The systems thinking paradigm
In contrast to classical system theory
Copyright 2017 Graham Berrisford. One of about 300 papers at http://avancier.website. Last updated 23/02/2018 00:28
The role of enterprise architects is to observe baseline systems, envisage target systems, and describe both.
They apply classical system theory – a relatively scientific approach to describing systems in a way that can be tested and implemented.
And you might assume they draw much from “systems thinking”; but this is not the case, for reasons to be explained.
If you are new to systems thinking, then perhaps best read Systems Thinkers before reading on.
Systems thinkers discussed in that article include. Ackoff, Ashby, Bateson, Beer, Bertalanffy, Boulding,
Carnap, Checkland, Darwin, Durkheim, Forrester, Habernas, Henderson, Luhmann, Marx and Engels,
Meadows, Pareto, Parsons, Schrödinger, Simon, Smith, Spencer, Tarde, Weber, Weiner, and Wittgenstein.
The system theory applied enterprise architecture frameworks is a relatively scientific approach to describing systems.
It is done to describe operational systems in a way that can be tested and implemented.
Much social system thinking is different.
A social entity is a group of inter-communicating actors.
The actors interact by exchanging information (descriptions, directions, decisions and requests for them).
Primitive social entities (e.g. bees in a beehive) can be regarded as deterministic systems.
The actors play pre-defined roles and communicate pre-defined types of information.
Classical system theory can be applied to such an animate social entity.
Classical system theory, when applied to a human society, is sometimes called “structuralism”.
Structuralists analyse aspects of human society in terms of relationships between elements in a conceptual or theoretical system.
Science presumes we can test a real-world or empirical system against a theoretical system.
However, many systems thinkers are concerned with societies that are less systematic.
Their focus is on social groups in which the actors are self-aware and regarded as having free will.
If need be, human actors can behave in a deterministic way; and communicate using pre-defined types of information.
And a deterministic system can accommodate out-of-ordinary cases as “exception paths”.
However, humans also create new information, communicate it to others, and respond in ad hoc ways.
The question arises: if little or no behaviour of the group can be tested against a description, in what useful sense is it a system?
David Seidl (2001) said a social system theorist has to choose the basic elements of a system.
“The sociological tradition suggests two alternatives: either persons or actions.”
The trouble is, taking the former view tends to eviscerate the concept of a system.
E.g. A street gang is a set of actors who perform activities as they choose.
Such an incoherent group - doing things that are not regular or repeated - is not well-called a system.
Russell Ackoff considered such entities to be out of scope of his systems thinking.
A group of people , who communicate with each other, may agree to cooperate in some way, for some time.
The trouble is, group members may come and go, they may change their mind about their aims and/or their activities.
Some members may obstruct or even undermine what other group members want to do.
Surely a social system is more than a group of people doing things they choose?
And to call a social entity a system - in a useful sense - is to imply the entity exhibits some orderly behavior?
Else, if its activities or aims are in continual flux, then they are not describable and testable.
What does it mean to call a church, a government or IBM “a system”?
Does it mean merely that it contains things that are interrelated in some way or another, which may be said of the entire universe?
Does it mean more specifically that it is a social entity in which people receive directions from above and communicate with each other?
System thinkers speak of social and business entities as complex adaptive systems.
Obviously, IBM is an entity in which countless systems may be found.
And at any one time, some of those systems will be changing.
So, one might glibly call IBM a complex adaptive system of systems.
How to measure IBM’s complexity, or its adaptations?
In what meaningful, testable or useful sense is IBM a system?
Much has been written on what are known as complex systems.
Where complex usually implies some or all of large, diverse, continually changing.
IBM is clearly a complicated entity, to what extent can it usefully be described as a testable system?
Imagine trying to describe all the actors and activities in a city.
It contains millions of actors who perform activities - mostly as they choose.
It does contain countless systems; a public transport system, a police authority, perhaps a shoe factory.
These activity systems are disparate systems, that are loosely or not at all coupled.
It is impossible to join them up in a holistic description, and test the city as a coherent whole.
And without reference to one description, to call the whole city one system has no useful meaning.
The entity is so vast and complex that it is beyond any one holistic description.
