal-type forms of problem context: simple^ unitary, simple^pluralist, simple^coercive, complex^unitary, complex^ pluralist and complex^coercive. This notion of ‘ideal type’ is crucial in understanding the SOSM and what it is seeking to convey. The grid does not wish to suggest that real-life problem situations can be de¢ned as ¢tting exactly within any of these boxes. Weber (1969), the originator of the notion, describes ideal types as stating logical extremes that can be used to construct abstract models of general realities. The grid presents some 20 Applied systems thinking abstract models that reveal various ways in which problem contexts might be typi¢ed by managers and management scientists. It is useful to us here if we are able to show, as we seek to do in the next subsection, that the developers of di¡erent systems methodologies have themselves been governed by particular ideal-type views of the nature of problem contexts in producing their systems approaches. 2.3.2 Systems methodologies related to problem contexts The ideal-type grid of problem contexts is useful in helping us to understand how applied systems thinking has developed over the last few decades. It enables us to grasp the variety of responses made by systems practitioners in their attempts to overcome the weaknesses of hard systems thinking in order to tackle more complex problem situations. We are able to discern a pattern in the history of the development of applied systems thinking. Let us consider initially the assumptions made by hard systems thinking about the nature of problem contexts. It is clear that they assume they are ‘simple^unitary’. In other words, hard systems approaches take it for granted that problem contexts are simple^unitary in character and recom- mend intervening accordingly. It is not surprising given the circumstances in which they were developed that they came to rely on there being a shared and, therefore, readily identi¢able goal. If you are trying to win a war or are engaged in postwar reconstruction, it is completely reasonable to make unitary assumptions. Later in the 1960s and 1970s, when hard systems approaches were taken into universities to be further ‘re¢ned’ by academics, an original bias toward quanti¢cation became an obsession with mathematically modelling the system of concern. To believe that this is possible you have to assume that the system you are dealing with is relatively simple. So the underlying assumptions of classical OR (and this is true, if to a lesser extent, of systems analysis and systems engineering) are simple^ unitary. Hard systems thinkers remain stuck in that area of the grid of problem contexts where it is assumed that people share values and beliefs and that systems are simple enough to be mathematically modelled. And it is true that these assumptions have served them well in tackling a whole variety of operational issues; in the case of OR for inventory, queuing, scheduling, routing, etc. problems. Unfortunately, di⁄culties arose when attempts were made to extend the range of application of hard systems approaches, exactly because of the assumptions embedded within them. As was mentioned earlier, it is often di⁄cult to de¢ne precise objectives on which all stakeholders can agree. In The development of applied systems thinking 21 these circumstances, methodologies demanding a prede¢ned goal cannot get started because they o¡er no way of bringing about any consensus or accommodation around a particular goal to be pursued. Similarly, if the system of concern is extremely complex, then any mathematical model pro- duced can only o¡er a limited and distorted view of reality from a particular perspective ^ and one which, in a turbulent situation, becomes quickly out of date. In the 1970s, therefore, came a general understanding of the lack of usefulness of hard systems thinking for more complex problem situations, and in problem contexts that were deemed to be more pluralist and coercive in character. It is to the credit of applied systems thinking that it has not remained stuck in its simple^unitary ghetto. The last 30 or so years have seen an attempt to extend the area of successful application of systems ideas by developing methodologies that assume that problem contexts are more complex, pluralist and/or coercive in nature. This is the progress in applied systems thinking that we now seek to chart. We begin with the vertical axis of the ideal-type grid of problem contexts, and our concern, therefore, is with those systems practitioners who wanted to move down the axis by assuming that problem contexts were more complex than hard systems thinkers believed. The aim of hard systems thinking was to optimize the system of concern in pursuit of a known goal, and to do this it appeared necessary to model the interactions between all those elements or subsystems that might a¡ect that system of concern. In complex systems, the vast numbers of relevant variables and the myriads of interactions make this an impossible requirement. The solution, suggested by those wishing to progress down the vertical axis, was to identify those key mechanisms or structures that govern the behaviour of the elements or subsystems and, therefore, are fundamental to system behaviour. It is regarded as impossible to mathematically model the relationships between all the variables that ‘on the surface’ appear to be involved in what the system does. You can, however, determine the most important structural aspects that lie behind system viability and performance. This ‘structuralist’ approach enables the analyst to determine, at a deeper level, what is going wrong with the present functioning of the system and to learn how to manip- ulate key design features so that the system can survive and be e¡ective over time by continually regulating itself, and self-organizing, as it adapts to internally and externally generated turbulence. The systems approaches responsible for making this shift down the vertical axis show a common concern for understanding the nature of complex adaptive systems and with ensuring they are designed to have a 22 Applied systems thinking capacity for goal seeking and remaining viable in turbulent environments. In this book we concentrate on ‘system dynamics’, ‘organizational cybernetics’ and ‘complexity theory’ as systems approaches that assume, in this manner, that problem contexts are extremely complex and need tackling in a ‘structur- alist’ fashion. In each case, as we shall see, they identify di¡erent key structural aspects that need to be understood and manipulated in dealing with complex- ity. In the case of system dynamics it is the relationships between positive and negative feedback loops that can give rise to ‘archetypes’ of system behaviour. In the case of organizational cybernetics it is cybernetic laws that can be derived from the concepts of black box, feedback and variety. With complexity theory it is ‘strange attractors’ and the variables that have to be adjusted to ensure that an ‘edge of chaos’ state is achieved. Applied systems thinkers have also made considerable progress along the horizontal axis of the ideal-type grid of problem contexts. If we move part way along that axis we ¢nd that a number of methodologies have been developed that assume that problem contexts are pluralist and provide recommendations for analysis and intervention on that basis. This tradition of work has become known as ‘soft systems thinking’ to distinguish it from the hard systems thinking that was left behind. Soft