paper msb

8/6/2019 Paper MSB

1/48

Systems Research and Behavioral Science

Syst. Res. 17, S11S58 (2000)

Copyright 2000 John Wiley & Sons, Ltd.

Research Paper

Soft Systems Methodology: A Thirty YearRetrospectivea

Peter Checkland*

25 Pinewood Avenue, Bolton-le-Sands, Carnforth, Lancashire, LA5 8AR, UK

INTRODUCTION

Although the history of thought reveals a numberof holistic thinkers Aristotle, Marx, Husserlamong them it was only in the 1950s that anyversion of holistic thinking became insti-tutionalized. The kind of holistic thinking whichthen came to the fore, and was the concern of anewly created organization, was that whichmakes explicit use of the concept of system, andtoday it is systems thinking in its various formswhich would be taken to be the very paradigmof thinking holistically. In 1954, as recounted inChapter 3 of Systems Thinking, Systems Practice,only one kind of systems thinking was on the

table: the development of a mathematicallyexpressed general theory of systems. It was sup-posed that this would provide a meta-level lan-guage and theory in which the problems of manydifferent disciplines could be expressed andsolved; and it was hoped that doing this wouldhelp to promote the unity of science.

These were the aspirations of the pioneers, butlooking back from 1999 we can see that the projecthas not succeeded. The literature contains verylittle of the kind of outcomes anticipated by thefounders of the Society for General Systems

Research; and scholars in the many subject areasto which a holistic approach is relevant have beenunderstandably reluctant to see their pet subjectas simply one more example of some broadergeneral system!

*Correspondence to: Peter Checkland, 25 Pinewood Avenue, Bolton-le-Sands, Carnforth, Lancashire, LA5 8AR.aReproduced from Soft Systems Methodology in Action, John Wiley &Sons, Ltd, Chichester, 1999.

But the fact that general systems theory (GST)has failed in its application does not mean thatsystems thinking itself has failed. It has in factflourished in several different ways which werenot anticipated in 1954. There has been devel-opment of systems ideas as such, development ofthe use of systems ideas in particular subjectareas, and combinations of the two. The devel-opment in the 1970s by Maturana and Varela(1980) of the concept of a system whose elementsgenerate the system itself provided a way of cap-turing the essence of an autonomous living sys-tem without resorting to use of an observersnotions of purpose, goal, information pro-cessing or function. (This contrasts with the the-

ory in Millers Living Systems (1978), whichprovides a general model of a living entity expre-ssed in the language of an observer, so that whatmakes the entity autonomous is not central tothe theory.) This provides a good example of thefurther development of systems ideas as such.The rethinking, by Chorley and Kennedy (1971),of physical geography as the study of the dynam-ics of systems of four kinds, is an example of theuse of systems thinking to illuminate a particularsubject area.

This paper provides an example of the third

kind of development: a combination of the twoillustrated above. We set out to see if systemsideas could help us to tackle the messy problemsof management, broadly defined.

In trying to do this we found ourselves havingto develop some new systems concepts as aresponse to the complexity of the everyday prob-lem situations we encountered, the kind of situ-ations which we all have to deal with in both our

8/6/2019 Paper MSB

2/48

professional and our private lives. The aim in theresearch process we adopted was to make neitherthe ideas nor the practical experience dominant.Rather the intention was to allow the tentativeideas to inform the practice which then becamethe source of enriched ideas and so on, round

a learning cycle. This is the action research cyclewhose emergence is described in Systems Think-ing, Systems Practice and whose use and furtherdevelopment is the subject of SSM in Action.

The action research programme at LancasterUniversity was initiated by the late Gwilym Jenk-ins, first Professor of Systems at a British univer-sity, and Philip Youle, the perspicacious managerin ICI who saw the need for the kind of col-laboration between universities and outsideorganizations which the action research pro-gramme required. Thirty years later that pro-

gramme still continues, and with the same aim:to find ways of understanding and coping withthe perplexing difficulties of taking action, bothindividually and in groups, to improve the situ-ations which day-to-day life continuously createsand continually changes. Specifically, the pro-gramme explores the value of the powerful bun-dle of ideas captured in the notion system, andthey have not been found wanting, though boththe ideas themselves and the ways of using themhave been extended as a result of the practicalexperiences.

The progress of the 30 years of research hasbeen chronicled and reflected upon since 1972 inabout 100 papers and four books which willbe referred to in the remainder of this chapter bythe initials of their titles. The nature of the booksis summarized briefly below.

Systems Thinking, Systems Practice (STSP)(Checkland 1981) makes sense of systems think-ing by seeing it as an attempt to avoid thereductionism of natural science, highly successfulthough that is when investigating naturalphenomena; it describes early experiences of try-ing to apply systems engineering outside thetechnical area for which it was developed, therethinking of systems thinking which earlyexperience made necessary, and sets out the firstdeveloped form of SSM as a seven-stage processof inquiry.

Systems: Concepts, Methodologies and Appli-

S12 Peter Checkland

Copyrightc 2000 John Wiley & Sons, Ltd. Syst. Res. 17, S11 S58 (2000)

RESEARCH PAPER Syst. Res.

cations (SCMA) (Wilson 1984, 2nd Edn 1990)describes the response of a professional controlengineer to experiences in the Lancaster pro-gramme of action research; less concerned withthe human and social aspects of problem situ-ations, it cleaves to the functional logic of engin-

eering and presents an approach which Holwell(1997) argues is best viewed as classic systemsengineering with the transforming addition ofhuman activity system modelling.

Soft Systems Methodology in Action (SSMA)(Checkland and Scholes 1990) describes the use ofa mature SSM in both limited and wide-rangingsituations in both public and private sectors; itmoves beyond the seven-stage model of themethodology (still useful for teaching purposesand occasionally in some real situations) tosee it as a sense-making approach, which, once

internalized, allows exploration of how peoplein a specific situation create for themselves themeaning of their world and so act intentionally;the book also initiates a wider discussion of theconcept of methodology, a discussion whichwill be extended below.

Information, Systems and Information Systems(ISIS) (Checkland and Holwell 1998) stems fromthe fact that in very many of the Lancaster actionresearch projects the creation of information sys-tems was usually a relevant, and often a core,concern; it attempts some conceptual cleansing

of the confused field of IS and IT, treating IS asbeing centrally concerned with the human act ofcreating meaning, and relates experiences basedon a mature use of SSM to a fundamental con-ceptualization of the field of IS/IT; it carries for-ward the discussion of SSM as methodology butless explicitly than will be attempted here.

It is important to understand the nature ofthese books if the aim of this chapter is itself tobe properly understood. The less than impressive but nevertheless sprawling literature of man-agement caters in different ways for several dif-ferent audiences. There is an apparent insatiableappetite for glib journalistic productions, offeringclaimed insights for little or no reader effort Distribution Management in an Afternoon: that

kind of thing. Such books are more often pur-chased than actually read. There is also a needfor textbooks which systematically display the

8/6/2019 Paper MSB

3/48

conventional wisdom of a subject for aspiringstudents. These need to be updated periodicallyin new editions. And also, more austerely, thereare books which carry the discussion which is thereal essence of any developing subject, and try toextend the boundaries of our knowledge. The

books described above are of this kind. It is notusually appropriate as it is with textbooks to update them in new editions. They are of their

time. But it is useful on republication to offerreflections on the further development of theideas as new experiences have accumulated sincethe books were written. That is what is done herefor STSP and SSMA.

A particular structure is adopted. First, theemergence of soft systems thinking is brieflyrevisited. Then the methodology as a whole isconsidered, since the way in which it is thought

about now is very different from the view of it inthe 1970s, when it was a redefined version ofsystems engineering. This consideration of themethodology as a whole frames reflection on theseparate parts which make up the whole (Analy-ses One, Two, Three; CATWOE; rich pictures;the three Es, etc.). This in turn yields a richerunderstanding of both the whole and its context.Such a structure, in which an initial considerationof the whole leads to an understanding of theparts, which in turn enables a richer under-standing of the whole to be gained, is itself an

example of Diltheys hermeneutic circle (Muel-ler-Vollmer 1986; Morse 1994, Chs 7 and 8). Here,it is a modest reflection of the same processthrough which SSM was itself developed, a pro-cess which tried to ensure that both whole andparts were continually honed and refined incycles of action.