Being self-aware and self-willed, human actors make the behaviour of a social entity less predictable.
Is that making the social entity more complex? Or merely less systematic, less describable as a system?
The complexity of the entity can only be measured with reference to what we can describe of it - which may be very little
And the complexity of an empirical system can only be measured with respect to the roles and rules of a theoretical system.
In short, a system is as complex as the roles and rules we describe or have in mind - no more, no less.
"Adaptive systems" and “self-organizing dynamics”
When early system theorists spoke of adaptation, they meant system state change.
When system thinkers speak of adaptation, they mean system generation change.
So an adaptive system is one that changes itself.
And some system thinkers presume a system has what may be called self-organizing dynamics.
This implies that a system can mutate continually, which is to undermine the very concept of a system
"Systems of systems"
von Bertalanffy wrote of a concept he called organicism.
He meant not only that larger systems can be decomposed into smaller ones, and vice-versa.
But also that a system depends on integration and cooperation between its subsystems.
You might think also of a designed, electro-mechanical entity like a radio or a car.
E.g. an automobile may be decomposed into chassis, wheels, steering, and engine, block, cylinder, valve.
However, a motor car does not depend on interaction and cooperation between its engine and its radio.
And some now use the term of system of systems much more loosely, to mean merely a container of systems.
IBM has a central management body, and it certainly contains many describable systems.
However those systems may duplicate or compete with each other.
We don’t know how well they cooperate; we don’t even know they are all necessary to the whole.
Today’s business probably differs from the institution Ackoff, Beer, Checkland etc. had in mind.
But the main point here applied then and now.
To say "the enterprise is a system" is meaningless unless you have a specific system description in mind.
Because the enterprise is as many different systems as you can describe and test.
We might say IBM is a complex adaptive system, but only in a circular sense.
Because the term “complex adaptive system” is merely a label for a large and complicated entity like IBM.
We might say "enterprise architecture regards the enterprise as a system, or a system of systems."
But that is vision rather than a reality, since large businesses suffer the diseconomies of scale.
The countless identifiable systems in a business may be disparate and even antagonistic.
To change and improve those systems does mean applying classical system theory and cybernetics.
The term “system” has a clear and useful meaning in classical system theory and cybernetics.
That meaning may be dismissed by “systems thinkers” as “engineering”.
Yet today’s business relies more than ever on the “engineering” of digitised business systems.
To be sure, culture of an enterprise has a huge impact on what systems can be changed or introduced at an operational level.
Culture also has a huge impact on the ability of enterprise architects to propose change in the first place.
Enterprise architects must be sensitive to cultures at both operational and strategic levels, and are influenced by them.
That does not mean that enterprise architects are employed to propose or design cultural change.
The social impacts of changes to activity systems are usually addressed in parallel, by a business change team.
In much systems thinking discussion, the term "system" is a noise word.
It adds no useful meaning to the discussion of a named institution or society.
Systems thinkers use terms such as autopoiesis, complexity, chaos, emergent properties, entropy, holistic, and value stream.
However, using scientific-sounding words does not make an essay scientifically valid or useful.
And the question remains: to what extent is a society meaningfully called a system?
Critical realism is philosophical approach associated with Roy Bhaskar (1944–2014).
It combines a general philosophy of science (transcendental realism) with a philosophy of social science (critical naturalism).
It describes an interface between the natural and social worlds.
Note that all philosophical positions have interpretations and variations.
And this section is written with reference to “The positive and the negative” (see references).
Like philosophers, socio-cultural essayists like comparing, contrasting and criticising past approaches.
“Bhaskar (1998) criticizes structuralist positions for determinism, and individualist positions for failing to account for the social context,
with the consequence being that for him the task is to link structure and agency.” (Cruickshank)
Obviously, the human actors in a society have some agency to shape its aims, activities and outcomes.
The question is to what extent that society is meaningfully called one system.
Inventing classifications and neologisms
“To do this [Bhaskar] turns to his natural scientific ontology to argue that social reality is a stratified open system”. (Cruickshank)
I presume one stratum is a social structure (whatever “structure” means) and the other is human actors who have a life and aims outside that structure.
The term “stratified open system” seem unrelated to the meaning of “open system” in classic system theory.