systems thinkers abandoned the notion that it was possible to assume easily identi¢able, agreed-on goals that could be used to provide an objective account of the system and its purposes. This was seen to be both impossible and undesirable given multiple values, beliefs and interests. Instead, attention had to be given to ensuring su⁄cient accommodation between di¡erent and sometimes con£icting world views in order that temporary coalitions could be fashioned in support of particular changes. The solution was to make subjectivity central, working with a variety of world views during the methodological process. In Checkland’s ‘soft systems methodology’ (1981), a highly developed approach of this kind, systems models expressing di¡erent viewpoints, and making explicit their various implications, are constructed so that alternative perspectives can be explored systemically, compared and contrasted. The aim is to generate a systemic learning process in which the participants in the problem situation came to appreciate more fully alternative world views, and the possibilities for change they o¡er, and as a result an accommodation, however temporary, becomes possible between those who started with and may still hold divergent values and beliefs. Systems practitioners seeking to progress along the horizontal dimension emphasize the crucial importance of values, beliefs and philosophies. Their primary interest is in exploring the culture and politics of organizations to The development of applied systems thinking 23 see what change is feasible and in gaining commitment from participants to agreed courses of action. Such soft systems thinkers are not trying to devise system models that can be used over and over again to reveal how real- world systems can be improved. This is felt not to be relevant or useful because of the widely di¡erent viewpoints about purposes that will be present in pluralist problem contexts. Instead, what is usefully replicated, as Checkland argues, is the methodology employed. The same approach to bringing about consensus or accommodation is tried again and again and is gradually improved. As well as studying Checkland’s ‘soft systems method- ology’ we will be considering ‘strategic assumption surfacing and testing’ and Acko¡ ’s ‘interactive planning’. All these soft systems approaches have by now been well researched. As a result we know much better than previously about some methodological processes that can assist in bringing about accommodations between di¡erent value positions and generate commitment among participants to implement agreed changes. If we shift further along the horizontal axis of the grid of problem contexts, the issue arises of how to intervene in problem situations that are regarded as coercive. Soft systems thinking fails to respond appropriately because of its pluralist bias that consensus, or at least accommodation, between di¡erent stakeholders can be achieved. Systems practitioners have, therefore, sought to formulate ‘emancipatory’ systems approaches based on the assumption that problem situations can be coercive. Ulrich’s ‘critical systems heuristics’ allows questions to be asked about who bene¢ts from par- ticular system designs and seeks to empower those a¡ected by management decisions but not involved in them. Beer’s ‘team syntegrity’ seeks to specify an arena and procedures that enable all stakeholders to debate openly and de- mocratically the issues with which they are confronted. Both these ap- proaches are considered. Finally, there are systems practitioners who worry about the claims of any systems methodology to be able to guarantee generalized improvement. They advocate postmodern systems practice in the face of the massive and impenetrable complexity and coercion that they see as inherent in all problem contexts. Suppressed viewpoints must be surfaced and diversity encouraged as in the emancipatory systems approach. All that is possible however is contested, local improvement justi¢ed on the basis that it feels right given local circumstances. Chapter 13 is devoted to this version of applied systems thinking. In short, the argument of this section is that applied systems thinking has developed over the past few decades taking into account the characteristics of a much wider range of the ideal-type problem contexts represented in 24 Applied systems thinking Figure 2.2 Systems approaches related to problem contexts in the System of Systems Methodologies (SOSM). the grid. It has progressed along the vertical dimension to take greater account of complexity. It has progressed along the horizontal dimension acknowledging that problem contexts can be de¢ned as pluralist and coercive. These conclusions are summarized in Figure 2.2. The intersecting lines that constructed the particular problem contexts in the grid of Figure 2.1 have been removed in this representation of the SOSM. This should be taken to mean that it is only indicative of the assumptions made by di¡erent systems approaches about the nature of problem contexts. There is no inten- tion to pigeon-hole methodologies and a more sophisticated treatment of their underlying assumptions will be presented in Part II. 2.4 THE MAIN STRANDS OF APPLIED SYSTEMS THINKING It is worth taking time at this point to build on the work of the previous section and to explain brie£y the rationale behind the grouping of systems approaches in Part II. In Part II, 10 systems approaches are divided into 4 types according to whether their primary orientation is improving goal seeking and viability, exploring purposes, ensuring fairness or promoting diversity. These are not mutually exclusive possibilities, but they o¡er a reasonable guideline as to where the main emphasis of an approach lies and, therefore, to what managerial end it most easily lends itself. The main strands of applied systems thinking 25 Type A systems approaches are dedicated to improving goal seeking and viability. This is a fairly broad category that ranges from optimizing approaches, single-mindedly concerned with reaching prede¢ned goals, to approaches where much more attention is given to capacity building in those areas of organizational behaviour and design perceived as necessary if viability is to be ensured, and so goal seeking made possible. In all cases, however, the measures of success are ‘e⁄ciency’ (are the minimum resources used in goal seeking?) and/or ‘e⁄cacy’ (do the means employed enable us to realize our goals?). The kinds of systems approach we are discussing here are those that have tended to take the nature of the purposes to be served by the system of concern for granted. They have assumed that participants are in a unitary relationship so that goals are already clear or can be easily determined. Their e¡orts have concentrated on the vertical axis of the grid of problem contexts where they have sought to optimize the system of concern to achieve its goals or recon¢gure it to enable it to deal with internally and externally generated complexity and turbulence. The original form of applied systems thinking, the hard systems approach, endeavours as we saw to ¢nd the best means of getting from the present state of the system to some optimum state. Mathematical modelling is often seen as crucial to the success of this. The other three approaches considered under Type A are more ‘structuralist’ in nature in terms of the analysis of Section 2.3. They seek to understand and manipulate the mechanisms, operating at a ‘deeper’ level, that give rise to system behaviour. System