THE EMERGENCE OF SOFT SYSTEMSTHINKING

The Starting Position

In the culture of the UK the word academic ismore often than not used in a pejorative sense.To describe something as academic is usually tocondemn it as unrelated to the rough and tumbleof practical affairs. This was certainly the outlook

S13Soft Systems Methodology


Syst. Res. RESEARCH PAPER

of Gwilym Jenkins when he moved to LancasterUniversity in the mid-1960s to found the firstsystems department in a UK university. He didnot want a department which could be dismissedas academic. He rejected the idea that the nameof the department should be Systems Analysis, in

favour of a Department of Systems Engineering.Analysis is not enough, he used to say hereti-cally. Beyond analysis it is important to putsomething together, to create, to engineersomething. Given this attitude it was not sur-prising that he initiated the programme of actionresearch in real-world organizations outside theuniversity. The intellectual starting point wasOptners concept (1965) that an organizationcould be taken to be a system with functionalsub-systems concerned with production, mar-keting, finance, human resources, etc. Jenkins

idea was that the real-world experiences wouldenable us gradually to build up knowledge ofsystems of various kinds: production systems,distribution systems, purchasing systems, etc.and that this knowledge would support the betterdesign and operation of such systems in real situ-ations. History did not, however, unfold in thisway. Instead, the practical experiences led us toreject the taken-as-given assumption underlyingthe initial expectation, so taking the thinking in avery different direction. In doing this we had todistinguish between two fundamentally different

stances within systems thinking: the two outlooksnow known as hard and soft systems thinking.

At the outset, by formulating a research aim touncover the fundamental characteristics of sys-tems of various kinds, we were making theunquestioned assumption that the world con-tained such systems. Along with this went asecond assumption that such systems could becharacterized by naming their objectives. It seemsobvious, for example, that a production systemwill have objectives which can be expressed as:to make product X with a certain quality, at acertain rate, with a certain use of resources, undervarious constraints (budgetary, legal, environ-mental, etc.). Given such an explicit definition ofan objective, then a system can in principle beengineered to achieve that end. This is the stanceof classic systems engineering (as described inChapter 5 of STSP). This was what constituted

8/6/2019 Paper MSB

4/48

systems thinking at the time our research star-ted, and its origins, as far as application to organ-izations goes, lie in the great contribution tomanagement science made by Herbert Simon inthe 1950s and 1960s (Simon 1960, 1977), whichpropounded the clarifying (but ultimately lim-

ited) concept that managing is to be thought of asdecision-taking in pursuit of goals or objectives.

The Learning Experience

We found that although we were armed with themethodology of systems engineering and wereeager to use its techniques to help engineer real-world systems to achieve their objectives, themanagement situations we worked in werealways too complex for straightforward appli-

cation of the systems engineering approach. Thedifficulty of answering such apparently simplequestions as: What is the system we are con-cerned with? and What are its objectives? wasusually a reason why the situation in questionhad come to be regarded as problematical. Wehad to accept that in the complexity of humanaffairs the unequivocal pursuit of objectiveswhich can be taken as given is very much theoccasional special case; it is certainly not thenorm. A current long-running example of the sur-prising difficulty in using the language of objec-

tives in human affairs is provided by thearguments which wax and wane over the Com-mon Agricultural Policy (CAP) of the EEC. TheTreaty of Rome boldly declares that the CAP hasthree equally important objectives: to increaseproductivity in the agricultural industry; to safe-guard jobs in the industry; and to provide thebest possible service to the consumer. No wonderthe CAP is a constant source of never resolvedissues: progress towards any one of its (equallyimportant) objectives will be at the expense of theother two! This is typical of the complexity wemeet in human affairs as soon as we move out ofthe more straight-forward area in which prob-lems can be technically defined: e.g. increase asmuch as possible the productivity of this phthalicanhydride plant, or make a device to produceradio waves with a 10 cm wavelength. (If youinsisted on using the language of objectives, you

S14 Peter Checkland



would have to conclude that the objective of theCAP is constantly to maintain and adjust a balance between the three incompatible objec-tives which is politically acceptable which isnot a very useful definition for engineeringpurposes.)

It was having to abandon the classic systemsengineering methodology which caused us toundertake the fundamental thinking described inChapters 24 of STSP. And it was this rethinkwhich led ultimately to the distinction betweenhard and soft systems thinking.

Four Key Thoughts

The process of learning by relating experience toideas is always both rich and confusing. But as

long as the interaction between the rhetoric andthe experienced reality is the subject of con-scious and continual reflection, there is a goodchance of recognizing and pinning down thelearning which has occurred. Looking back at thedevelopment of SSM with this kind of reflectivehindsight, it is possible to find four key thoughtswhich dictated the overall shape of the devel-opment of SSM and the direction it took (Check-land 1995).

Firstly, in getting away from thinking in termsof some real-world systems in need of repair or

improvement, we began to focus on the fact that,at a higher level, every situation in which weundertook action research was a human situationin which people were attempting to take pur-poseful action which was meaningful for them.Occasionally, that purposeful action might be thepursuit of a well-defined objective, so that thisbroader concept included goal seeking but was notrestricted to it. This led to the idea of modellingpurposeful human activity systems as sets oflinked activities which together could exhibit theemergent property of purposefulness. Ways ofbuilding such models were developed.

Secondly, as you begin to work with the ideaof modelling purposeful activity in order toexplore real-world action it quickly becomesobvious that many interpretations of anydeclared purpose are possible. Before modellingcan begin choices have to be made and declared.

8/6/2019 Paper MSB

5/48

Thus, given the complexity of any situation inhuman affairs, there will be a huge number ofhuman activity system models which could bebuilt; so the first choice to be made is of whichones are likely to be most relevant (or insightful)in exploring the situation. That choice made, it is

then necessary to decide for each selected pur-poseful activity the perspective or viewpointfrom which the model will be built, the Wel-tanschauung upon which it is based. Thus whenDavid Farrah, a director of the then British Air-craft Corporation asked us to use our systemsengineering approach to see how the Concordeproject might be improved, possible relevant sys-tems might have included a system to managerelations with the British Government (since theywere funding it) or a system to sustain a Eur-opean precision engineering industry (since

Concorde would help to stimulate such activity).Thinking like systems engineers at the time (Whatis the system? What are its objectives?) DaveThomas and I in fact proceeded only with themost basic and obvious of possible choices: asystem to carry out the project. Neither did thesecond choice give us pause: how would we con-ceptualize that project? Again, with our systemsengineering blinkers firmly in place, it did notoccur to us to think of it as anything other thanan engineering project. But given its origins, at atime when President de Gaulle of France was

vetoing British entry into the European CommonMarket, a defensible alternative world-viewwould be to treat it as a political project. On theday the Concorde project agreement was signedthe British Government let it be known that itexpected Britain to join the European Communitywithin a year, while de Gaulle a few weeks latertold a press conference that it was probable thatnegotiations for British entry might not succeed;in fact he made the supersonic aircraft project atouchstone of Britains sincerity in applying formembership (Wilson 1973, pp. 3132). So a modelof the project based on a political world-viewmight be as useful as or perhaps more usefulthan the more obvious one based on a tech-nical world-view.

The learning here was that in making the ideaof modelling purposeful activity a usableconcept, we had to accept that it was necessary to




declare both a world-view which made a chosenmodel relevant, and a world-view which wouldthen determine the model content. Equally,because interpretations of purpose will alwaysbe many and various, there would always be anumber of models in play, never simply one

model purporting to describe what is the case.This moved us a good way away from classic

systems engineering, and the next key thought inunderstanding our experience recognized this. Itwas the thought which can now be seen to haveestablished the shape of SSM as an inquiring pro-cess. And that in turn established the hard/softdistinction in systems thinking, though that toowas not immediately recognized at the time.

We had moved away from working with theidea of an obvious problem which requiredsolution, to that of working with the idea of a

situation which some people, for various reasons,may regard as problematical. We had developedthe idea of building models of concepts of pur-poseful activity which seemed relevant to makingprogress in tackling the problem situation. Next,since there would always be many possible mod-els it seemed obvious that the best way to proceedwould be to make an initial handful of modelsand conscious of them as embodying onlypure ideas of purposeful activity rather thanbeing descriptions of parts of the real world touse them as a source of questions to ask of the

real situation. SSM was thus inevitably emergingas an organized learning system. And since theinitial choice of the first handful of models, whenused to question the real situation, led to newknowledge and insights concerning the problemsituation, this leading to further ideas for relevantmodels, it was clear that the learning process wasin principle ongoing. What would bring it to anend, and lead to action being taken, was thedevelopment of an accommodation among peo-ple in the situation that a certain course of actionwas both desirable in terms of this analysis andfeasible for these people with their particular his-tory, relationships, culture and aspirations.

