Systems thinking, systems practice

Systems thinking, systems practice

405 Book Reviews Peter C H E C K L A N D Systems Thinking, Systems Practice Wiley, Chichester, 1981, £11.95 Planning is a very old profession, but t...

282KB Sizes 0 Downloads 38 Views


Book Reviews Peter C H E C K L A N D

Systems Thinking, Systems Practice Wiley, Chichester, 1981, £11.95 Planning is a very old profession, but the twentieth century seems to be witnessing a shift in its methodology, although it is probably much too soon to recognize the exact nature of the shift. In any event, the way of labeling what planners do ~, clearly changing; the most prominent label by far is 'systems': systems analysis, systems approach, etc. Indeed, those planners who write books on the subject are well advised to mclude 'systems' somehow in the titles; 1 should know because two of my rec~,:nt books [1,2] have done just this. The idea seems to have struck Checkland as a good one since he has doubled the dose, so to speak. One of the phenomena surrounding the use of 'system' is that the literature on systems planning is quite fragmented. Some writers believe that systems analysis is a branch of mathematics, some that it is a foundation of biology, some that it is control theory, or linear programming, or operations research, and so on. I have been making an informal check on citations in the 'systems books' as they appear, and the lack of interconnections is quite apparent. Checkland is an exception. He has tried to include as many systems writers as he found useful, including the Frankfurt School of Sociology. There are a few omissions that 1 found strange, e.g., Donald N. Michael's On Learmng to P l a n - - a n d Planning to Learn [4], Erich Jantsch's The Self-Organizing Universe [3], and Donald A. Schon's Technology and Change [5]. But on the whole, the book is valuable as a compendium, since Checkland has a fine knack of explaining what other writers have tried to do. I'd like to make this review perform a rather Operational researchers wishing to revww books and pubhshers washing to have new books rewewed please conta¢t CB. Tdanus, Ezndhoven Unwerstty of TechnoloKv P 0 Box 513. 5600 M B Etndhoven, Netherlands. Tel. 31-40-473601.

North-Holland Ptlblishing Company European Journa', of Operatmnal Research 11 (1982) 405-412

unusual critical function. Checkland's main argument is that it is probably a mistake to assume that an organization has a set of problems and that planners should set about solving each problem, one by one. Furthermore, '"Hard' systems thmkmg assumes that 'the s ) s t e n f (together with Its sub-systems and the wider systems of which it is ~tself a part) are not problematic, they are 'obvious'. they can be taken as given, and the problem then resides m defining their objecuves and examimng alternative ways of meeting them. In ~ague problem s~tuatlons, ~t was again apparent, no ,,)stems tuerart.h~ relevant to the problem could be taken as given. Problem defiml,on again depended upon the part~tular vJe~ adopted a n d agam it seemed necessary to maAe that t'wwpmnt e~phttt and work out the systemic consequence,, of adopting it "" (p. 160).

Checkland proceeds to spell out the stage.,, of his 'soft' systems approach and to illustrate qmte clearly how his approach works. Instead of identifying places where 1 have some doubts or where 1 wish he had been clearer, I'd rather step aside and play the role of a reflective epistemolog~st because 1 believe the epistemology of planning is not hke the epistemology of the standard sciences at all: ~t is not empiricist, rationalist, or any of the derivatives or generalizations of these phdosophies. There is a common image of the deosion-maker, which is found in many texts. He ~s supposed to be faced by a number of feasible alternatives which are depicted as though they were so many doors which he can open; but the image tells us he can open only one, and that's it: the tiger or the lady. Each door leads to a set of results which vary in their quality. Some results have very good features; some, bad; all have a mixture of values for the decision-maker. All results need not occur at the same time. The image tells us how to find a unifying utility function over the results, which includes time and probabilities. Finally, the image tells us which door he should open. For example, Reno and Las Vegas provide a large number of options for the gamt~ler. If he brings a modern planner along, the planner can tell the decision-maker where to gamble--perhaps. The trouble is that for the managers of pubhc or private organizations the image is wrong, and for a very good reason. One can begm to under-

0377-2217/82/0000-0000/$02.75 © 1982 North-Holland


Book Rev:ew.~

stand the reason it one compares slot machines and blackjack. For the sake of the innocents, I need to explain that slot machines operate by the gambler's inserting a coin and pulling down a handle. Usually three pictures of, say, different fruits appear in a window. Certain sequences o[ the pictures pay money; certain do not. Blackjack is a game of cards between the gambler and the dealer. N o w in the case of the slot machine, once the gambler has pulled the handle, he has nothing to do but wait for the result. In the case of blackiack, however, once the gambler is dealt the initial cards, he usually ha~ to make another decision, namely, whether to ask for another c a r d - - h e does not just wait for the outcome. One should understand that there is quite a difference between these two gambling games. Perhaps, based on my own experience, 1 can put it this way: within reasonable limits, it really does not matter ~hether a gambler has had several drinks before playing a slot machine, but for most gamblers it matters a great deal when playing blackjack. In philosophical iaaguage, the slot machine is mechanical, i.e., ateleological, whereas blackjack is teleological. In management language, the slot machine needs no management once the handle is pulled, whereas blackjack requires management. In systems thinking, the slot machine and the gambler are a bounded system, but not the blackjack dealer and the gambler because the system of the dealer and the gambler unfolds into the system of the gambler and drinking, or of the gambler and fatigue, or of the gambler and intelligence and experience. Most managers, in addition to opening the doors, have to manage the passage to the results. This means that the 'management informatio,a" is totally different from a set of empirical data or management "principles'. The needed information is information about how to manage the situation that each door-opening demands. This means tha! 'management information' is incredibly difficult to obtain because it means knowing how to manage a plethora of other systems. Consider inventory, e.g., the 'problem' of how many 15-1/2. 33 white shirts to order for a men's store. The 'doors' are the integers from 0 to n. Operations-research texts tell the poor, unprepared students to ao,. some sort of inventory model. The texts point out that if the students open door 20, they may ~.. ve shirts that sit on the shelf for days