dynamics sees the key to system behaviour as lying in the inter- relationships between the positive and negative feedback loops within which important system elements are bound. If these can be understood, then the manager can be guided as to how he or she should intervene in order that system behaviour is controlled close to what is regarded as desir- able. Organizational cybernetics uses a cybernetic model, the Viable System Model (VSM), to try to manage issues of complexity and turbulence that are beyond the capacity of hard systems approaches to handle. The VSM seeks to help managers to design complex organizations according to cybernetic prescriptions so that they remain viable in rapidly changing environments. Managers can learn how to use the VSM to diagnose problems in organizations and put them right so that viability is secured and goal seeking becomes possible. Complexity theory is often associated with unpredictability and with the study of disorder. However, an equally important ¢nding of complexity theory is that, underlying chaos, it is poss- ible over time to recognize patterns occurring in the way systems develop. 26 Applied systems thinking Managers with access to these patterns can identify points of leverage that they can exploit to ensure that desirable system behaviour is forthcoming. Type B systems approaches are dedicated to exploring and clarifying the purposes stakeholders want to pursue through the operations or organiza- tion in which they have an interest. The three approaches covered are alternative examples of ‘soft systems thinking’ and so advocate facilitating a learning process in which the importance of subjectivity is fully respected. Stakeholders can bene¢t from being made aware of the systemic implications of the values and beliefs they hold and by being confronted with di¡erent visions of the future and the changes necessary to achieve it. Debate can then be organized around the di¡erent viewpoints about purposes that exist and accommodations teased out that stakeholders can commit to in planning systemic improvement. The measures of success for soft methodol- ogies are ‘e¡ectiveness’ (are we actually achieving what we want to achieve?) and elegance (do the stakeholders ¢nd what is proposed tasteful?). The kinds of systems approach we are discussing here are those that have concentrated their e¡orts on the horizontal axis of the grid of problem contexts. They have seen the main failing of hard systems thinking as being its inability to deal with pluralism. They see much the most important task of systems thinking as being able to handle the disagreements and con£icts that occur between stakeholders because of the di¡erent values, beliefs and philosophies they hold. If these can be managed, then solutions to problems become more or less straightforward. Strategic assumption surfacing and testing concentrates attention on the di¡erent assumptions, multiple perspectives and diverse world views that are likely to exist in any problem situation. It takes advantage of these to articulate a dialectical learning process of thesis, antithesis and synthesis. Con£ict is thus harnessed to assist with problem resolution. Interactive planning seeks to win stakeholder approval for and commitment to an ‘idealized design’ for the system they are involved with. This is meant to ensure that the maximum creativity is brought to the process of dissolving the current mess the stakeholders are confronted by and replacing it with a future they all desire. Appropriate means for achieving the idealized design are then sought. Soft systems methodology enables managers to work with and change the value systems, cultures and philosophies that exist in organizations. It aims to institutionalize continuous learning by seeking and challenging accommodations between the world views of the di¡erent stakeholders concerned with a problem situation. Type C systems approaches are dedicated to ensuring fairness in systems design and in the consequences that follow from it. The two approaches The main strands of applied systems thinking 27 considered are examples of ‘emancipatory systems thinking’ that have ven- tured along the horizontal axis of the grid of problem contexts, into areas where the value of soft systems approaches is threatened by lack of fairness or by coercion. To that extent their aims are similar. They want to support those disadvantaged by present systemic arrangements so that they can make their full contribution to systems design and receive the bene¢ts to which they are entitled from the operation of the system of concern. This may not be happening at the moment for all sorts of reasons. There may be a lack of recognition of the rights of some stakeholder group. And it may be that this is the result of some form of conscious or unconscious discrimina- tion based on class, sex, race, sexual orientation, disability, etc. Emancipatory approaches focus attention on matters of this kind that can easily be missed by other sorts of systems thinking. They are, of course, of huge signi¢cance in society and of increasing importance for all organizations. The measures of success for emancipatory approaches are ‘empowerment’ (are all individuals and groups able to contribute to decision-making and action?) and ‘emancipation’ (are disadvantaged groups being assisted to get what they are entitled to?). Critical systems heuristics and team syntegrity address these emancipatory concerns from di¡ering perspectives. The former seeks to ensure the full participation of those who are a¡ected by systems designs who might not otherwise be involved. The latter provides for the creation of a democratic milieu in which outcomes result from con- sensus and the better argument rather than power, status and/or hierarchy. Type D are postmodern systems approaches that seek to promote diver- sity in problem resolution. Such approaches are, in a sense, antisystemic in that systems of domination (e.g., dominating discourses) have to be challenged and broken down in order to let suppressed voices have their say. They are less well established than other types of systems methodology (because they are more recent) and one chapter is enough to hint at the value of some of the postmodern methods now being developed. Post- modern systems thinkers are phased by what they see as the immense com- plexity and coercion that are intertwined in all problem situations. They are therefore sceptical of appeal to any universal guarantees for the success of action. They would however want to justify and evaluate their interventions on the basis of ‘exception’ (what otherwise marginalized viewpoints have we managed to bring to the fore?) and ‘emotion’ (does the action that is now being proposed feel appropriate and good in the local circumstances in which we are acting?). This section has sought to link our account of the development of applied systems thinking to the arrangement of systems approaches that will be 28 Applied systems thinking found in Part II. Further justi¢cation for seeing four major strands in contemporary systems thinking will be given in the next chapter when the four strands are linked to four overarching social science paradigms. The systems approaches detailed as part of the four strands are not, of course, an exhaustive set. In particular, I regret the omission due to space constraints of Miller’s ‘living systems theory’, which contributes signi¢cantly to dealing with systems complexity, and War¢eld’s ‘interactive management’, a well-regarded soft systems approach. More information about other systems approaches (including these two), together with full references, can be found in Jackson (2000). 