SSM thus gradually took the form shown inFigure 1.3 of SSMA (p. 7), repeated with someembellishment here as Figure A1. This was theform of representation of SSM which eventuallytook hold, and is the one now normally used. The

8/6/2019 Paper MSB

6/48

Figure A1. The inquiring/learning cycle of SSM

initial version of it was the seven-stage modelwhich is shown in Figure 6 in STSP, p. 163 andFigure 2.5 in SSMA, p. 27. This version, thoughstill often used for initial teaching purposes, hasa rather mechanistic flavour and can give the falseimpression that SSM is a prescriptive processwhich has to be followed systematically, henceits fall from favour.

These three key thoughts capture succinctly thelearning which accumulated with experience ofusing SSM, and they make sense of its devel-

S16 Peter Checkland



opment. The fourth such thought, that models ofpurposeful activity can provide an entry to workon information systems (which are less than idealin virtually every real-world situation) is not ourconcern here, this aspect of SSMs use being thedetailed subject of ISIS.

Hard and Soft Systems Thinking

Our final concern in this section is the majorthought which came from these particular experi-

8/6/2019 Paper MSB

7/48

ences of relating systems thinking to systemspractice: the hardsoft distinction. This wasfirst sharply expressed in a paper written twoyears after the publication of STSP in 1981(Checkland 1983). It took some time for this ideato sink in!

In systems engineering (and also similarapproaches based on the same fundamentalideas, such as RAND Corporation systems analy-sis and classic OR) the word system is usedsimply as a label for something taken to exist inthe world outside ourselves. The taken-as-givenassumption is that the world can be taken to be aset of interacting systems, some of which do notwork very well and can be engineered to work better. In the thinking embodied in SSM thetaken-as-given assumptions are quite different.The world is taken to be very complex, prob-

lematical, mysterious. However, our coping withit, the process of inquiry into it, it is assumed, canitself be organized as a learning system. Thus theuse of the word system is no longer applied tothe world, it is instead applied to the processof our dealing with the world. It is this shift ofsystemicity (or systemness) from the world tothe process of inquiry into the world which is thecrucial intellectual distinction between the twofundamental forms of systems thinking, hardand soft.

In the literature it is often stated that hard

systems thinking is appropriate in well-definedtechnical problems and that soft systems think-ing is more appropriate in fuzzy ill-defined situ-ations involving human beings and culturalconsiderations. This is not untrue, but it doesnot define the difference between hard and softthinking. The definition stems from how theword system is used, that is from the attributionof systemicity.

Experience shows that this distinction is a slip-pery concept which many people find it very hardto grasp; or, grasped one week it is gone the next.Probably this is because very deeply embeddedwithin our habits is the way we use the wordsystem in everyday language. In everyday talkwe constantly use it as if it were simply a label-word for a part of the world, as when we talkabout the legal system, health care systems, theeducation system, the transport system, etc. even




though many of these things named as systemsdo not in fact exhibit the characteristics associatedwith the word system when it is used properly.This day-by-day use unconsciously but steadilyreinforces the assumptions of the hard systemsparadigm; and the speaking habits of a lifetime

are hard to break!As the thinking about SSM gradually evolved,

the formation of this precise definition of hardand soft systems thinking did not arrive in thedramatic way events unfold in adventure storiesfor children (With one bound, Jack was free!).Rather the ultimate definition is the result of ourfeeling our way to the difference between hardand soft, as experience accumulated, via a number of different formulations. These have beenspotted and extracted by Holwell (1997, Table4.2, p. 126) who collects eight different ways of

discussing the hard/soft distinction between1971 and 1990. These begin unpromisingly judged by todays criteria by assuming thathard and soft systems (roughly, determinateand indeterminate respectively) exist in theworld. The shift in thinking comes between thepublication of STSP and SSMA, its very firstexplicit appearance being in Checkland (1983), apaper which can now also be seen as part of thedevelopments which have made the phrase softOR meaningful.

The eventual definition of the hard-soft dis-

tinction is succinctly expressed in Figure 2.3 ofSSMA (p. 23), but this diagram is over-rich formany, and so here it is supplemented by FigureA2, a further attempt to make clear the differencebetween hard systems thinking and soft systemsthinking. Understanding this idea is the crucialstep in understanding SSM.

SOFT SYSTEMS METHODOLOGY THEWHOLE

Learning from books or lectures is relatively easy,at least for those with an academic bent, but learn-ing from experience is difficult for everyone.Everyday life develops in all of us trusted intel-lectual structures which to us seem good enoughto make sense of our experiences, and in generalwe are reluctant to abandon or modify them even

8/6/2019 Paper MSB

8/48

Figure A2. The hard and soft systems stances

when new experience implies that they are shaky.Even professional researchers, who ought to be

ready to welcome change in taken-as-given struc-tures of thinking, show the same tendency todistort perceptions of the world rather thanchange the mental structures we use to give usour bearings. So we were lucky in our researchprogramme that the failure of classic systemsengineering in rich management problem situ-ations, broadly defined, was dramatic enough tosend us scurrying to examine the adequacy of thesystems thinking upon which systems engin-eering was based. (The early experiences aredescribed in STSP, Chapter 7.) But in spite of thisit is still the case that the story of our learning isalso the story of our gradually managing to shedthe blinkered thinking which we started out withas a result of taking classic systems engineeringas given.

Holwell (1997) has an appendix to her thesiswhich collects four different representations of

S18 Peter Checkland



SSM between 1972 and 1990 and correctly sug-gests that these show how the methodology has

become less structured and broader as it hasdeveloped (p. 450). It is useful briefly to reviewthis changing perception of the methodology asa whole before moving on to a consideration ofits parts.

1972 Blocks and Arrows

The first studies in the research programme werecarried out in 1969, and the first account of whatbecame SSM (though that phrase was not used atthe time) was published three years later in apaper: Towards a systems-based methodologyfor real-world problem solving (Checkland1972). The paper argues the need for meth-odology of practical use in real-world problems[sic](p. 88), reviews the context provided by thesystems movement, introduces the case for action

8/6/2019 Paper MSB

9/48

research as the research method, describes threeprojects in detail, refers to six others, anddescribes the emerging methodology. It finisheswith the very important argument that any meth-odology which will be used by human beingscannot, as methodology, be proved to be useful:

Thus, if a reader tells the author I have used yourmethodology and it works, the author will have toreply How do you know that better results mightnot have been obtained by an ad hoc approach? Ifthe assertion is: The methodology does not workthe author may reply, ungraciously but with logic,How do you know the poor results were not duesimply to your incompetence in using the meth-odology? (p. 114)

With reference to human situations, neither ofthese questions can be answered. Methodology,as such, remains undecidable.

Nearly 30 years later the paper has a somewhatquaint air, though not embarrassingly so. Apartfrom the reference noted above to real-worldproblems, rather than problem situations, themain inadequacy now is in the legacy of hardsystems thinking which leads to reference beingmade to both hard systems and soft systems asexisting in the real world; thus we find a fewremarks of the kind: In soft systems like those ofthe three studies under discussion. . . . (p. 96).Such statements would not have been made a fewyears later. Also the methodology is presented

as a sequence of stages with iteration back toprevious stages, the sequence being: analysis;root definition of relevant systems; con-ceptualization; comparison and definition ofchanges; selection of change to implement; designof change and implementation; appraisal.

The focus on implementing change rather thanintroducing or improving a system is a signal thatthe thinking was on the move as a result of theseearly experiences, even if the straight arrows inthe diagrams and the rectangular blocks in someof the models do now cause a little pain!

1981 Seven Stages

By the time the first book about SSM was written(STSP, 1981) the engineering-like sequence of the1972 paper was being presented as a cluster of




seven activities in a circular learning process: theseven-stage model, versions of which are Figure6 in STSP (p. 163) and Figure 2.5 in SSMA (p. 7).In this model the first two stages entail enteringthe problem situation, finding out about it andexpressing its nature. Enough of this has to be

done to enable some first choices to be made ofrelevant activity systems. These are expressed asroot definitions in stage three and modelled instage four. The next stages use the models tostructure the further questioning of the situation(the stage five comparison) and to seek to definethe changes which could improve the situation,the changes meeting the two criteria of desirablein principle and feasible to implement (stagesix). Stage seven then takes the action to improvethe problem situation, so changing it andenabling the cycle to begin again. The arrows

which link the seven stages simply show the logi-cal structure of the mosaic of actions which makeup the overall process; it has always beenemphasized that the work done in a real studywill not slavishly follow the sequence from stageone to stage seven in a flat-footed or dogged way.Thus, to give one example, the stage five com-parison cannot but enhance the finding out aboutthe situation, leading to new ideas for more rel-evant systems to model. Similarly, the processcan take a real-world change being implementedto be an example of stage seven; you can then

work backwards to construct the notional com-parison which would lead to this change beingselected, thus teasing out what world-views arebeing taken as given by people in the situation.

