Ackoff’s system classification (1996-2003)

Copyright 2014-17 Graham Berrisford.

One of about 300 papers at Last updated 15/01/2017 22:00


Russell L Ackoff (1919-2009) was an American organizational theorist, operations researcher, systems thinker and management scientist.

He was a well-known systems thinker, respected for a large body of work focused on human society.

This earlier paper analyses 30 ideas Ackoff felt to be definitive of systems in 1971, including an early system classification.

The paper below analyses his system classification as it reappeared about 30 years later.

Most text in blue below is quote from On The Mismatch Between Systems And Their Models” - Russell L. Ackoff and Jamshid Gharajedaghi.


The four system classes. 1

First thoughts. 3

Deterministic (Mechanistic) Systems - in detail 5

Animate Systems - in detail 7

Social Systems - in detail 8

Ecological Systems - in detail 9

On Ackoff’s conclusions. 10


The four system classes

“All organizations are social systems”

Ackoff was not a fan of general system theory.

He deprecated the mechanistic, biological and animalistic views of systems taken by other system theorists.

His primary interest was the organisations of interest to “management science”.

So his system classification is bent to this end.


“Most contemporary social systems are failing.”

Ackoff sometimes seemed to be presenting a political manifesto.

Surely some businesses succeed, some fail, and most succeed a bit and fail a bit?

Also, some government agencies succeed, some fail, and most succeed a bit and fail a bit?


A system is a whole that is defined by its function(s) in one or more containing systems.

OK: A system can be defined by its interaction with entities and activities in a wider environment.


Every system contains at least two essential parts and these must satisfy three conditions defined in the paper.

OK: A system contains two or more interrelated parts. The “three conditions” are questionable, but I’ll leave those for another time.


Ackoff proposed “choice” to be the characteristic that distinguishes his four classes of systems.

His system classification is almost certainly different from your first intuitive understanding of it!

Type of System Model





1 Deterministic / mechanistic

Clock, Tree, Bee

No choice

No choice

Not just machines! This includes plants and “lower” animals

2 Animate


No choice


Not all animals! This excludes “lower” animals.

It includes only “higher” animals that Ackoff considers to exercise free will

3 Social

Church, Corporation



Not all social groups! This excludes “lower” animal groups and informal groups.

It is primarily if not only formal organisations

4 Ecological



No Choice

Not an external environment!


It is easy to misread Ackoff’s distinctions.

How does he (or you) distinguish between animals that make choices freely and those that do not?

He presumes machines, plants and “lower” animals choose between actions deterministically - based on current state information or a memory of past events.

But surely you also do that some of the time? (Boulding assumed you do that all of the time.)

How do social systems make choices independently of choices made by actors they contain?

How do ecologies make choices independently of choices made by actors they contain?


It is important to note that there is a hierarchical relationship among the four categories shown in Table 1: each type has or uses parts that are systems of the type above it in Table 1.

Really? Isn’t the human body (4) the ecological system for the bacteria in its gut which are (1) deterministic systems – with no intermediate social (3) or animate system (2)?


There are many different ways of classifying systems. Different classifications have different uses.

OK. The authors don’t intend us to use this classification for every purpose; alternatives are possible.


Since we believe choice (and purposeful behavior which derives from it) is at the heart of human and social development,

our classification depends on whether the essential parts of a system or the whole can display choice and, therefore, have purposes.


We have found the classification we present here to be the most useful for understanding the poor performances of social systems.

Who says social systems perform poorly? By what measure? 

Are the authors implicitly adopting a political position with respect to government institutions?

There are many ways of understanding why a public sector enterprise’s performance can be disappointing.

Powerful explanations can be found in Parkinson’s law, the Peter Principle, and the challenges of recruiting and motivating employees.

First thoughts

Ackoff’s 4-way classification has questionable theoretical foundations.

The paper isn’t academically rigorous.

There is a questionable version of business history, contentious points, tenuous connections and conclusions.

The paper is more rationalist than empiricist; how to test his assertions?

Despite the authors’ attack on organismic or animalist approaches, there is an anthropomorphic tendency.


Is there a polemical aim?