2.5 CONCLUSION In this chapter we have covered a lot of ground, looking at the development of applied systems thinking. The systems approaches available today have resulted from attempts to correct the original problems found when trying to use hard systems thinking in practice. They have also arisen from theo- retical developments in the transdiscipline of systems thinking as new problem contexts have been envisioned and their implications for practice have been explored. As we have seen it is reasonable to conclude that there are now four main strands of applied systems thinking embracing a whole variety of individual systems approaches. Awareness of the di¡erent strands of applied systems thinking and of the variety of systems methodologies leads us to ask whether it might not assist creative problem solving to use them in combination in the same interven- tion. The SOSM has after all helped to demonstrate the relationships between the di¡erent approaches and made it possible to understand that they do not necessarily clash with one another. They all do rather di¡erent things. It is this insight ^ that we need to make creative use of the di¡erent forms of holistic inquiry ^ which inspired the ‘creative holism’ that is the focus of Part III. Before we can appreciate that, however, we need to under- stand exactly how creativity can be enhanced by using systems approaches in combination. That is the subject of the next chapter. REFERENCES Checkland, P.B. (1981). Systems Thinking, Systems Practice. John Wiley & Sons, Chichester, UK. References 29 Flood, R.L. and Jackson, M.C. (1991). Creative Problem Solving: Total Systems Intervention. John Wiley & Sons, Chichester, UK. Jackson, M.C. (1993). The system of systems methodologies: A guide to researchers. Journal of the Operational Research Society, 44, 208^209. Jackson, M.C. (2000). Systems Approaches to Management. Kluwer/Plenum, New York. Jackson, M.C. and Keys, P. (1984). Towards a system of systems methodologies. Journal of the Operational Research Society, 35, 473^486. Weber, M. (1969). The Methodology of the Social Sciences. Free Press, New York. Hard Systems Thinking 4 Many elements of such [sociotechnical] systems exhibit forms of regular behaviour, and scientific scrutiny has yielded much knowledge about these regularities. Thus, many of the problems that arise in socio- technical systems can be addressed by focusing such knowledge in appropriate ways by means of the logical, quantitative, and structural tools of modern science and technology. Quade and Miser (1985) 4.1 INTRODUCTION Hard systems thinking, as we saw in Chapter 2, is a generic name given by Checkland (1981) to various systems approaches for solving real-world problems developed during and immediately after the Second World War. The approaches most commonly associated with this label are operational research (operations research in the USA), systems analysis and systems engineering. These, however, gave rise to a myriad of other variants of hard systems thinking, such as decision science, cost^bene¢t analysis, planning^programming^budgeting systems and policy analysis. All these approaches took on a common form, which Checkland identi¢ed and classi- ¢ed as ‘hard systems thinking’ and which we will be examining below. The pioneers of hard systems thinking were immensely proud of the fact that they applied the scienti¢c method to problems of real signi¢cance to decision-makers. They were not the ¢rst to do this. Frederick Taylor had abandoned the laboratory as the place to practise science much earlier in the century, when he invented scienti¢c management. They were, however, the ¢rst to recognize that in modifying the scienti¢c method, to make it applicable to real-world problems, one of its main tenets ^ reductionism ^ had to be thoroughly questioned. Might not holism o¡er a better handle on the complex sociotechnical problems that managers face? 48 Hard systems thinking 4.2 DESCRIPTION OF HARD SYSTEMS THINKING 4.2.1 Historical development The term ‘operational research’ was invented about 1937 in the context of a project in which UK scientists sought to assist military leaders to maximize the bene¢ts to be gained from using radar to detect enemy aircraft. What justi¢ed the new name was that this was scienti¢c research carried out into operational processes rather than into natural phenomena. From the RAF, operational research soon spread to the army and navy and to other countries, such as Canada, the USA, France and Australia. In the USA its ¢rst usage was in the Naval Ordnance group dealing with mine warfare. After the war it found civilian application in government departments and, particularly, in the newly nationalized industries of the UK such as coal, gas, steel and transport. The 1950s saw professional societies being formed to promote Operational Research (OR) and the beginnings of the academic study of the subject. Systems analysis is said by its protagonists to have emerged out of operations research and to be broader in scope. The name was ¢rst applied to research being done for the US Air Force on future weapon systems in the late 1940s. In the 1950s and 1960s the approach was promoted by the in£uential RAND (an acronym for ‘Research ANd Development’) Corpora- tion, a non-pro¢t body in the advice-giving business, and its use became widespread in the defence and aerospace industries. In 1965 President Johnson gave systems analysis (under the label ‘planning^programming^ budgeting systems’) a further boost by ordering its adoption in all departments of the US federal government. In 1972 the International Institute for Applied Systems Analysis (IIASA) was established in Austria, on the initiative of the academies of science (or equivalent) of 12 nations, with the remit to apply systems analysis to world problems (e.g., energy, food supply and the environment). Since that time IIASA has become the o⁄cial guardian of the development of systems analysis as a discipline and profession. Systems engineering grew out of engineering in the 1940s and 1950s as that discipline sought to extend its scope to the design of more complex systems involving many interacting components. It was pioneered in the USA at Bell Telephone Laboratories to meet the networking challenges faced in the communications industry. It spread rapidly to the defence, space and energy industries and, in the 1960s and 1970s, various guidelines and standards were established for the use of systems engineering to Description of hard systems thinking 49 develop military systems and in civilian aerospace and energy programmes. In manufacturing industry, systems engineering had to encompass even more of the ‘whole’ as it was forced to concern itself with interacting sets of processes (e.g., in petrochemical plants) and how these could be optimized in the prevailing market conditions. Today, the International Council on Systems Engineering (INCOSE) sees the approach as relevant to problems as wide and diverse as transportation, housing, infrastructure renewal and environmental systems (www.incose.org). 