System stability and change

Copyright Graham Berrisford 2017. Last updated 13/01/2019 13:54

One of a hundred papers on the System Theory page at

Find that page for a link to the next System Theory for Architects Tutorial in London.



Heraclitus of Ephesus was a Greek philosopher known for his doctrine of change being central to the universe.

Plato quoted him as saying “Everything changes and nothing stands still.

Yet central to human existence is that some things are stable, or stable enough, for a while.

And history repeats - some conditions and reactions to them recur – which is symptomatic of a system.


This paper discusses system stability and change with reference to Ashby’s ideas.

A conclusion is that it is vital to distinguish different kinds of change.


Note: This paper is not about social network analysis (SNA) or the problems that it solves.

Here the term social network is used more generally, to mean any society, social entity or social group.


A few terms (recap) 1

Social system stability. 2

Two kinds of system change. 3

Two kinds of system state change. 4

Two kinds of system mutation. 5

“Complex adaptive systems”. 7

Conclusions. 7

Footnotes. 8


A few terms (recap)

The term system is widely used, but loosely, and with a variety of meanings.

At its most vacuous, it means only "a set of things that are related to each other, if only by their relationship to something else.”


The concept was defined more purposefully, in his introduction to cybernetics, by Ross Ashby.

The question is not so much "what is this system?" as ''what does it do?"

General system theory is mostly about systems that exhibit regular or repeatable behaviors.


A concrete system contains actors and their actions on objects or variables.

An abstract system contains the roles and rules that actors and their activities are supposed to adhere to.

The latter is abstraction from the infinite complexity of any entity that realises it.

E.g. The roles and rules of the stickleback mating ritual are realised by countless pairs of sticklebacks.

In discussion and testing of the mating ritual, no attention is paid to the complexity of a stickleback’s internal biochemistry.


General system theory concepts are ubiquitous in modern systems analysis and design methods.

This table presents them in a way that can be used in discussion of social entities.


Abstract social system

A set of roles and rules (the logic or laws actors follow)

Concrete social system

Actors playing the roles and acting according to the rules

Social network

Actors who inter-communicate and act as they choose


By the way, a social cell is a social network in which actors find that realising the roles and rules of a particular social system is so attractive they resist any change to it.

Social system stability

The systems of interest here are stable islands of orderly behavior in the ever-unfolding process of the universe.

W Ross Ashby said they exhibit behaviors behaviors that are “regular, or determinate, or reproducible.”


You might think of the solar system, riding a bicycle or the cardio-vascular system.

The interest here is a social network in which actors act in response to information encoded in messages and memories.

The physical actors in the system may be replaced, but its roles and rules persist.


The stability of some systems – both natural and designed - is central to human existence.

We expect at least temporary stability in the observable behavior of

·         banking services

·         church services.

·         electricity supply

·         elevators

·         health care operations

·         package delivery operations

·         payroll operations

·         tennis matches

·         traffic control systems

·         voting systems.


Stable does not mean static - all the above are dynamic activity systems

It means that system roles and rules are stable for a generation.

At least, long enough to pass whatever system tests are deemed as definitive of the system.


E.g. The structural roles (sun and planets) and behaviors (planetary orbits) of the solar system are stable for eons.

At least, long enough to match an astronomer’s description of them.


Stable does not mean unchanging   the roles and roles of the system can be changed.

It means only that system mutations happen in discrete steps.

In business systems, so-called “continuous improvement” is typically managed in discrete cycles – under change control.

Two kinds of system change

Ashby distinguished two kinds of change thus:

“The word "change" if applied to [an entity repeating a behavior] can refer to two very different things.

·         change from state to state, which is the machine's behavior, and which occurs under its own internal drive, and

·         change from transformation to transformation, which is a change of its way of behaving, and occurs at the whim of the experimenter or some other outside factor.

The distinction is fundamental and must on no account be slighted.” Ashby 1956


Ashby’s two kinds of system change might be defined thus:

·         System state change: e.g. regulating the values of defined state variables to stay within a desired range.

·         System mutation: e.g. reorganizing or changing the state variables, or the rules that update variable values.


Ashby’s distinction may be expressed in terms of three principles.


Principle 1: The state of an activity system may change.