earning no money at all for the store. It they had opened door 10, they would have had in the form of cash ten times the cost of a shirt to the students. What could they do with the cash? Most texts tell them to use the current interest rate to measure the value of tbis 'lost opportunity'. But is that the best way to manage cash? Hence, the manager needs to have a cash flow model to solve the inventory problem. Now, one can understand why Checkland's approach is more realistic than the 'separate problem" approach. If 1 have some image (Weltanschauung) of the retail store, 1 may be able to make the necessary approximations of a measure like opportunity costs. Or, consider demand on inventory. Most texts tell students how to make probability distributions of past demands, thereby assuming that there is no need to manage demand. But, of course, any wise manager knows that managing demand by pricing, advertising, etc., is the heart of retailing. Another way to explain the unusual aspect of the epistemology of planning is that the conclusion of the intellectual effort is an action and not a proposition. A mathematician who proves a theorem ends his task with a set of symbols on a piece of paper; writing the symbols appropriately is the only action he takes. An experimenter also writes symbols, the numbers he obtains from his observations and the analysis he performs of his results. But the planner has to consider the actions of humans and for this purpose, according to Checkland, must use the construct of a "human activity system'. He may derive from his method an instruction that reads 'reduce manpower 25% in Division 1', but that instruction is not the action. The action consists of managing personnel reduction so that it does not harm the organization and the people who are to be fired. At the end Checkland sujgests a conversation between Vickers, Habermas, and me. Of course, since Vickers has died, the conversation cannot be real, but it well could be imagined. I find imagined conversations useful in systems work. They enable one to bnng out and highlight conflicting and complementary ideas. I do hope Checkland will try his hand at this.

C West CHURCttMAN University of California Berkeley, CA, U.S.A.

Book Ret'te. s

References [I] C.W. Churchman, The Destgn of inqulrtng Systems I Baslc Books, New York, 1968). [2] C.W. Churchman, The S w'tems Approath and Its Enemte.~ (Basic Books, New York, 1979), [31 E. Jantsch, The Self-Orgamzmg Umverse (Pergamon Pres,,, New York, 1980). [4] D.N. Michael, On Learning to Plan --and Phmmng to Learn (Jossey-Bass. San Franosco, 1973) [5] D.A. Schon. Technologe and Change (Delacorte Press, New York, 1967),

William E. S O U D E R

Management Decision Meflaods: For Managers o| Engineering and Research Van Nostrand Reinhold, Ne~, York, 1980, xii + 317 pages, £19.15 R & D managers continuously invest in the future. They are concerned with estimates of the likelihood of the technical and commercial success and of capacity and resources needed in the future, with the development of legislation, and with how all these factors affect the whole R & D programme, etc. On the other hand they feel.that what they are doing will actually change the future to a certain extent. How can these uncertainties be included in a decision-making process? Professor S o u d e r ~ a n authority in this field-attempts to solve this problem by means of a Structured Decision Process (SDP) based on combined decisions concerning technologies and people and taking the different behavioural aspects into account. The author divides the decision process into a sequence of seven steps (problem deftnition, specification of the goals to be achieved. enumeration of possible decisions, evaluation of alternatives, selection, post-optimal analysis and controlled implementation). This considerably reduces most of the specific shortcomings of traditional management and helps managers to maintain strong leadership. A first introductory part in which the foundations of the SDP are laid and the usual decision tech.-.iques described, is followed by a chapter stressing the great need for a structured decision process. In group and organizational settings too many behavioural factors interfere with rationality. The explicit consideration of the impact of group dynamic principles governing the interdependence of individuals within a group is a new


aspect in quantitative decision-making, and enabling highly complex processes to be kept under rational control. Parts Ill and IV tackle the core of the problem m detail, namely the preparation of a decision and a systematic approach to planning and budgeting, and give a lucid description of the methods used. Key factors like cost, benefit, value and risk, characterizing the potential implications of each alternative decision are analysed. The state of the art in scheduling methods is summarized in Part V, while Part VI concentrates on resource allocation and management and related problems O.e. inventory and resource programming). The management of the decision system itself is dealt w~th in the last part, m which the main factors affecting its effectiveness (quality, implementation and contr,~l of a decision) are described. Subjective probabilities are used as a vehicle to reach expected values of decision alternative:c However, another approach to demsion-makmg ~s possible, though not mentioned. A well-known trend rejects the use of the subjective probability of a consequence of a particular decision as unsound, because any decision could have an infinite range of consequences, many of which are perfectly possible, The decision-maker has no valid reason for disregarding any of them. so that the decision criteria should be the value, or obstructi,~eness, of each decision sequel. This methodology unfortunately cannot be tied into a simple mathematical formulation as precisely as can subjective probability, except perhaps by means of the fuzzy-sets theory. The presentation is concise, the graphics used efficient, and the tables clear. This, in all respects, excellent "handbook of management decisaonmaking" must be recommended for a wide readership ranging from students of business admimstration to R& D group leaders. Roherto A.A. B O S C H ! CIBA -GEIG )" A. G. Basle. Swatzerhmd