4.2.2 Philosophy and theory Much of the philosophy and theory underpinning hard systems thinking is taken for granted and not declared openly. This is not surprising because so much of it is borrowed directly from the natural sciences. Discussion does however sometimes focus on the adjustments that have to be made to the scienti¢c method to make it applicable to the real-world problems that interest hard systems thinkers. We can tease out these adjustments by looking at the common features of well-known de¢nitions of OR, Systems Analysis (SA) and Systems Engineering (SE). The British Operational Society for many years de¢ned OR as: the application of the methods of science to complex problems arising in the direction and management of large systems of men, machines, materials and money in industry, business, government and defence. The distinctive approach is to develop a scientific model of the system, incorporating measurements of factors such as chance and risk, with which to predict and compare the outcomes of alternative decisions, strategies or controls. The purpose is to help management determine its policy and actions scientifically. Quade and Miser (1985), in the ¢rst Handbook of Systems Analysis, state that: the central purpose of systems analysis is to help public and private decision and policy-makers to solve the problems and resolve the policy issues that they face. It does this by improving the basis for their judge- ment by generating information and marshalling evidence bearing on their problems and, in particular, on possible actions that may be suggested to alleviate them. Thus commonly, a systems analysis focuses on a problem arising from the operations of a sociotechnical system, considers various responses to this problem and supplies 50 Hard systems thinking evidence about the costs, benefits, and other consequences of these responses. INCOSE propounds a de¢nition of systems engineering with several components: (1) it is an interdisciplinary approach and means to establish a sound system concept, (2) it defines and validates clear and concise system requirements, (3) it creates an effective system design or solution, and (4) it ensures that the developed system meets client and user objec- tives in the operational environment. Although the commitment to science is explicit in all these de¢nitions, it is also clear that the purpose of using science di¡ers from that normally asso- ciated with the scienti¢c enterprise. Its primary purpose in hard systems thinking is to serve the interests of clients, managers, decision-makers, policy-makers, etc., not to bring about the advancement of knowledge for its own sake. Another point follows. In hard systems thinking scientists are required to address real-world problems and the solutions they produce must work in the operational domain, not in the laboratory. Furthermore, it is usually too costly or simply unethical to carry out experiments using large socio- technical systems. They are cut o¡, therefore, from the usual experimental methods employed to test hypotheses under controlled laboratory conditions. An alternative to the laboratory has to be found. All varieties of hard systems thinking propose that models, primarily mathematical models, can perform in management science the role that the laboratory plays in the natural sciences. Models, in hard systems thinking, are designed to capture the essential features of the real world. Sometimes these will be regularities in behaviour, which detailed observation and measurement reveal in particular types of sociotechnical system. At other times the systems practitioner will have to rely on insight and whatever incomplete information that happens to be available. Whatever is the case, it is seen as essential that some type of model is built. Models are so crucial in hard systems thinking because they aim to capture as accurately as possible the workings of the system underlying the problems being investigated. Forced to deal with complex problem situations in the real world, the hard systems thinker replaces the traditional notion of a scien- ti¢c object with that of ‘system’ as the focus of study. Once the model has been constructed it can be used to explore how the real-world system Description of hard systems thinking 51 behaves without actually taking any action that might alter and damage the real-world system itself. In particular, di¡erent possible ways of improving system behaviour from the point of view of the clients can be tested. Finally, it is clearly recognized in the de¢nitions o¡ered that no one ¢eld of science is likely to be able to deal with any real-world problem. Such problems simply do not ¢t into the domains of the established scienti¢c disciplines. The hard systems thinker, being problem- rather than discipline-centred, will therefore have to draw on a range of disciplinary areas or be interdisciplinary in his or her approach. 4.2.3 Methodology Cutting through the arguments of the advocates of di¡erent strands of hard systems thinking that their favoured approach is more comprehensive than the others, Checkland (1981) used an examination of methodology to demonstrate that all variants of hard systems thinking are in fact similar in character. Methodology in applied systems thinking, the reader will recall, refers to the guidance given to practitioners about how to translate the philosophy and theory of an approach into practical application. Looking at the methodologies proposed by hard systems thinkers, Checkland concluded that they all take the same form. They largely assume that they can de¢ne an objective for the system they are seeking to improve and see their task as the systematic pursuit of the most e⁄cient means of achieving that objective. We can now review this conclusion in relation to the speci¢c methodologies of OR, SA and SE. The ¢rst full expression of the classical OR methodology appeared in Churchman, Acko¡ and Arno¡’s textbook on OR published in 1957. The authors establish that OR is the application of the most advanced scienti¢c techniques by interdisciplinary teams to the overall problems of complex organizations and that a systems approach is essential. They then set out a six-stage methodology:. formulating the problem;. constructing a mathematical model to represent the system under study;. deriving a solution from the model;. testing the model and the solution derived from it;. establishing controls over the solution;. putting the solution to work (implementation). 52 Hard systems thinking There is, therefore, the expected emphasis on problem formulation (specify- ing the decision-makers and their objectives, and the system involved), on a modelling phase and an implementation phase. Many di¡erent styles of systems analysis developed out of the early RAND Corporation military applications. It is, however, reasonable to take the three IIASA handbooks, edited by Miser (1995), and Miser and Quade (1985, 1988), as representing the current state of the art as far as systems analysis methodology is concerned. The handbooks make clear that systems analysis always starts with the recognition by someone involved with a sociotechnical system that a problem exists. This problem will require proper formulation. Once that is achieved the methodology pre- scribes a research phase during which a scienti¢c approach is brought to bear on the problem. The research should be multidisciplinary. It requires identifying alternative ways of tackling the problem and building models that can be used to test the alternatives. The alternative means are then evaluated and ranked according to the decision-makers’ preferences, bearing in mind costs, bene¢ts and other consequences. Finally, assistance is given with implementation and with evaluation of outcomes. The Figure 4.1 representation of systems analysis methodology appears in all three of the handbooks. For A.D. Hall (see Keys, 1991), re£ecting on his experiences with the Bell Telephone Laboratories, systems exist in hierarchies and should be engineered with this in mind to best achieve their objectives. The systems engineer is charged with co-ordinating a multidisciplinary team that must discover the objectives and then ensure the optimum integration and consis- tency of system and subsystems in pursuit of those objectives. Jenkins (1972), a British systems engineer, provides a detailed elaboration of the steps required: 1. Systems analysis ^ 1.1 formulation of the problem; 1.2 organization of the project; 1.3 de¢nition of the system; 1.4 de¢nition of the wider system; 1.5 objectives of the wider system; 1.6 objectives of the system; 1.7 de¢nition of an overall economic criterion; 1.8 information and data collection. 