The seven-stage model of SSM has provedresilient, not least because it is easy to understandas a sequence which unfolds logically. This makesit easy to teach, and that too helps explain itsresilience. Certainly it has three virtues worthnoting before we begin to undermine it in whatfollows.

Firstly an intangible, aesthetic point, but animportant one its fried-egg shapes and curvedarrows begin to undermine the apparent certaintyconveyed by straight arrows and rectangularboxes. These are typical of work in science andengineering, and the style conveys the impli-cation: this is the case. The more organic styleof the seven-stage model (and of the rich pictures

8/6/2019 Paper MSB

10/48

and hand-drawn models in SSMA) is meant toindicate that the status of all these artefacts is thatthey are working models, currently relevant nowin this study, not claiming permanent ontologicalstatus. They are also meant to look more human,more natural than the ruled lines and right angles

of science and engineering.Secondly, it is a happy chance that the learning

cycle of this model of the process has sevenstages. Millers well-known account of laboratoryexperiments on perception (1956) suggests thatthe channel capacity of our brains is such that wecan cope with about seven items or concepts atonce, hence the title of his famous paper: Themagical number seven, plus or minus two: somelimits on our capacity for processing infor-mation. (He reminds us that there are seven daysof the week, seven wonders of the world, seven

ages of man, seven levels of hell, seven noteson the musical scale, seven primary colours. . . .)Irrespective of whether or not seven is truly acrucial number in human culture, thecomfortablesize of the model of the SSM process does meanthat you can easily retain it in your mind. You donot have to look it up in a book, and this is veryuseful when using it flexibly in practice.

Another feature of the seven-stage model wor-thy of note is that the stages of forming mostdefinitions and building models from them(stages three and four) were separated from the

other stages by a line which separates the sys-tems thinking world below the line from theeveryday world of the problem situation abovethe line. This distinguishing between the every-day world and the systems thinking about it wasintended to draw attention to the conscious use ofsystems language in developing the intellectualdevices (the activity models) which are con-sciously used to structure debate. The purpose ofthe line was essentially heuristic, and its elim-ination from the 1990 model of SSM will be dis-cussed later in this paper.

Finally, as far as the 1981 model is concerned,it was important at that stage of development tothink about what it was you had to do in a sys-tems study if you wished to claim to be usingSSM. This problem was first addressed byNaughton (1977). He was tackling the problem ofteaching SSM to Open University students, and

S20 Peter Checkland



for the sake of clarity in teaching, distinguished between Constitutive Rules which had to beobeyed if the SSM claim was to be made, andStrategic Rules which allowed a number ofoptions among which the user could choose. Ver-sions of these rules endorsed in STSP are given

in Table 6 (p. 253). This was a very useful devel-opment in its time, though this is another areawhich will be further discussed in the light ofcurrent thinking.

In summary, formulation of SSM in the 1981book was at least rich enough to enable it to betaught and used; accounts began to appear ofuses of SSM by people other than its early devel-opers. See, for example, Watson and Smith (1988)for an account of 18 studies carried out in Aus-tralia between 1977 and 1987.

1988 Two Streams

All of the action research which developed andused SSM was carried out in the spirit of GwilymJenkins remark quoted earlier, that Analysis isnot enough. The overall aim in all the projectsundertaken was to facilitate action, and it wasalways apparent that making things happen inreal situations is a complex and subtle process,something which will not happen simply becausesome good ideas have been generated or a soph-

isticated analysis developed. Ideas are not usu-ally enough to trigger action and that is whyindustrial companies value highly their shakersand movers: they are a much rarer breed thanintelligent analysts. So, although a debate struc-tured by questioning perceptions of the real situ-ation by means of purposeful activity models wasalways insightful, moving on to action entailedbroader considerations.

In the very first research in the programme, forexample, in the failing textile company describedin Checkland and Griffin (1970) and in STSP (p.156), we were brought into the situation by arecently appointed marketing director. He hadbeen brought into the company because the crisisdue to falling revenues and disappearing pro-fitability had at last been recognized by a rela-tively unsophisticated and rather inbred groupof managers. This was the first instance in that

8/6/2019 Paper MSB

11/48

companys history of appointing a senior man-ager from outside. The newcomer was thus notpart of what had become a closed tribe, andthough his previous experience gave him manyideas relevant to improving company perform-ance, his effectiveness was profoundly affected

by suspicion of the off-comer. Understandingthat, and taking it into account in influencingthinking in the company was crucial to initiatingaction.

It was thus important always to gain an under-standing of the culture of the situations in whichour work was done. For some years this was doneinformally, but we hoped with insightsfrom experience, since all the original actionresearchers developing SSM were ex-managersrather than career academics who are oftennave about life in unsubsidized organizations.

During those years much reflection went onconcerning how we went about reading situ-ations culturally and politically, and it was a sig-nificant step forward when SSM was presentedas an approach embodying not only a logic-basedstream of analysis (via activity models) but alsoa cultural and political stream which enabled judgements to be made about the accom-modations between conflicting interests whichmight be reachable by the people concerned andwhich would enable action to be taken. This two-stream model of SSM (SSMA, Figure 2.6, p. 29)

was first expounded at a plenary session of theAnnual Meeting of the International Societyfor General Systems Research in 1987, andwas published the following year (Checkland1988).

This version of SSM as a whole recognizes thecrucially important role of history in humanaffairs. It is their history which determines, for agiven group of people, both what will be noticedas significant and how what is noticed will be judged. It reminds us that in working in realsituations we are dealing with something whichis both perceived differently by different peopleand is continually changing.

Also, it is worth noting that this particularexpression of SSM as a whole omits the dividingline between the world of the problem situationand the systems thinking world. It had served itsheuristic purpose.




1990 Four Main Activities

Published in 1981, STSP covered broadly the firstdecade of development of SSM. The seven-stagemodel gave a version of the approach which wasby then sufficiently well founded to be applied

in new real-world situations, large and small, inboth the public and the private sector. That waswhat happened during the second decade ofdevelopment, some of those experiences beingdescribed in SSMA. They cover action researchin different organizational settings (industry, theCivil Service, the NHS) and include involvementswhich took from a few hours (ICL, Chapter 6, pp.164171) to more than a year (Shell, Chapter 9).

When it came to expressing the shape of themethodology in the 1990 book, the seven-stagemodel was no longer felt able to capture the now

more flexible use of SSM; and even the two-stre-ams model was felt to carry a more formal airthan mature practice was now suggesting char-acterized SSM use, at least by those who hadinternalized it. The version presented was thefour-activities model (SSMA, Figure 1.3, p. 7) ofwhich Figure A1 in this chapter is a contemporaryform. This is iconic rather than descriptive, andsubsumes the cultural stream of analysis in thefour activities, which it implies rather thandeclares.

The four activities are, however, capable of

sharp definition:

1. Finding out about a problem situation, includ-ing culturally/politically;

2. Formulating somerelevant purposeful activitymodels;

3. Debating the situation, using the models, seek-ing from that debate both(a) changes which would improve the situ-

ation and are regarded as both desirableand (culturally) feasible, and

(b) the accommodations between conflicting

interests which will enable action-to-improve to be taken;

4. Taking action in the situation to bring aboutimprovement.

((a) and (b) of course are intimately connectedand will gradually create each other.)

A decade after SSMA was published this iconic

8/6/2019 Paper MSB

12/48

model of SSM is still relevant. Why that is so willbe discussed when we return to discussing themethodology as a whole. But first it is usefulto review the evolving thinking about the partswhich make up the whole.

SOFT SYSTEMS METHODOLOGY THEPARTS

The gradual change in the way SSM as a wholehas been thought about, described above, hasbeen paralleled by more substantive changes tosome of the separate parts which make up thewhole. Many of these represent consciousattempts to improve and enrich such things as

model building, or the uses to which rich picturesare put; some have entailed dropping earlierways of doing things, for example the shift awayfrom using structure/process/climate as aframework for initial finding outabout a situation(STSP, pp. 163, 164, 166), or the deliberate drop-ping of the formal system model (STSP, Figure9; SSMA pp. 41, 42). But whether the changes tothe parts were additions or deletions, they werenever made by sitting at desks being academic.They have always been made as a result of experi-ences in using the approach in a complex world,

and they have played their part in changing per-ceptions of SSM as a whole. This section willreview the changes to the parts of SSM, the review being structured by the four activities whichunderpin the mature icon for SSM which is FigureA1 here.

