System Classifications – including naďve ones

Copyright 2017 Graham Berrisford. One of about 300 papers at Last updated 22/09/2019 10:51


In his paper on applying system theory to management science, Boulding (1956) outlined two approaches to system theory:

1.     Definition of common system properties - identification of features common to systems in different disciplines.

2.     Classification of system types – e.g. arrangement of system types into a hierarchy of complexity.


Boulding’s view of system properties was discussed in another paper.

He also arranged system types in a 9-level complexity hierarchy.

This paper challenges his classification of system, and then challenges whole idea of classifying systems.


A system classification based on system properties. 1

Boulding’s questionable classification. 2

Ackoff’s questionable classification. 3

A classification you might find interesting. 4

Aside on static systems (passive structures) 4

Conclusions and remarks. 5


A system classification based on system properties

This work discusses many kinds of system, including physical and biological systems.

It discusses natural systems (like the solar system) which evolve so as to behave in a regular or orderly fashion, with no given aim.

However, the systems of most interest to us are one or more of the following:

·       A dynamic system, noting that it can maintain a passive structure, such as a record of system state data.

·       A designed system in which actors perform activities to meet given aims. E.g. a cyclist pressing pedals to move a bicycle forward.

·       An open system which consumes inputs and produces, noting that the boundary is always a matter of choice.

·       A social system in which actors exchange meaningful messages. E.g. a business activity system.

·       A scripted system in which actors perform prescribed activities. E.g. some violinists following the score of a symphony.


This table is an attempt to arrange system kinds in a taxonomy.

It is flawed, since (for example) social systems can be designed and open systems can be natural.


Discrete entity


disorderly entity

chaotic (and so

not describable)


organised, orderly, stable (in described ways)

Passive structure

does not act

Dynamic system

acts an orderly or rule-bound way

Natural system


Designed system

e.g. symphony or software system


e.g. solar system


e.g. tree,


Social system

e.g. bee hive,

hunting party

Closed system

e.g. System

Dynamics model

Open system

I/O exchange

across boundary


This work makes no assertion about the usefulness of this classification; it is only a vehicle for helping people to appreciate the breadth of system varieties.

Boulding’s questionable classification

Boulding’s view of system properties was discussed earlier.

In the same paper he arranged system types in a 9-level complexity hierarchy, summarised in the table below.

This classification is highly questionable.

To begin with, Boulding confuses two scales: simple to complex, and part or component to composition.

Complexity level

Boulding says (p202 to 205)

Some questions

1 - static structure

Boulding’s examples are static descriptions

What about static concrete structures – such as buildings?

2 - dynamic system

e.g. clocks and other machines

Boulding’s dynamics seem limited to motions of matter. What about flows of energy and information? 

3 - control system

e.g. thermostats and other information processing systems

Ashby says the state of a control system (level 3) is infinitely simpler than the state of a machine it controls (at level 2).

4 - open system

simple life, self-sustaining, self-reproducing, like a cell

What about other system types that are “open”, such as a business system?

And note that cell is a component of classes 5 and 6, rather than a simpler form.

5 - genetic-societal

a plant composed of cells - little or no information processing

What about animals considered not to be self-aware (fleas, ants, fish, worms, bees, oysters)?

6 - self-aware animals

image-dependent processing

How is the constructed image different from any constructed state in a level 3 system?

7 - self-aware humans

whose internal state/memory is complex beyond our reckoning

What about apes, mammals and any other animal demonstrably self aware?

8 - social systems

a system of human roles

Surely a single human (or even a plant) is more complex than a social system of roles

9 - transcendental



Boulding made several interesting points; some clearly have some truth to them; some might be challenged; and some fall into both those categories


Boulding considered the processing of information (including images) to be a significant feature of self-aware systems.

Yet today, a number plate recognition system is an image-processing information system.

And modern biology regards all life (including plant life) to be a kind of information processing.


Boulding considered social systems to be more complex than the actors in them.

At one point he wrote "in a sense, each level incorporates the levels below it"

Yet elsewhere suggested a social system incorporates roles rather than actors.

And the roles of a system can be simple.