2. Systems design ^ 2.1 forecasting; Description of hard systems thinking 53 Figure 4.1 The systems analysis methodology. From Miser and Quade (1988), reproduced by permission of John Wiley & Sons. 2.2 model building and simulation; 2.3 optimization; 2.4 control; 2.5 reliability. 54 Hard systems thinking 3. Implementation ^ 3.1 documentation and sanction approval; 3.2 construction. 4. Operation ^ 4.1 initial operation; 4.2 retrospective appraisal; 4.3 improved operation. INCOSE’s brief guide to systems engineering sees this process as bringing to projects a disciplined vision of stakeholders’ expectations together with a disciplined focus on the end product, its enabling products, and its internal and external operational environment (i.e., a system view). At the beginning of this subsection we stated that Checkland recognized a commonality in the form of all types of hard systems methodology. In essence, he argued, they take what is required (the ends and objectives) as being easy to ascertain and see their task as undertaking a systematic investi- gation to discover the most e⁄cient ‘how’ that will realize the prede¢ned objectives. Hard systems thinking presupposes that real-world problems can be tackled on the basis of the following assumptions: 1. there is a desired state of the system S1 , which is known; 2. there is a present state of the system S0 ; 3. there are alternative ways of getting from S0 to S1 ; 4. it is the role of the systems person to ¢nd the most e⁄cient means of getting from S0 to S1. 4.2.4 Methods It is not entirely correct to say that hard systems thinking has concentrated on methods of model building at the expense of methods to support other stages or phases of the methodological process. The three IIASA handbooks on systems analysis contain a good deal on the craft skills necessary to support problem formulation, communication with decision-makers and implementation. The rapid development of ‘soft OR’ in the 1980s and 1990s in the UK is evidence of a tradition of work that has paid attention to the process of operational research. Successful practitioners of all strands of hard systems thinking have of necessity had to develop well-tuned social and political skills. Nevertheless, it is true that the mainstream academic literature in journals and textbooks certainly does show an overwhelming bias in the direction of perfecting methods of modelling. It is not surprising, Description of hard systems thinking 55 therefore, that this is the area in which hard systems thinking has most to o¡er. Models are explicit, simplifying interpretations of aspects of reality relevant to the purpose at hand. They seek to capture the most important variables and interactions giving rise to system behaviour. They are used to experiment on as surrogates for the real-world system. The literature of hard systems thinking identi¢es various types of models: iconic, analogic, analytic, simulations, gaming, judgemental and conceptual. Let us consider these in turn. Iconic models are simply scale (usually reduced scale) representations of what is being modelled, such as an aircraft model used in wind tunnel testing or an architect’s three dimensional model of a new building. Analogue models are very di¡erent in appearance to the reality, but, never- theless, seek to mimic the behaviour of what they represent. An example would be an electrical network used to represent water £owing through pipes. Analytic models are mathematical models that are used to represent the logical relationships that are believed to govern the behaviour of the system being investigated. They are widely used in operational research. Wilson (1990) helpfully provides a matrix (reproduced as Figure 4.2) that divides analytic models into four classes depending on whether they repre- sent behaviour over time (dynamic) or at one point in time (steady state) and whether the behaviour is described by ¢xed rules (deterministic) or statistical distribution (non-deterministic). Algebraic equations can be used to formulate, for example, problems about the most appropriate way to allocate productive resources in order to maximize pro¢t when many alternatives exist and resources are limited. The linear programming technique has been developed to provide an Figure 4.2 Types of analytic model. 56 Hard systems thinking optimal solution in this type of situation. Statistical and probability relation- ships can be employed to determine the degree of dependence of one variable on another in the absence of complete knowledge about the interrelation- ships in a system. Wilson shows how the linear regression technique can provide a model, for example, showing how total electricity sales are related to the level of industrial production. Di¡erential equations provide a modelling language for the dynamic deterministic category. Wilson demonstrates how the problem of designing a suspension system for a vehicle can be tackled on the basis of di¡erential equations solved by computer. Dynamic non-deterministic systems require simulation, which we are treating here as another type of modelling. Simulation of quite complex systems, in which the relationships between variables are not well understood and change over time, has been made possible with the advance in computing power. It refers to the process of mapping item by item and step by step the essential features of the system we are interested in. The model produced is subject to a series of experiments and the outcomes documented. The likely behaviour of the system can then be predicted using statistical analysis. For example, the essential features of a tra⁄c £ow system can be represented in a computer simulation as long as the key factors impacting on tra⁄c £ow can be identi¢ed. Computer- generated random numbers determine amount of tra⁄c, numbers turning left, etc., so that the system’s behaviour can be monitored under di¡erent conditions. A sophisticated type of simulation modelling, system dynamics, is described in Chapter 5. Gaming is a kind of modelling in which human actors play out the roles of signi¢cant decision-makers in a system. They are supposed to behave as would their real-world counterparts in order that matters of choice, judge- ment, values and politics can be investigated. Judgemental models usually rest on group opinion of the likelihood of particular events taking place. Techniques such as ‘Delphi’ (developed at the RAND Corporation) and ‘scenario writing’ are used in systems analysis to develop the best group models from the individual mental models of members of multidisciplinary teams. Delphi employs an anonymous, itera- tive