Finding Out about a Problem Situation

Rich Picture BuildingMaking drawings to indicate the many elementsin any human situation is something which hascharacterized SSM from the start. Its rationale liesin the fact that the complexity of human affairsis always a complexity of multiple interactingrelationships; and pictures are a better mediumthan linear prose for expressing relationships.Pictures can be taken in as a whole and help to

S22 Peter Checkland



encourage holistic rather than reductionist think-ing about a situation.

Producing such graphics is very natural forsome people, very difficult for others. If it doesnot come naturally to you, it is a skill worth cul-tivating, but experience suggests that its for-

malization via use of ready-made fragments, suchas is advocated by Waring (1989) is not usually agood idea, except perhaps as a way of making astart. Users need to develop skill in making richpictures in ways they are comfortable with, wayswhich are as natural as possible for them as indi-viduals.

As far as use of such pictures is concerned, wehave found them invaluable as an item whichcan be tabled as the starting point of exploratorydiscussion with people in a problem situation. Indoing so we are saying, in effect This is how we

see this situation at present, its main stakeholdersand issues. Have we got it right from your per-spective? For example, when researching thesubtle relationship between a health authorityand one of its acute hospitals a few years ago(during the short-lived experiment with con-tracting in the NHS) we assembled from a greatmany semi-structured interviews a somewhatlarge and complicated picture though evenvery elaborate pictures are of course selections.(Bryant (1989) is correct to emphasize that Selec-tion of the key features of a situation is a crucial

skill in developing a picture (p. 260).) The picturein question became known as the briar patch,since that was the impression it gave at firstglance! Nevertheless it was found extremelyuseful, in a second round of interviews, to talkpeople through it and ask them for both theircomments about things we had got wrong, asthey saw it, and for their views on what werethe main issues concerning contracting (Duxbury1994). Their responses not only improved the pic-ture, and hence our holistic view of the situation,but also contributed to our understanding of thesocial and cultural features of the situation thesubject, in SSM of Analyses One, Two and Three(discussed below).

In recent work in the Health Service a new rolefor rich-picture-like illustrations has emerged. InDecember 1997 the Government White Paper TheNew NHS (HMSO 1997) described a new concept

8/6/2019 Paper MSB

13/48

of the NHS, which was to exhibit such featuresas: led from the front line of health care (primarycare by family doctors and other local services);founded on evidence-based medicine, withnational standards and guidelines; and sup-ported by modern information systems. Achiev-

ing this, according to the Minister of Healthresponsible for it, involved a demanding ten yearprogramme of development (p. 5). In 1998 thenecessary information strategy to support thisvision was published, the two documents beingcoherently linked (Burns 1998). Together, thesetwo publications represent the best conceptualthinking about the NHS for 20 years, though real-izing the vision will be an immense and difficulttask for medics who are usually not very inter-ested in thinking deeply about managing theirwork (as opposed to its professional execution)

and for an organization in which sophisticatedinformatics skills are scarce.

The White Paper and the information strategyare documents of 86 and 123 pages respectively;absorbing their message is not an easy task forpeople as busy as health care professionals andHealth Service managers. We have found itexceptionally useful, in work commissioned bythe centre of the NHS on the information systemimplications of the new concept for acute hospi-tals, to turn these excellent but overwhelmingdocuments into picture form. (The documents

themselves, being products of a Government ser-vice in which prose rules, contain only a handfulof rather unadventurous diagrams.) For theWhite Paper, Figure A3 gives the basic shape ofthe concept, while Figure A4 adds much moredetail to this simple picture. The informationstrategy, more complicated at a detailed levelthan the White Paper, was converted into a suiteof eight pictures covering its core processes andstructures, as well as the intended technical solu-tion: electronic patient records which graduallyevolve into each persons lifelong electronichealth record. These picture versions of longdocuments have been very useful in con-ceptualizingour work, and no NHS audience seesthem without asking for copies. This experiencedoes suggest that there is a useful role for picturesof this kind wherever there is detailed writtenexposition of plans and strategies at least until




thehappy times when such documents will them-selves use seriously the medium of pictures aswell as words.

Figures A3 and A4 can be seen as rep-resentations of combined structures and pro-cesses which enable the relation between the two

elements to be debated. But the use of structure,process and the relation between them as aformal framework for finding out in SSM,emphasized in the 1972 paper and in STSP (pp.163, 164, 166), has not survived. I believe per-sonally that I still use that framework mentally,without giving it much focused thought, but itsmore formal use, as described in 1972, has falleninto disuse. This seems to be because when youare faced with the energy and confusion whichgreet you whenever you enter any human prob-lem situation, that particular framework seems

highly abstract, a long way away from enablingyou to grapple with pressing issues. However, asalways with methodology, if it seems useful toyou, then use it!

Analyses One, Two and ThreeIn addition to rich picture building, other frame-works which help to make the grasp of the prob-lem situation as rich as possible are provided byAnalyses One, Two and Three (STSP, pp. 194198, 229233; SSMA, pp. 45 53). Analysis One isan examination of the intervention itself, and its

development was a direct result of our experienceof research for the late Kenneth Wardell, arespected mining engineering consultant in thatindustry. (He is the Mr Cliff of STSP (pp. 194198).) This analysis is now a deeply embeddedpart of the thinking. The rich pictures will drawattention to the (usually) many people or groupswho could be seen as stake-holders in any humansituation, and Analysis Ones list of possible,plausible problem owners, selected by theproblem solver, is always a main source of ideasfor relevant systems which might usefully bemodelled.

The freedom of the person or group inter-vening in a problem situation to answer the ques-tion: Who could I/we take the problem ownerto be? is important in achieving a grasp of thesituation which is as holistic as possible. Thusin work which helped a community centre in

8/6/2019 Paper MSB

14/48

Figure A3. The core concept of the NHS White Paper 1997 ( HA = health authority; HIP = health improvement plan;PCG = primary care group; PCT = primary care trust)

Liverpool to rethink its role in a run-down partof that city, it was relevant to consider LiverpoolSocial Services Department as one among manypossible problem owners, even though at the timethe relationship between the centre and thedepartment had not surfaced as an issue for any-one in the department. This kind of choice is whattrying to be as holistic as possible entails eventhough the whole will always remain an unre-achable grail. To adopt the counter-view sug-gested by Bryant (1989) that to be a problemowner you have to be aware of owning the prob-lem, would put a completely unnecessary con-straint on interventions founded on soft systemsthinking.

Analyses Two and Three, comprising a frame-work for the social and political analyses, arealso now thoroughly embedded in praxis. Somecommentators have suggested that they are less

S24 Peter Checkland



highly developed than some of the other partsof SSM, such as model building, but that is tomisunderstand them. The roles/norms/valuesframework and the ongoing analysis of com-modities which embody power are certainly sim-ply expressed. That is the point of them. You cankeep them in your head, and they can constantlyguide all of the thinking which goes on through-out an intervention. But though they are simplein expression they reflect one of the main under-lying conclusions from the whole 30 years of SSMdevelopment: that to make sense of it you haveto adopt the view argued in Chapter 8 of STSP(pp. 264285), namely that social reality is noreified entity out there, waiting to be inves-tigated. Rather, it is to be seen as continuouslysocially constructed and reconstructed by indi-viduals and groups (the latter never perfectlycoherent). This represents an intellectual stance

8/6/2019 Paper MSB

15/48

Figure A4. The White Paper concept of the New NHS 1997 (D of H = Department of Health; HAZ= health action zone;HI = health improvement; PH= public health)

defined by such features as: deriving from thework of Max Weber; articulated, for example, inthe sociology of Alfred Schutz; and underpinned by the philosophy of Edmund Husserl (Luck-mann 1978, pp. 7 13). In practical terms, theusable framework which underpins AnalysesTwo and Three was found in the autopoieticmodel teased out of the work of Vickers onappreciative systems (Checkland and Casar




1986). That will be discussed further towards theend of this chapter.

Analysis Three moves beyond the model ofan appreciative system but is compatible with it.(The appreciative system model describes a socialprocess; Analysis Three covers one of the maindeterminants of the outcomes of that process: thedistribution of power in the social situation.) Thisanalysis is avowedly practical, a highly sig-

8/6/2019 Paper MSB

16/48

nificant contribution to the development of SSMfrom the action research carried out by Stowell ina light engineering company and in an edu-cational publisher (Stowell 1989). He reviews theextensive social science literature on power, buthis main aim is not to add to that literature

which is strong on words, less interested inaction but to find practical ways of enablingopen discussions to take place on topics whichare usually taboo, or emerge only obliquely in thelocal organizational jokes. These are discussionsfocused on power, its manifestations and the pat-tern of its distribution.