E.g. In Forrester's System Dynamics, the stocks in system model may grow, shrink or even be exhausted.

Similarly, the state variables (temperature, salinity, etc) of a biological entity vary over time.


Principle 2: The roles and rules of a system are fixed for a system generation.

E.g. In Forrester's System Dynamics, the roles of stocks, and the rules governing flows between stocks, are fixed for a generation.

Similarly, the form and functions of a biological entity are encoded for a generation in its DNA.


Principle 3: Changing the roles or rules makes a new system or system version.

Changing roles and rules makes a different system - or at least, a new system version or generation.

In Forrester's System Dynamics, the model of stocks and flows can be changed, and the model rerun to see how its outcomes differ.

In Darwin's theory of evolution, the form and functions of a species changes (very slightly) as its DNA changes from generation to generation.


In short, it is important not to confuse system state change with system mutation.

Evolutionary biology plays two roles in these papers.

It explains why animals retain mental models of the world, which helps us to answer questions about the description-reality dichotomy.

It also helps us distinguish system state changes from system mutations.


·         Individual organisms experience system stage changes – as discussed in cybernetics and general system theory.

·         A species undergoes system mutations - via reproduction and evolution.

Two kinds of system state change

This section distinguishes discrete state change from continuous state change.


System state change: a change to the state of a system, which changes the value of at least one variable.

State changes happen during the regular processes of the system.

E.g. winning a set in a tennis match; or the changing positions of planets in their orbits.


Early system theorists were especially interested in homeostatic systems.

They studied dampening feedback loops that maintain system state variables (e.g. body temperature) in a desirable range. 

System theorists called such system state changes adaptations, meaning they adapt the entity so as to maintain homeostasis.


But not all state changes are homeostatic, or well called “adaptations”.

Some systems continually amplify or increase a variable, such as income, profit, or happiness.

And most information systems update state variable values to reflect changes in a reality of interest.


So, system state change varieties include the following.

·         Homeostasis – the process by which a system responds to internal and external state changes so as to maintain its state variables in acceptable ranges.

·         Information update – the process by which a system’s information state is changed to reflect the state of its environment.

·         Autopoiesis – the process by which a biological cell, given simple chemical inputs, sustains/replicates its own complex chemistry.


The dynamics of a system are how the state of that system changes over time.

State changes may happen in discrete steps, in response to discrete events, or continuously.


Discrete (or digital) system state change

This means a system’s state advances incrementally in response to discrete events.

The term digital is used to describe signals that are chunked into discrete units, as in a clock that displays the time as numbers. 


Continuous (or analogue) system state change

This means a system’s state advances continually in response to continuous forces or inputs.

The term analogue is used to describe signals that vary continuously, as in a clock with revolving hands.


Most business systems are driven by discrete events and experience discrete state changes.

So they are naturally modelled as what is called discrete event-driven systems (DEDS).

However, continuous systems can also be modelled as discrete event-driven systems.

“We shall assume that the change occurs by a measurable jump, as the money in a bank account changes by at least a penny.

consideration of the case in which all differences are finite loses nothing; it gives a clear and simple foundation;

and it can always be converted to the continuous form if that is desired.” Ashby 1956

Two kinds of system mutation

This section distinguishes discrete state mutation from continuous system mutation.


System mutation: a change to the nature of a system, which changes the type of at least one variable or behavior.

Mutations happen when the processes of the system are changed

E.g. The Lawn Tennis Association publishes changes to the laws of tennis.

Or an asteroid knocks a planet out of its orbit.


System mutation varieties include the following.

·         Reproduction with modification - the process that creates a new, different, descendant system (generation N + 1).

·         Maturation – the process by which seeds or eggs develop into adult forms.

·         Learning – the process by which intelligent actors (animal or machine) respond differently to a new input after recognising its resemblance to past inputs (implies fuzzy logic).


Reproduction with modification happens in clearly identifiable discrete steps, each system generation replacing the last.

By contrast, maturation and learning may happen continuously, or in such small steps it is regarded as continuous.


Meta system: a system that transforms a system from one generation to the next.

Meta systems include biological evolution, which depends on the process of sexual reproduction and DNA to define and change an organic system.

And system architects who use some kind of methodology to define and change systems.