Borrowing words from classifical system theory
“It is stratified because social structures are held to be emergent properties that exist in interaction with agents who are conditioned but not determined by structures.”
The term emergent property seems unrelated to emergent properties in the classic system theory (bicycle and rider) sense.
If capitalism and religions are emergent properties, they seem to be system descriptions.
Does Bhaskar mean a whole system, one that has evolved and stabilised inside a society or social entity?
Focusing on people as human rather than role players
“Archer (1995) argues that culture is an emergent property.”
A property of what? How is the existence of a culture detected, measured or tested?
Presenting classifications as though they are scientific theories.
“[Archer’s critical realism discusses] the interplay of
· structural emergent properties (SEP) such as capitalism;
· cultural emergent properties (CEPs) such as religion; and
· people’s emergent properties (PEPs), such as religious groups, political groups or trades’ union groups.”
Again this is to use the term emergent properties differently from in classical system theory.
How does this three-way classification help?
Surely the classes overlap, since people defined capitalism and all religions (bar your own if you insist)
And one person can influence SEP and CEP – depending on how big those systems are – and what their own role is.
Grand-sounding statements to cover a weakness.
“Given that the outcome of this interaction is contingent and given that SEPs and PEPs can interact in contingent ways, the social system has to be an open system characterized by change at the level of observable events.”
Here, contingent seems to mean that chance plays such a large role in what happens we can’t use our theory to predict anything.
“The role of theory in social science therefore is to interpret empirical phenomenon in terms of how observed events are the contingent outcomes of the interaction of unobservable processes.”
Meaning, we cannot see or define the processes of the system, or predict their outcomes.
So, all we can do is explain history in terms of a classification or theory we made up? Like Marxists do?
In short, it is difficult in sociology to conduct experiments to test a theory.
Consequently, much systems thinking is the stuff of a socio-cultural essayist rather than a scientist.
Systems thinkers have devised an astonishing profusion of classifications.
The classifications are sometimes used to explain things in retrospect.
It is often unclear how the classification helps you do or decide something in future
Sociologists have classified people into types.
E.g. the categories defined by Belbin, Myers-Briggs, and used in DISC behavior assessment.
Such schemes are open to abuse by managers who pigeon hole people and then limit their opportunities.
Social systems thinkers have also classified systems in many ways.
Typically, a clock is placed near the bottom of a scale, and a human society is placed near the top (Boulding).
Or a human being is at the bottom, and each higher level is a successively wider social entity or context
Most system classifications are questionable and/or incoherent for one or more of these reasons.
· there is confusion between a type/subtype hierarchy and a composite/part hierarchy.
· the categories overlap
· the categories are incomparable, not of the same nature.
· the categories are placed on a questionable scale, or ranked in a questionable hierarchy.
This example was suggested to me by Rick Anderson.
Bronfenbremner’s five "systems" are a questionable way to classify things that influence a human's behavior.
Do the five categories represent steps on a scale?
No, categories 1, 2 and 3 suggest a widening of the social context; whereas categories 4 and 5 are of a different nature.
The following definitions and examples are edited from <https://explorable.com/ecological-systems-theory>.
1 "Micro system
the setting in which you have direct social interactions.
family, friends, classmates, teachers, neighbors
and others you contact directly."
Is that one micro system or several?
In what sense do the people who contact you form a system?
2 "Meso system
the relationships between microenvironments in your life.
your family experience related to your school experience."
There are relationships both within and between micro systems.
How does 2 differ from 1?
3 "Exo system
the setting that links the context you have no active role in
to the context you participate in."
Surely the context you have no role in is the rest of universe?
4 "Macro system
the actual culture of an individual.
being born to a poor family makes a person work harder every day."
Culture is a vague concept; the example suggests a social prejudice.
5 "Chron system
the transitions and shifts in one's life.
divorce may affect not only your marital relationship
but also your children's behavior."
Surely such events are continuous in all the systems above?
What does Bronfenbremner’s hypothesis predict?
Does it predict an influential person or event fits in only one category? No, a person or event may appear in more than category.
Does it predict each category has a different influence on us? No, there seems no correspondence between categories and influences.