Analysis Three is not based on an answer tothe question: What is power? It works with thefact that everyone who participates in the life ofany social grouping quickly acquires a sense ofwhat you have to do to influence people, to cause

things to happen, to stop possible courses ofaction, to significantly affect the actions the groupor members of it take. The metaphor of the com-modities which embody power is used toencourage discussion of these matters. Views can be elicited on what you have to possess to bepowerful in this group or this organization. Isit knowledge, a particular role, skills, charisma,experience, clubbability, impudence, commit-ment, insouciance . . . etc? Recent history of theorganization or group can be questioned and/orillustrated in these terms, all with the aim of find-

ing out as deeply as possible how this particularculture works, what change might be feasibleand what difficulties would attend that change.Stowell (1989) describes the use of the metaphorcommodity thus:

Commodity, then, is the proposed means of pro-viding organisational members involved in changewith a practical means of addressing power.Acknowledging, with Giddens, that speech andlanguage provide us with useful clues as to how toconceptualise processes of social production and re-production, what has been suggested within this

thesis is an idea by which the notion of power can be articulated in terms which are appropriate toa given organizational culture and which can beunderstood by those most affected (p. 246).

The aspiration of openness here is admirable, butdo not be surprised if Analysis Three has to becarried out with great sensitivity and tact. In

S26 Peter Checkland



many human situations there is not the con-fidence necessary for open discussion of issueshinging on power.

Before moving on from the finding outactivity, it is worth reiterating that finding outis never finished; it goes on throughout a study,

and must never be thought of as a preliminarytask which can be completed before modellingstarts.

Building Purposeful Activity Models

The Role of Modelling in SSMThe purposeful activity models used in SSM aredevices intellectual devices whose role is tohelp structure an exploration of the problem situ-ation being addressed. This is not an easy thought

to absorb for many people, since the normal con-notation of the word model, in a culture dren-ched in scientific and technological thinking, isthat it refers to some representation of some partof the world outside ourselves. This is the case,for example, for models as used within classicoperational research. If an operational researcherbuilds a model of a production facility, then thereis a need, before experimenting on the model toobtain results which can be used to improve thereal-world performance, to first show that themodel is a valid representation. This might be

done by showing that the model, fed with the lastsix months input, can generate something whichis close to the actual output produced over thatperiod. But models in SSM are not at all likethis. They do not purport to be representations ofanything in the real situation. They are accountsof concepts of pure purposeful activity, basedon declared world-views, which can be used tostimulate cogent questions in debate about thereal situation and the desirable changes to it. Theyare thus not models of. . . anything; they are mod-els relevant to debate about the situation perceivedas problematical. They are simply devices tostimulate, feed and structure that debate.

In the early stages of SSMs development,devices of only one type were built. Blinkered byour starting position in systems engineering, wetended only to make models whose (systems) boundaries corresponded to real-world organ-

8/6/2019 Paper MSB

17/48

izational boundaries. This self-imposed limi-tation derived, we can now see, from the systemsengineers view that the world consists of inter-acting systems. Thus, working in, say, a manu-facturing company with a conventionalfunctional organization structure, we would

make models of a production system, an R&Dsystem, or a marketing system. These would mapon to Production, R&D and Marketing depart-ments. But organizations have to carry out, cor-porately, many more purposeful activities thanthe handful which can be institutionalized in anorganization structure. For example, suppose themanufacturing company in question to be in thepetrochemicals business. Such companies, inorder to survive in a science-based internationalbusiness full of very smart competitors, have tobe technological innovators. In a systems study

carried out in just such a company (the study being concerned with improving relationsbetween R&D and other functions) it was foundvery useful to make a model based on the coreidea of innovating in that industry. That model (a. . . system . . . to innovate . . . in the petrochemicalindustry . . .) had a boundary which did notcoincide with the organizational boundaries ofthe (functionally defined) existing departments.Not surprisingly, many of the activities in thatmodel were actually taking place in the company:some in R&D, some in Production, some in Mar-

keting. Also, some of the activities in the modelwere missing in the real situation. The great valueof the model was that its boundary cut across theorganizational boundaries of the actual depart-ments. This was very helpful in stimulating dis-cussion and debate within the company, whenthe model was used to question the existing situ-ation.

Models which map existing organization struc-tures (such as a system to carry out R&D in thisexample) are thought of as primary task models;models like that of the innovation system areissue-based the notional issue here being thatsomehow or other this particular company hasto ensure that it has the ongoing capability toinnovate. This primary task/issue-based dis-tinction (SSMA, p. 31) has been found to be asource of confusion for many. This is probablybecause the distinction is not absolute. The petro-




chemical company, if its thinking had been a littledifferent, might well have brought together peo-ple with the appropriate skills and expertise tostaff an Innovation Department. Had they doneso the issue-based model here would then have been a primary task model. Pragmatically, to

make sure that the useful provocation providedby models whose boundaries cut across existingorganization boundaries is not neglected, the rulefrom experience is simple: make sure that you donot think only in terms of models which mapexisting structures. This will help ensure that themodelling fulfils its intended role in SSM: to liftthe thinking in the situation out of its normal,unnoticed, comfortable grooves.

Root Definitions, CATWOE and Multi-levelThinking

To build a model of a concept of a complex pur-poseful activity for use in a study using SSM,you require a clear definition of the purposefulactivity to be modelled. These definitional state-ments, SSMs root definitions, are constructedaround an expression of a purposeful activityas a transformation process T. Any purposefulactivity can be expressed in this form, in whichan entity, the input to the transforming process,is changed into a different state or form, sobecoming the output of the process. A bold sparsestatement of T could stand as a root definition,

for example a system to make electric toasters,but this would necessarily yield a very generalmodel. Greater specificity leads to more usefulmodels in most situations, so the T is elaboratedby defining the other elements which make upthe mnemonic CATWOE, as described in STSP(Chapters 6, 7, Appendix 1) and SSMA (Chapter2; illustrated passim).

These are not abstruse ideas; the skills requiredfor model building are not arcane: logical thoughtand an ability to see the wood and the trees; also,any model should be built in about 20 minutes.Nevertheless there are classic errors which recurtime and time again. The most common error,often found in the literature, is to confuse theinput which gets transformed into the outputwith the resources needed to carry out the trans-formation process. This conflates two differentideas: input and resources, which coherency

8/6/2019 Paper MSB

18/48

requires be kept separate. Also, when people real-ize that there is a formula (an abstract one) whichwill always produce a formulation which is atleast technically correct, namely: need for Xtransformed into need for X met, they seize onthis with glee. Unfortunately, they then often slip

into writing down such transformations as needfor food transformed into food. What a fortuneyou could make in the catering industry if youknew how to bring off that remarkable trans-formation! It is evidently not easy to rememberthat in a transformation what comes out is thesame as what went in, but in a changed (trans-formed) state.

In recent years experience has shown the valueof not only including CATWOE elements in defi-nitions but also casting root definitions in theform: do P by Q in order to contribute to achiev-

ing R, which answers the three questions: Whatto do (P), How to do it (Q) and Why do it (R)?[This formulation was, alas, initially given interms of XYZ rather than PQR (SSMA, p. 36).Using P, Q and R avoids the chance that Y may be confused with why?] The simplest possibledefinition is of a system to do P. Do P by Q isricher, answering the question: how? And alsoforcing the model builder to be sure that there isa plausible theory as to why Q is an appropriatemeans of doing P. For example: communicate (P)by letter writing (Q) is certainly plausible, but

would provoke examination of the reasons fordoing this communication (i.e. the R question) bythis chosen means. In this particular case, thequestion of required timing would have to bethought about. This could lead to examining, forexample, whether there was a case for replacingthe cultural resonance which goes with writing aletter to someone by the more brutal but quickere-mail.

The formal aim of this kind of thinking priorto building the model is to ensure that there isclarity of thought about the purposeful activitywhich is regarded as relevant to the particularproblem situation addressed. The idea of levels,or layers (or hierarchy, though that word tendsto carry connotations of authoritarianism whichare not relevant here) is absolutely fundamentalto systems thinking. Much human conversationis dogged by the confusion which follows from

S28 Peter Checkland



the common inability to organize thoughts andexpression consciously in several layers. Thus,the Chair of the Tennis Club Social Committeeopens a meeting by saying that the committeeneeds to decide whether or not to organize theclub fete this year, given the wet day last July and

the unfortunate arrival of a gang of unruly bikers!As you begin to think about this, sitting in com-mittee, you are surprised (but should not be) thatthe first member to speak says: My sister and Iwill do the cake stall as usual. Systems thinkersare adept at consciously separating whetherfrom what and how.