A simple system

Customer role

Supplier role

Place order

Send invoice

Send payment

Send receipt


This social system is infinitely simpler than the biology of the actors who play the roles.

And playing the roles may require little more than pressing buttons on a computer screen..


Boulding’s system classification is not the continuum he thought it was.

It mixes up abstract system descriptions with concrete system realisations.

The complexity of a system depends entirely in its describers.

Every concrete system realisation is infinitely complex – since it contains countless features irrelevant to the described system.

E.g. the complexity of life on earth is irrelevant to the role of the earth in the solar system described by astronomers.


Some of Boulding’s "levels" might better be considered as "views" of a system.

Each view can be detailed to at any level of abstraction or complexity you choose.


Remember, systems complexity is entirely in the hands of the system describer.

You cannot measure the complexity of an operational system directly - you can only measure the complexity of what appears in your description of that system.

Your description of a tree as a system might be much simpler than a biologist’s description of a single cell a tree leaf.

System theory relies our ability to form abstract system descriptions that hide the internal complexity of component parts.

A higher-level system may be seen as coarse-grained and simple, whereas its lower level components may be seen as complex.

Any higher-level system can contain many lower-level systems whose complexity is completely ignored in the higher level system description.


In short, a social system (at level 8) is far simpler than a human being (at level 7).

And a control system such as thermostat (at level 3) is simpler than any dynamic system such as a heating system (at level 2) that it controls.


Boulding questioned the adequacy of theoretical models above level 2.

And questioned whether there is even rudimentary theory above level 4.

Ackoff’s questionable classification

Ackoff proposed a 4 way classification in which he uses “choice” as the key differentiator between system classes.

Ackoff’s type of System Model




Deterministic (sometimes mechanistic)

No choice

No choice



No choice









No Choice



Ackoff’s classification is also highly questionable, and is challenged in this other paper.

A classification you might find interesting

This is not a formal classification; it is only a way to get you thinking about the application of system theory concepts.

Five overlapping kinds of activity system are tabled below.


A cooperation of

Biological entity

biological organs and cells.

Bio-social organisation

biological entities

Human cognition

brain cells involved in thinking

Human organisation

humans capable of human cognition

Deterministic system

components that execute deterministic processes.


Does the term “system” mean the same thing in each domain?

What is the difference between self-sustaining and self-replicating, between organised and self-organising, and between adapting to variety and adapting to novelty?

Aside on static systems (passive structures)

The loosest definition of a system as "a set of interrelated parts" allows that a system may lack any behaviour.

It could be passive hierarchical, lattice or network structure.


·       A description: a map, the Dewey Decimal System.

·       A fence, a bird table, a necklace, stained glass window, a church building, a mountain.

·       A musical score, a computer program listing, a model of a DNA molecule

·       An activity system at rest, a stationary bicycle, or any idle machine.

·       A dead activity system: a log, a corpse, a closed power station.


Bear in mind, you can choose to regard any of these as an activity system if you wish.

A system is what you describe it to be, provided your description is consistent with a recognised theory of what properties a system has in general.


Again, our interest is in activity systems that feature both actors and activities.

Enterprise architects are specifically interested human and computer activity systems (rather than machines such as photocopiers, cars or guns).

Conclusions and remarks

It is easy to invent classifications of system types, problem types, society types and personality types.

The classification schema may take the form of a simple scale, a hierarchy, a grid, a matrix or other.

There are many such system classifications, and most of them questionable.


People are drawn to schemas that promise to simplify the choices they have to make.

You may be tempted to use a schema to:

·       rewrite world history to match it, build a world view around it

·       help you choose one path or solution option over another

·       decide how people will be organised and/or what roles they play.

·       pigeon-hole somebody and start treating them as a type rather than an individual.


The trouble is – it is difficult to test the validity of schemas as a decision-making tool.

The danger is that a classification scheme can be used to:

·       avoid dealing with problems as individuals

·       avoid taking responsibility for a decision

·       hide personal prejudices behind a superficial rationale.


And since alternative choices are never explored, any or every use may be presented as evidence of the scheme’s validity.



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..