Does it explain things - influences on us, or interactions within or between categories? No, it only classifies them.
Are the five categories well-called "systems"?
No, you could replace "system" by "context" or "environment", with no effect on the meaning of the categories.
There is no testable system description, no definition of roles and rules.
There is only the notion of people making contact with each other, and unpredictable events happening.
Yes, the classification does predict that people, their habits and events in the world (both near and far) have an influence on our lives.
But this is more a statement of the obvious than a theory of scientific value.
Again, this is the stuff of a socio-cultural essayist rather than a scientist.
This example was suggested to me by Rick Anderson.
Consider the discrete entity known as a Sound System.
At the heart of it is a set of equipment for the reproduction and amplification of sound
When people switch it on and listen to it, it becomes an activity system.
The system boundary can be expanded to embrace the roles of equipment controller and sound receiver.
The regular behaviors of actors playing these roles in the system can be defined, predicted and tested.
However, in Charles Handy’s discussion of the topic, the Sound System took on a cultural aspect.
In Jamaican dance halls, the Notting Hill carnival and elsewhere, the "system" evolved to include the role of the DJ as a superstar with social influence.
This become a worldwide phenomenon, and businesses grew up around it.
Thus, Handy buried the original Sound System inside an entirely different, large and diffuse, social entity
Many well-respected social systems thinkers (e.g. Ackoff and Meadows) make the kind of leap Handy did.
Within a few sentences, they leap from speaking of a discrete, describable and testable system to making assertions about a large and diffuse social entity, cultural phenomenon or ecology.
In doing this, they eviscerate the meaning of "system" to the point it conveys no particular meaning beyond "the entity contains things that are interrelated in some way or another".
Again, this is the stuff of a socio-cultural essayist rather than a scientist.
In some ways, enterprise architects have inherited the mantle of what used to be called the "operational research" department.
They are concerned with the efficiency and effectiveness of business operations, and digitization of them.
They don't presume to define the aims or strategy of the business; but they must of course strive to support them.
Natural entities (a hurricane, the solar system, a bacterium or a beehive) evolve by chance with no aims.
Meaning, the system's non-organic actors have no aims of their own.
Here, the term "aim" refers to a result or outcome that is envisaged and intended by an organic actor.
You might say all organic actors have aims predetermined in their DNA.
However, at some point in evolution, some biological actors became recognisably what Ackoff called “purposeful”.
This means actors are self-aware and self-willed enough to
· envisage the future,
· predict the outcome of activities that might be performed, then
· choose a course of action with outcomes in mind.
So here, an aim is an outcome envisaged by a purposeful actor, or agreed by a group of them, of activities to be performed.
And given a social entity of interest, two kinds of aim can be identified.
External sponsors and other stakeholders may give aims to an entity.
In designing mechanical and social entities, designers trade off between conflicting directors' aims.
In a beehive, the aims of actors playing roles have evolved to match those of the social entity.
In a business, the human actors are purposeful, their own aims may conflict with directors' aims.
Management scientists often speak of a corporation as a system.
And often, they are thinking of its management structure and directors' aims.
Directors declare aims; managers cascade them to people working in the “operational system”.
However, directors' aims may be incomplete, inconsistent or ineffective.
What workers do may be more determined by other things – local procedures, actors' aims and market forces.
A hierarchical management structure and aim cascade is only one way to describe how enterprises work in reality.
Yes, we can encapsulate a social entity by giving it some aims.
Then (without ever describing its internal actors and activities) test that the social entity meets those aims.
But different people will ascribe different aims to the same entity.
Meaning that, even if we consider only its aims, the entity is different systems to different people.
And to describe external aims without internal activities is to eviscerate the concept of a system.
For more, see papers on "SYSTEMS THINKING" on the "System Theory" page at avancier.website.
(Assessing critical realism and social constructionism as post-positivist approaches to empirical research in the social sciences.)
Working Papers Paper 42, August 2011, Justin Cruickshank
Published by the International Migration Institute (IMI), University of Oxford
All free-to-read materials on the http://avancier,web site are paid for out of income from Avancier’s training courses and methods licences.
If you find them helpful, please spread the word and link to the site in whichever social media you use.