A desire to treat an organisation as deserving of the same moral and legal status as a human being?

A desire to place management science at the same level of “science” as biology, physics or maths?


The classification is often misread as corresponding to mental models other people hold.


Management consultant or Enterprise architect

Mechanistic system

Computing or other Technology

Animate system

People or Human actors

Social system.

Organisation or Enterprise

e.g. a kind of Social System that works to goals given to it by one or more directors.

e.g. a kind of Organisation whose directors have control of a shared budget.




Other papers classify system types in different ways, for example:

Actor or system type



Ackoff class?



Solar System, Tidal System

None (unless you count maintaining state).



Clock, Car, Computer

Goals of sponsors, designers and users.




Tree, Bee, Human

Produce progeny, and all to that end

Deterministic? Animate

Social group

Tennis club, Church, IBM

Goals of sponsors, designers and users

+ goals members bring to their roles

Deterministic? Society


The table below maps 4 of Avancier’s system scales to Ackoff’s 4 system classes.

Note that deterministic means predictable in theory, predictable means predictable in practice.

Ackoff’s system class

Avancier’s system scales






















This mapping is useful when we come to look at the Cybefin framework

Deterministic (Mechanistic) Systems - in detail

Ackoff Commentary


All mechanisms are deterministic systems.

Is the implication that all deterministic systems are mechanistic systems? That is questionable, but seems to be assumed later.


Their behavior and the behavior of their parts are determined by their internal structure and their environments.

·         A clock, for example, continues to run only if rewound or has its battery replaced, or is fed electricity, or is kept moving for much of the time.

·         An automobile’s behavior is determined by its users and the behavior of its essential parts.

But the behavior of its essential parts is determined not by their choices, because they cannot make any, but on their physical state or that of their environment.

The authors define “choice” as excluding mechanistic choices made by machines and non-self-aware biological entities like plants or ants.

They do not explain how non-mechanistic choice differs from a mechanistic choice.


Although deterministic systems have no purposes of their own, they normally serve the purpose(s) of one or more complex systems; for example, their creators, controllers, or users.

Clearly, an automobile has the purposes conferred on it by its designers, users and other observers.

But what about a rose bush? A honey bee? Or a slug? Does it have purposes?


The only completely closed system is the universe.

All systems modelled using Forrester’s System Dynamics are closed systems.


Computers…appear to make choices, but it is not so.

Their behavior is completely determined by their structure together with the information and program put into them by external sources.

Hmm  computers and people both choose how to respond to an input with respect to their internal state or memory of past events. What is the difference?


If we know these, in principle if not in practice, we can determine what a computer will do in any specified situation.

Hmm… the processes humans use to make decisions might be equally predictable in principle

The difficulty is that we do not know their internal state or the rules they apply to inputs, so cannot predict their results in practice.


The more functions a deterministic system performs, the more complex it is.

This is misleading. A system with many simple functions can be simple. A system with a few complex functions can be complex.


The measure of complexity of a system is the number of variables and their interactions required to explain the behavior of the system.

This is misleading. The term variable here begs definition, but counting variables is not a good measure. Ashby recommends counting variable values, which is also a poor measure.


Animate Systems - in detail

Ackoff Commentary


Not all biological systems are animate. Plants, for example, do not make choices but respond to external and internal conditions deterministically.

Hmm… Would the authors consider a slug, ant, flea, bacterium or virus to be animate?

Surely not. So is the defining characteristic of an animate entity really its self-awareness?


Animate organisms such as humans can make choices but their parts cannot. (Unless otherwise indicated, we will use the term ‘organism’ to refer only to animated systems.)

The definition of animate organisms seems so anthropomorphic the category might as well be called human.

Does a human system include choosing clothing to maintain its body temperature – which seems a mechanistic homeostatic action?

Does a human system include the bacteria in its gut, without which it could not survive?

Does a human system include the spreadsheet it created yesterday to help it make choices today?


Plants react to external conditions in such a way as to make their survival possible. However, their reactions are determined, not matters of choice. Reactions are determined; responses involve choice.

Hmm… It is far from clear that choice and determinism are incompatible.