process to guide experts and other knowledgeable individuals toward a reasonable consensus about an issue. Scenario writing explores the likelihood of particular future states of a¡airs coming about. Conceptual models, as the name suggests, are qualitative models used to make explicit the particular mental models held by parties interested in a decision. They are more frequently employed in soft systems thinking than in hard systems thinking. Hard systems thinking in action 57 4.2.5 Recent developments The most important recent developments in hard systems thinking relate to continued attempts to extend the scope of the approach beyond the some- what technical problem situations, in which it has proved very successful, to situations of greater complexity and in which people and politics play the central role. INCOSE has declared its intention to expand systems engineering into non-traditional domains, such as transportation, housing, infrastructure renewal and environmental systems ^ although, it must be said, little work has been done yet on how the approach might have to change to make it so applicable. The IIASA handbooks declare that systems analysis needs to be extended to more ‘people-dominated’ problems. In this ¢eld considerable e¡ort has gone into documenting the ‘craft skills’ needed to cope with people and politics. But this still remains a long way from developing appropriate concepts and an appropriate language that would allow these matters to be discussed theoretically so that continuous learning can be generated. The greatest progress has been made in OR, with the establishment of ‘soft OR’ in the UK as a complementary practice to ‘hard OR’. Soft OR di¡ers markedly from the classical version we have been studying in this chapter and does put people at the centre of problem resolving and decision-making. The collection of papers in Rosenhead and Mingers (2001) is a good introduction. In general, the question needs asking whether hard systems thinking really can be adapted in the way certain visionaries believe. Would it not lose the capacity to be good at what it does well now? Would it not be better for us to look to other systems approaches, developed for other purposes, to complement hard systems thinking where it is weak? 4.3 HARD SYSTEMS THINKING IN ACTION Given the range of hard systems approaches covered in this chapter, it is di⁄cult to provide one representative example of hard systems thinking in action. We shall tackle this problem by showing di¡erent aspects of use for each of OR, SE and SA. Once the original pioneering spirit had faded, operational researchers, or at least those of a more academic persuasion, began to concentrate their e¡orts on developing mathematical models to apply to what they recognized as frequently occurring types of problems. Each problem type was assumed to have a particular form and structure, which determined its nature and 58 Hard systems thinking how it could be tackled, regardless of the context in which it was found ^ military, manufacturing industry, service sector, etc. Fortuin et al. (1996) present 15 case studies of OR at work in application areas as diverse as transport and logistics, product and process design, maintenance and ¢nancial services, health care and environmental decision-making. Keys (1991), and Cavaleri and Obloj (1993), provide good introductory material on the most common OR problems; a typical list being:. queuing problems;. inventory problems;. allocation problems;. replacement problems;. co-ordination problems;. routing problems;. competitive problems;. search problems. Queuing models seek an optimum trade-o¡ between the costs of providing service capacity and keeping customers happy. Inventory models aim to establish the optimum reorder point for stocks of resources so that pro- duction £ow can be maintained while the costs of holding excess inventory are minimized. Allocation models seek to apportion scarce resources in the most e⁄cient manner, maximizing output or minimizing costs, while achieving overall objectives. Keys comments on an example involving a farming enterprise that both reared cattle for beef and produced crops that could themselves be sold or, alternatively, used to feed the cattle. A linear programming-type model was constructed containing 640 constraints and 1,801 variables. A solution that maximized pro¢t was discovered in 34 seconds of computer time. Replacement models help to minimize costs by identifying the point at which acquisition of new assets is justi¢able. Co-ordination techniques, such as PERT (Programme Evaluation and Review Technique) and critical path analysis, calculate how tasks must be sequenced in a project to ensure completion in minimum time and at minimum cost. The goal of routing models is to determine the most e⁄cient route between di¡erent locations in a network. Competitive problems are conceptualized in terms of games, the aim being to maximize outcomes for one or more participants. Search models try to maximize the e⁄ciency of a search (say, for a location for a new factory) by minimizing both costs and the risks of error. Hard systems thinking in action 59 A recent INCOSE document (see www.incose.org) sets out systems engineering pro¢les for 18 di¡erent application domains: agriculture, commercial aircraft, commercial avionics, criminal justice system and legal processes, emergency services, energy systems, environmental restoration, facilities systems engineering, geographic information systems, health care, highway transportation systems, information systems, manufacturing, medical devices, motor vehicles, natural resources management, space systems, and telecommunications. There are also pro¢les for seven cross- application domains: e-commerce, high-performance computing, human factors engineering, Internet-based applications, Internet banking, logistics, and modelling and simulation. Not surprisingly, a number of these pro¢les are rudimentary, with the most extensive being in areas of traditional systems engineering practice, such as the design and development of commercial aircraft. The commercial aircraft industry operates in a very competitive environ- ment and depends on complex manufacturing processes arising from highly integrated subsystems, advanced technologies, use of advanced materials, detailed speci¢cations and very rigorous testing. The systems engineering speci¢cations for this domain insist on the principle that commercial aircraft are considered as wholes, and not as collections of parts. Both customer and regulatory requirements are ¢rst identi¢ed. Aircraft architecture is then seen as a hierarchy in which the functions and constraints operating at the top level, the aircraft system itself, £ow down into require- ments for the subsystems. A typical decomposition of the aircraft system into parts would identify the mechanical, propulsion, environmental, airframe, avionics, interiors, electrical and auxiliary subsystems. These sub- systems are then further decomposed into subordinate components with their own requirements deriving from those of the subsystems. Thorough monitoring and control is essential at all stages of design and construction to ensure that requirements at the di¡erent levels are veri¢ed and validated by testing. The IIASA handbooks provide some comprehensive descriptions of SA applications, which are then referred to and analysed throughout the three volumes. The main illustrations are of improving blood availability and utilization (also described in Jackson, 2000), improving ¢re protection, protecting an estuary from £ooding, achieving adequate amounts of energy for the long-range future, providing housing for low-income families and controlling a forest pest in Canada. Of these examples, the ¢re protection case is regarded as one that closely follows the prescribed systems analysis methodology. 