In selecting some hopefully relevant systemsto model, there are in principle always a numberof levels available, and it is necessary to decidefor each root definition which level will be thatof the system, the level at which will sit the T of

CATWOE. This makes the next lower level thesub-system level: that of the individual activitieswhich, linked together, meet the requirements ofthe definition. The next higher level is thendefined automatically as that of the wider sys-tem: the system of which the system defined byT is itself only a sub-system. In SSM this higherlevel is the level at which a decision to stop thesystem operating would be taken: it is the levelof the system owner, i.e. the O of CATWOE.Thus, this intellectual apparatus of T, CATWOE,root definition and PQR, ensures that the think-

ing being done covers at least three levels, thoseof system, sub-systems and wider system. It pre-vents the thinking from being too narrow, andstimulates thoughts about whether or not to buildother models. For example it might be decidedalso to model at the wider-system level, or toexpand some of the individual activities in theinitial model by making them sources of furtherroot definitions. (Figure 8.14 in SSMA, p. 231, forexample, shows a model in which activity 1 hasbeen expanded into four more detailed activities.Similar structuring is shown in Figure 5.6, p. 136,and Figure 7.8 of ISIS, p. 209 shows a simplemodel in which most of the activities have beenexpanded in this way.) Figure A5 summarizesthe importance of thinking consciously at severaldifferent levels, and also makes the point thatdifferent people might well make different judge-ments about which level to take as that of the

8/6/2019 Paper MSB

19/48

Figure A5. Systems thinking entails thinking in layers defined by an observer

system. What and how, system and sub-sys-tem are relative, not absolute concepts.

Measures of PerformanceIt is obvious from the form of SSM (as in FigureA1) that it would be possible to use the approachwithout creating systems models as the devicesused to shape the exploration of the situationaddressed. It would be possible to use insteadmodels based on theories of, say, aesthetics, psy-chology, religion, or even, if you were foolishenough to abandon rationality completely, astrol-ogy! We use systems models because our focus ison coping with the complexity in everyday life,and that complexity is always, at least in part, a




complexity of interacting and overlappingrelationships. Systems ideas are intrinsically con-cerned with relationships, and so systems modelsseem a sensible choice; and since they have beenfound, time after time, to lead to insights, theyhave not been abandoned.

Now, the core systems image is that of thewhole entity which can adapt and survive in achanging environment. So our models, to use sys-tems insights, need to be cast in a form which inprinciple allows the system to adapt in the lightof changing circumstances. That is why modelsof purposeful activity are built as sets of linkedactivities (an operational system to carry out theT of CATWOE) together with another set of

8/6/2019 Paper MSB

20/48

activities which monitor the operational systemand take control action if necessary. Since there isno such thing as completely neutral monitoring, itis necessary to define the criteria by which theperformance of the system as a whole will be judged. Hence the core structure of the moni-

toring and control sub-system is always the same:a monitor activity contingent upon definition ofthe criteria by which system performance will bejudged, and an activity rendered as take controlaction which is contingent upon the monitoring.This can of course be augmented if justified inparticular cases as in the model in STSP, p.291. The basic structure is seen in many of themodels in SSMA and ISIS.

For many years the concept of measures ofperformance was felt to be sufficient for use inmodels, but was then enriched by an analysis

which flows from the consideration that SSMsmodels are simply logical machines for carryingout a purposeful transformation process expre-ssed in a root definition. Measuring the per-formance of a logical machine can be expressedthrough an instrumental logic which focuses onthree issues: checking that the output is pro-duced; checking whether minimum resources areused to obtain it; and checking, at a higher level,that this transformation is worth doing becauseit makes a contribution to some higher level orlonger-term aim. This gives definitions of the

3Es which will be relevant for every model: thecriteria of efficacy (E1), efficiency (E2), and effec-tiveness (E3), first developed in 1987 (Forbes andCheckland 1987; Checkland etal. 1990; SSMA, pp.38, 39). This core set of criteria can be extendedin particular cases for example by adding E4for ethicality (is this transformation morally cor-rect?) and E5 for elegance (is this an aestheticallypleasing transformation?). Since it will not bepossible to name the criteria for effectivenesswithout thinking about the aspirations of thenotional system owner (O in CATWOE), thisanalysis is another contribution which preventsthe modellers thinking being restricted only toone level, that of the system itself.

Model BuildingGiven the preliminary thinking expressed in rootdefinition, CATWOE, the three Es and PQR,

S30 Peter Checkland



assembling an activity model ought not to betoo difficult: simply a matter of assembling theactivities required to obtain the input to T, trans-form it, and dispose of the output, ensuring thatactivities required by the other CATWOEelements are also covered; then link the activities

according to whether or not they are dependentupon other activities. And the task ought not tobe an elaborate one either, given the oft-provedvalue of the heuristic rule that the overall activityof the operational sub-system should be capturedin Millers (1968) magical number 722 indi-vidual activities (any of which can if necessarybe made the source of a more detailed model).Nevertheless some people manage to makemodel building a task fraught with difficulty.This is probably because there are in fact subtlefeatures of the process which are masked in the

simple account just given.These subtleties are illustrated by the fact that,

for example, the distribution manager of a manu-facturing company is probably not the personwho will find it very easy to build a model froma root definition of a system to distribute manu-factured products. The difficulty for such a per-son is to focus only on unpacking and displayingthe concept in the root definition; the tendency willbe to slip into describing the real-world arrange-ments for distributing products in his or her owncompany. Equally, inexperienced users, fresh-

from-school undergraduates especially, find allsuch models difficult to build because they knowso little about real-world arrangements. The factis and this is where the unobvious difficultiesof modelling lie it is not usually possible toconstruct a model exclusively on the basis of a rootdefinition, CATWOE, three Es and PQR; real-world knowledge does inform model building,but, crucially, must not dominate it. The craft skillis to build a model using a background of real-world knowledge without including features oftypical practice which are not justified by the rootdefinition, CATWOE, 3Es and PQR. As alwayswith craft skills, practice, practice, practice is thewatchword.

Because most practitioners initially feel theirway to a method of modelling comfortable forthem, it may be helpful to provide some tem-plates which derive purely from the logic of the

8/6/2019 Paper MSB

21/48

Figure A6. A logical procedure for building activity models

process and which may provide help for thosejust starting to use the process of SSM. Two suchtemplates are provided here; they are meant tobe abandoned as experience grows.

Figure A6 sets out a logical procedure for mod-elling purposeful activity systems in a series ofsteps; Figure A7 expresses the process in FigureA6 as a partial activity model. These are self-explanatory, though it may be remarked withreference to Figure A6 that although the stagescan be carried out on a computer screen, there isa good case, as long as you can manage it ingood visual style, for producing the final modelin hand-drawn form. The reason for this ispsychological, and is the same as that for drawingegg or cloud shapes rather than rectangular




boxes: it acknowledges the models role as prag-matic devices, not definitive once-and-for-allstatements. In Figure A7 the process formemphasizes the exercise of judgement duringmodelling. Iteration around activities 2, 3, 4 con-tinues until it is felt that the minimum but necess-ary cluster of activities has been assembled; thewider iterations around activities 1 to 6, andaround 1645 represent the checks that themodel is defensible in relation to the concept beingexpressed.

Once a model is constructed by such a process,the golden rule for reading a model some-thing which the many people unconsciouslystraitjacketed in linear thinking find difficult is always to start from the activities which are not

8/6/2019 Paper MSB

22/48

Figure A7. The process of modelling in SSM, embodying the logic of Figure A6.

dependent upon other activities but have others

dependent upon them, i.e. those which havearrows from them but none to them.