Animate systems may be unpredictable, they may be aware of choices they make; that is not the same as saying their responses are not deterministic.


The organs of a human body like machines are mechanisms.

The behavior of one's heart and lungs is determined in the same way that the behavior of a motor of an automobile is.

Their function is to make life possible, but they do not make choices.

Hmm… body organs and people both choose how to respond to an input with respect to their internal state or memory of past events.

What is the difference? Where do the authors explain that?


An animal’s purpose to survive and reproduce is not something it decides, creates or owns.

This fundamental purpose is conferred on the animal by God, or by evolution (as encoded in it specification – its DNA).

Hmm…  Consider the purposes to eat, make love, make friends, retain friends through social interaction, conform to social norms, earn a living, make a home.

Aren’t these encoded, if indirectly, in DNA? Can’t other aims and actions be seen as subordinate to them, or side effects of them?


Social Systems - in detail

Ackoff Commentary


All organizations are social systems.

Really? A car is not an organisation? A plant is not an organisation? A body is not an organisation?

They surely cannot mean all social organisations are social systems, since they exclude “lower” animals.

They probably mean all human social organisations are social systems – which sound tautologous.


They are significantly different from organisms. They contain parts that are animate systems that display choice.

Yes, groups are different from individuals

But the parts of a social organisation are not animate entities; they are roles played by those animate entities.

And an animate entity can play role in several social organisations

If the eleven actors in football team form a choral group – that is two social organisations?


(They may also use or have mechanical systems as parts; for example, production equipment in a manufacturing companies.)

Does Ackoff think only of human social systems? What about chimpanzee troupes? ant colonies? schools of fish?


Organizations display choice.

Do they? Would you say an investment bank’s trading division chooses which stocks to buy and sell?

The choices used to be made by animate human actors.

They are now mostly made by mechanistic computer actors (proxies for humans).

In what sense does the organisation display choice?

How does any social system make a choice independently of choices made by one or more of its animate or mechanistic parts?


We commonly refer to choices made by corporations, government, and other types of social systems; for example, schools, hospitals, and government agencies.

Some do refer to choices made by social systems, but some find that unnatural.

We know the choices are in fact made by individual actors who play roles in the system.


Clearly, an automobile has the purposes conferred on it by its observers - designers, users and others.

A social system may also have purposes conferred on it by its observers - designers, users, members or others.

A human actor always has personal goals or purposes outside any social system they act in.

Geoff Elliot tells me that in the socio-cultural perspective, only people can be “purposeful”.

But a human actor always has personal goals or purposes outside any social system they act in.

They may have purposes that are private, are in conflict, are in flux, are unconscious (they don’t recognise themselves).

They may act to make a social system inefficient, or even to sabotage it.

So, which of all those purposes might Ackoff consider as making the social system purposeful?


Note also that social systems are usually parts of larger social systems and that these usually contain other social systems.

For example a corporation is part of an economic system that contains other corporations.

A university is part of an educational system that contains many other schools.

What does “contain” mean? Physical enclosure? Logical dependency?

Surely a corporation can be part of several distinct economic systems? And move from one to another?


We are not aware of effort to model organisms or mechanical systems as social systems.

Some suggest biological evolution assembled organisms from what were formerly independent parts, they view our body as a society of organs and cells.


However, social systems have often been, and still are, modeled as organisms (Beer, 1972).

Beer’s didn’t actually claim a biological basis for his Viable System Model – he made only an analogy.

There are ways in which an enterprise acts like an organisation; there is stimulus-and-response across the system boundary.

(Though biological organisms are staggering complex and adaptable in a completely different way from how an enterprise must be adaptable._

There are ways in which enterprise architecture treats an enterprise as a species  the enterprise adapts when better business systems replace ill-fitting ones.


and even deterministically (Forrester, 1961 and 1971).

Forrester’s system dynamics is by the by.

Enterprise architects can and do model a business (a formalised social system) as a network of deterministic processes.

This is not to deny the importance of human intelligence in a business.

It is simply to recognise that every large business has to assume a reasonable degree of predictability in the processes it executes and the services it delivers.