60 Hard systems thinking The ¢re protection study began in 1973 in Wilmington, DE and was conducted by a local project team with technical assistance from the New York-based RAND Institute. The eight existing ¢rehouses in Wilmington were getting old and the mayor wanted to ¢nd out if they o¡ered adequate protection, whether they were located in the right places and whether any new ¢rehouses needed building. The main objectives of ¢re protection were pretty obviously to protect lives and safeguard property while, at the same time, keeping costs low. Unfortunately, there was no reliable way of evaluating how di¡erent deployment strategies related directly to these objectives. Three ‘proxy’ measures were therefore developed: approximate travel time to individual locations, average travel time in a region and company workload. The consequences of changes in locations and numbers of ¢rehouses were then considered against these. The next stage required the analysts to build models that could be used to test various deployment alternatives. The primary tools employed to encap- sulate the data were a parametric allocation model, based on a mathematical formula for allocating companies to di¡erent regions, and a more descriptive, simulation model, known as the ¢rehouse site evaluation model. The transparency of this latter model was crucial as it enabled city o⁄cials to be involved in suggesting alternatives. The recommendations to close one of the ¢re companies and reposition most of the remainder provoked a long battle with the ¢re¢ghters union before they were eventually implemented. When the results were ¢nally evaluated it was found that the ¢re protection service was just as e¡ective as before, but with costs signi¢cantly reduced. 4.4 CRITIQUE OF HARD SYSTEMS THINKING Hard systems thinking has sought to bring scienti¢c rigour to the solution of management problems. It wants to produce objective results, free from the taint of personality and vested interests, through a process in which assump- tions, data and calculations are made clear, and which is validated in order to inform the work of other scientists facing similar problems. At the same time, hard systems thinkers are clear that they do not seek knowledge for its own sake. They do research aimed at serving the interests of clients, decision-makers and problem owners. This shift to valuing knowl- edge directly relevant to application rather than simply to the advancement of a scienti¢c discipline was revolutionary and enabled management scientists to steal a march on other disciplinary areas that, to this day, are Critique of hard systems thinking 61 still struggling to put in place the conceptual apparatus that would enable them to make their ¢ndings more relevant (e.g., see Tran¢eld and Starkey, 1998, on the struggle to establish more application-oriented management research). In tackling actual management problems hard systems thinkers pioneered the use of multidisciplinary teams of researchers and became advocates of an interdisciplinary approach. Particularly valuable to them, in this respect, was the existence of various systems ideas and concepts. Reductionism was useless because of the complexity and unbounded character of real-world problems and because of the interactive nature of their parts. What was required was a more holistic, integrating approach that sought to be compre- hensive by drawing the boundaries of the system of concern more widely. The systems language, employing concepts such as system, subsystem, hierarchy, boundary and control, was perfect for this purpose. Another problem that hard systems thinkers were able to overcome was how to test the hypotheses they developed. They could not carry out experiments directly on the systems they were hoping to improve ^ it was too dangerous because of expense, ethics or both. Unlike natural scientists, the problems they faced were too interconnected to be broken up and taken into the laboratory for analysis. The solution was to construct a model or models that accurately captured the behaviour of the real-world system and to run tests on those. Considerable progress had to be made by hard systems thinkers on the techniques of mathematical and computer- based modelling if this approach was to succeed. We considered the main weaknesses of hard systems thinking in Chapter 2, when we were looking at why other strands of systems thinking had emerged and established themselves over the last few decades. To recap, these related to the failure of hard systems thinking in the face of extreme complexity, multiple perceptions of reality and the need for radical change. The extreme complexity of the problem situations that managers confront, and the fact that they are subject to very di¡erent interpretations, frustrate hard systems thinkers in their search for an objective account of the system of concern that can be used to construct a mathematical model. Modelling is about simpli¢cation, but it is often not clear how complex problem situations can be simpli¢ed without bias creeping in. There is also the danger of leaving out of account factors that cannot be quanti¢ed. Hard systems approaches demand that the goal of the system of concern be clearly established before analysis can proceed. This makes it di⁄cult even to get started in many problem situations, where multiple stakeholders bring di¡erent perceptions to bear on the nature of the system and its 62 Hard systems thinking objectives. Hard systems thinking tends to leave the human aspect of systems aside. People are treated as components to be engineered, not as actors whose commitment must be won if solutions are to be implemented and plans realized. Hard systems thinking is also accused of conservatism. It privileges the values and interests of its clients and customers, and lends its apparent exper- tise to their realization. It thus gives the facade of objectivity to changes that help to secure the status quo. In general terms, despite its many strengths and achievements, hard systems thinking is today thought of as having a limited domain of applica- tion. It is ¢ne when world views converge and the problem becomes one of ¢nding the most e⁄cient means of arriving at agreed-on objectives. Such well-structured problems, usually arising at the tactical level in organizations, are meat and drink to an approach that employs a systematic methodology to seek out alternative means and evaluate them against well-de¢ned measures of performance. All this is easily understood when we recognize how totally hard systems thinking embraces the functionalist paradigm. Its interest is in ensuring the e⁄cient engineering of systems to achieve known goals. Their behaviour has to be predicted and they have to be regulated in pursuit of their con- trollers’ objectives. The concerns of the interpretive paradigm in bringing about mutual understanding among those with di¡erent values and beliefs, of the emancipatory paradigm in alleviating disadvantage and the post- modern paradigm in unpredictability and diversity, do not get a look-in. Within functionalism, hard systems thinking is further constrained by its adherence to the machine metaphor. The language of g