Finally, on modelling, a few remarks about theformal system model are in order (STSP, pp. 173177; SSMA, pp. 4142). As formulated in Figure9 of STSP (p. 175) this was useful when we wereacquiring a sense of what is meant to treat pur-poseful activity seriously as a systems concept.In SSMA it is said that it can now be cheerfullydropped (p. 92). Its language was the problem.Since it was built using concepts such as bound-ary, sub-systems, decision-takers, resources, etc.the unfortunate effect of its use was to reinforcethe wrong impression that the devices calledhuman activity systems are in some way to bethought of as would-be descriptions of real-worldpurposeful action. Since that is a main source ofmisunderstanding about SSM, and since what itoffers conceptually is captured in CATWOE, the

S32 Peter Checkland



3Es and PQR, it can indeed be cheerfully drop-

ped. The same argument speaks against thephrase human activity system, but that is prob-ably too deeply embedded to be prised out ofSSM and ditched. The best antidote to thesedangerous phrases is undoubtedly to encouragethe use of Arthur Koestlers neologism for theabstract concept of a whole, namely holon(Checkland 1988). That is what models in SSMare: holons for use in structuring debate.

Exploring the Situation and Taking Action

As human beings experience the unrolling fluxof happenings and thoughts which make up day-to-day life, both professional and private, theyare all the time likely to see parts of that flux assituations, and certain features of it as prob-lems, or issues. These concepts and this kind of

8/6/2019 Paper MSB

23/48

language of situations, issues, problems are very commonly used in everyday talk, but

they are subtle concepts, and we need to bewareof giving them a status they do not deserve. Wemust not reify them; they do not exist out there,beyond ourselves, as we can assume that beech

tree and that dog scratching itself do. This situ-ation, and this problem indicate dispositions tothink about (parts of) the flux in particular ways,and they are themselves generated by humanbeings; also, no two people will see them in exactlythe same way. If, for example, the senior man-agers of a company all agree in discussion thatthey have a problem due to the failure of a newproduct to build up sales following its launch, notwo of them will have precisely the same per-ception of this situation and/or this problem.What is more, some among those who agree

about the situation/problem may privately beseeking to ensure the failure of the new productin order that more resources can then come theirway! (Remember, we can never know for surewhat is going on inside the head of another per-son; and we cannot assume that their wordsnecessarily reveal it.)

These are bleak thoughts, but necessary onesif applied social science is to be pursued withadequate intellectual rigour. They mean that nei-ther problem situations nor problem types can beclassified and made the basis of pigeon holes into

which particular examples may be slotted, forone persons major issue or serious problemmay well be anothers unruffled normality. Boththe existence of a problem situation and itsinterpretation are human judgements, andhuman beings are not like-thinking automata.

A result of this is that the later stages of a studyusing SSM cannot be pinned down and as sharplydefined as the early stages, in which a situationcan be tentatively defined and explored, plausibleproblem owners named, relevant systemsselected and models built. The many uses of SSMdescribed in STSP, SSMA, SCMA and ISIS, aswell as the many accounts in the literature frompeople outside the Lancaster group, reveal thevariety which is possible. This ranges from quick,short, tactical studies to much longer onesoriented to strategy. Because of this, commentson the later parts of SSM are bound to be




generalizations from experiences which are verydiverse, those generalizations being themselvessubject to change as the flux of experiencerolls on. Nevertheless a few very broad gen-eralizations from a rich mass of experience canbe entertained.

The initial ways of using models, described inSTSP (pp. 177180) and SSMA (pp. 4244 andpassim in the cases described), are still the mostcommon way of initiating the comparison stageof SSM, in which well-structured debate aboutpossible change is sought. Most common, at leastasan initiator of debate, is the completion of chartsin which questions derived from the modelsbrought to the debate are answered from per-ceptions of current reality on the part of peoplein the situation addressed. But do not expect thedebate to be tidy or predictable; be deft, light on

your feet, ready to follow where the debate leads,unready to follow any dogmatic line.

Looking back over experiences in the lastdecade, an emerging pattern can be discerned inwhich there are two common foci of the laterstages of SSM, during which the driving principleis to bring the study to some sort of conclusion.The first of these is the original one: SSM as anaction-oriented approach, seeking the accom-modations which enable action to improve to betaken. This is exemplified in the work in IndexPublishing and Printing Company, described in

STSP, pp. 183189. Here action was taken toimprove the working relationships between pub-lishers and printers, who represent two very dif-ferent cultures. A new process to deal with issuessurrounding the decision where to print wasestablished, and a new unit to carry out the workwas set up. The second focus, very prevalent inthegreat complexity which characterizes the pub-lic sector, is on SSM as a sense-making approach.This is exemplified in recent work in the NHS,and is discussed below.

In the first (action-oriented) case the changesought can usefully be thought about in terms ofstructural change, process change and changes ofoutlook or attitude. Normally in human affairsany explicitly organized change will entail allthree, and the relationship and interactionsbetween the three need careful thought. Of coursethe easy option to take in the public sector for

8/6/2019 Paper MSB

24/48

Figure A8. Thinking about desirable and feasible change

Government or, in other organizations, for seniormanagers is to impose structural change; andthat is often done without serious attention to theother two dimensions: process and attitude. Thelong series of changes imposed by the UKGovernment on the Health Service, for example,give a good illustration of imposed structuralchange with relatively little attention to the pro-cess and attitudinal change also required (Ham

1992; Rivett 1998; Webster 1998). [It has been sig-nificant, recently, that an experienced com-mentator on the Health Service, Chris Ham, hasdetected that the obsession with structuralchange that has dominated health policy in recentyears has given way to a focus on how staff andservices can be developed. . . . (Ham 1996). Thatis a much-needed change.]

In general, thinking about desirable and feas-ible change can initially be structured in the wayshown in Figure A8. A most important feature ofthis is the need in human affairs to think not onlyabout the substance of the intended change itself but also about the additional things you normallyhave to do in human situations to enable changeto occur. (In introducing a clinical informationsystem in a big acute hospital, for example, aproject described in ISIS (pp. 192198), its insti-gator Peter Wood, Chairman of the District

S34 Peter Checkland



Health Authority had already spent several yearspreparing the ground with hospital consultantsinevitably suspicious that such systems couldlead to greater control of their clinical activity byhospital managers.)

The second broad category of use to whichSSM-style activity models can be put is to usethem to make sense of complex situations(though that sense making may of course also

lead on to action being taken). It is significant thatthis category of use has grown markedly in thelast decade of SSM development, as conceptssuch as organization, function, professionand career have all become more fluid.

Sense-making use of models is well illustratedin recent research in the NHS. The work has beendescribed in some detail in Checkland (1997) andin ISIS, pp. 165172 and will only be sketched inhere. Setting out to research the new contract-based relationship between purchasers of healthcare for a given population and providers of thatcare (such as acute hospitals, for example) aresearch team from Lancaster University Man-agement School, using SSM, first built an activitymodel of the contracting process (Figure 6.2 inISIS). The concept expressed in this model didnot rely at all upon observation of the NHS. Itreflected simply the interests of the multi-disci-

8/6/2019 Paper MSB

25/48

plinary research team: information support,organization change, etc. This model was used asa source of structure for open-ended interviewswith more than 60 NHS professionals. This pro-duced a daunting mass of interview material.This was analysed by extracting from it the nouns

and verbs used by NHS professionals in describing the contracting process and their expec-tations of it during its first year. These nounsand verbs were fashioned into the elements of anactivity model, and these elements were com-bined to make an activity model relevant to thecontracting process as it was initially being inter-preted by both purchasers and providers ofhealth care. (This backwards modelling notbased on a root definition but teased out of theinterview material represented an innovationwithin SSM. See ISIS, pp. 165 172.) The dif-

ference between the first model (based on theresearchers world-views) and the second one(based on the world-views of NHS professionals)defined the learning achieved in this first phaseof the research. This led to 10 pieces of actionresearch in the NHS, and eventually to anothersense-making model which helped to unpack andilluminate the purchaser-provider relationship.

This second sense-making model sought toflesh out coherently the complex interactions between a particular purchaser (a health auth-ority) and a particular provider (an acute hospi-

tal), interactions to which we had had access overa two-year period. In order to find our way to amodel which would richly express all we hadobserved, 47 previous models relevant to NHSpurchasing/providing were first examined (Dux-bury 1994). (These came from earlier SSM-basedwork in the NHS.) This established what lan-guage had been found relevant to describing pur-chasing or providing. This language, togetherwith the recorded observations of what had hap-pened in the present experience between the col-laborating hospital and their local healthauthority, yielded an activity model which makessense of all that had been observed. The deri-vation of the model [Figure 8.4 in the book aboutresearch in the NHS edited by Flynn and Wil-liams (1997)] was a subtle process. The guide tothat process was the question: What activitymodel could generate all the happenings




observed over the two-year period? Its role wasto provide a coherent frame for the 10 furtherpieces of action research at NHS sites.

This completes the necessarily tentative dis-cussion of the variegated later stages of SSM-based studies or projects. Enoug

paper msb

Documents