Ecological Systems - in detail

Where is the boundary of the ecology? If it is unbounded, then surely it is better called the environment?

Ackoff Commentary


Nature is the most familiar and inclusive ecological system.

Are there any other examples? The earth? An ocean? An island? A national park? A zoo?

Aren’t these nothing more or less than 3 dimensional spaces?


Some parts of ecological systems can display choice but the whole can not.

Why not? The authors say we refer to social systems making choices. We also refer to mother nature, her moods and her favours.


We can affect our environments and often do, but the ways the environments react to our actions are determined.

So the environment is mechanistic? I am baffled here.


Ecological systems are living systems in the sense contained in Zeleny’s definition.

This seems an unsupported assertion. In what sense is the earth, an ocean, an island, a national park or a zoo alive?


They are self-organizing and self-maintaining.

Really? Is this the Gaian hypothesis? The scientifically accepted form of the Gaian hypothesis ("influential Gaia") states only that

·         biota organisms in an environment minimally influence certain aspects of

·         the abiotic world their environment, e.g. temperature and atmosphere.


Life is currently defined in terms of autopoiesis: the maintenance of units and wholeness, while components themselves are being continuously or periodically disassembled and rebuilt, created and decimated, produced and consumed. (Zeleny, 1981, p. 5).

As we will see, it follows from this definition that animated, as well as, social systems are living systems.

Some of the authors’ mechanistic systems (plants and body parts) are also alive. If autopoiesis is not a distinguishing feature of any category, then why mention it?


Animate and social systems are frequently confronted with situations in which their choices can affect their effectiveness, either positively or negatively.

Does this applies equally to choices made by systems the authors regard as mechanistic (automobiles, plants, software systems, bacteria, lice, ants, slugs)?


Such situations are problematic. In other words, problems are situations in which a system's choice can make a significant difference to that system.

And the challenging situations that face a slug, a plant, our lungs, or any homeostatic device are problems in this same sense?

On Ackoff’s conclusions

We have argued that it is useful to cast systems and their models into one of four types: deterministic, ecological, animate, and social systemic.

I’m not sure how useful the categorisation might be. People do seem to like it, yet misread it. And the class distinctions seem highly questionable to me.


The difference between them is a matter of "choice."

·         Deterministic systems and their parts display no choice.

·         The parts of ecological systems can display choice but not the whole.

·         Animate systems can display choice but their parts can't.

·         Social systemic systems display choice and their parts do as well. In addition, they are usually parts of larger systems that also display choice and contain other systems that do so as well.

I feel the authors make a self-referential argument for using their definitions of “choice” and “purpose” by inventing a classification that depends on us accepting those definitions.

I don’t see their definitions as a good or effective way to distinguish the classes they have named.

Surely the two distinguishing features of human choice (compared with machine choice) are self-awareness and unpredictability?


Our point has been that when models of one type are applied to systems of a different type, at least as much harm is done as good.

Deterministic models are certainly not enough, but we do need them, and EA is much about filling that need.

Business systems can be seen as formalised social systems

We have to design human roles and processes in business on the assumption that inputs, choices, actions and results are reasonably predictable.


The amount of harm (hence good) that is done depends on the level of maturity that social systems have reached.

What does maturity mean? Some regard the digitisation of human systems as progress. Is this another circular definition?


Society and the principal private and public organizations and institutions that it contains have reached a level of maturity that eliminates whatever effectiveness applying deterministic and animalistic models to social systems may once have had.

Isn’t the reverse equally true? Human organisations and institutions started as informal social systems and have been incrementally mechanised?

I’d be the first to suggest that mechanisation of human activities can be a bad thing, but if business owners didn’t find it effective, they wouldn’t do it.

And what is Facebook but an attempt to mechanise friendship relationships?


Finally, we showed five characteristics that we believe social systems designed as social systems should have in order to function as effectively as possible.

I discuss these five characteristics elsewhere. It does turns out that Ackoff’s socio-systemic view of organisations closely parallels TOGAF’s process for EA.


All free-to-read materials at are paid for out of income from Avancier’s training courses and methods licences.

If you find the web site helpful, please spread the word and link to in whichever social media you use.