Meta System


System roles

<define>                        <idealise>

System architects  <observe and envisage>  System actors



System actors  <perform>  System Roles


Activity systems are often called transformations.

1.      At the operations level: system actors perform roles to <transform> business inputs into outputs.

2.      At a meta level: system architecture functions <helps managers to transform> system roles.

3.      At a meta meta level: methodologists <help managers to transform> system architecture functions.


System actors can act as architects of the system they work in.

Actors in a given organisation structure may change their roles and processes.

Actors performing given processes may change their organisation structure.

When doing this, when changing the system they work in, they are actors in the meta system.

Discrete system mutation

Discrete system mutation: a step change from one system generation to the next.


To change a biological organism is to change its DNA.

Discrete-step generational system change is the natural course of events in biology.


To change a System Dynamics model is to change its stocks and flows.

Discrete-step generational system changes are made by the system modellers to improve the model.


To change a business system is to change its roles and rules.

Change requests may come from owners, sponsors, observers, designers and actors in the system.

In business, so-called “continuous improvement” is typically managed in discrete cycles

The cycle is sometimes called PDCA (plan–do–check–act/adjust), the Deming circle/cycle/wheel, the Shewhart cycle, or the control circle/cycle.


The purpose of an enterprise architecture function is to design and plan discrete-step mutations to business systems.

The enterprise function is a meta system to the business systems it observes and envisages.


How to distinguish descendants from ancestors?

Most parents give children different names; however, the veteran golfer Davis Love III is now a caddy for Davis Love IV.

Version numbering usually implies ancestors will eventually be replaced by descendants.


The ancestor system

the descendant system is named

or else

is replaced by a descendant

using the same name

the same name with a new generation or version number

continues in parallel with a descendant

using a new name


Evolution by discrete step mutation

The process of Darwinian evolution can be separated from biology, and from the processes that sustain a system during its life time.

It is an inter-generational process that reproduces system version N with small modifications in system version N+1.

The better that system N+1 is fitted to its environment, the more likely it will be reproduced again.


One ancestor system may give rise to two or more descendent systems at generation N+1.

The two descendant systems will likely be regarded as relatives or members of the same species.

Successive generation changes will likely lead to increasingly diverse descendant systems.

Eventually the descendant systems will likely diverge so far that they can no longer be recognised as relatives or members of the same species.


“Since offspring tend to vary slightly from their parents,

mutations which make an organism better adapted to its environment will be encouraged and developed by the pressures of natural selection,

leading to the evolution of new species [eventually] differing widely from one another and from their common ancestors.”

Continuous system mutation (or re-organisation)

Above, a system is a stable island of orderly behavior in the ever-unfolding process of the universe.

The notion of continuous system mutation or reorganisation is an oxymoron; since it undermines the basic notion of a system.


Obviously, human actors have the ability to stand back from their role in a system and propose changes to it.

But what if actors can continually change the system they act in?

If the roles and rules of a system are continually modified, there is no describable or testable system.

There is ever-unfolding (and unpredictable) process rather than a system.

“Complex adaptive systems”

A critique of this concept needs more space, and is now presented in a separate paper.

Read Complex adaptive systems for more.


The first conclusion below is repeated from Social networks versus social systems.

The second is the conclusion of this paper.


How to restore the system concept to systems thinking?

It is commonly said that “enterprise architecture views the enterprise as a system, or system of systems”.

But there are misunderstandings of what this means.

To call every problem, situation or business “a system” is unhelpful.

It is important to distinguish:

·         a social network in which people communicate

·         a social system in which people realise role and rules.


An enterprise is one social network that realises many social and technological systems.

Architects may find some systems overlap, duplicate or even conflict with each other.


Changing the roles or rules of a system makes a new system or system version.

Enterprise architects apply some change control to large-scale, generational, system change.


How to extend system theory to embrace “self-organisation”?

Ashby insisted we distinguish two kinds of system change, which need different names here.

·         System state change: e.g. increasing turnover, decreasing profit.

·         System mutation: e.g. reorganising to offer new services, or offer old ones differently


We also need to distinguish:

·         a system – actors playing roles in regular business operations

·         the meta system – actors playing roles in operational research to study and define the system

And allow actors in a social network to switch between roles as rule followers in an operational system and rule definers in a meta system.


If the roles and rules of a system are continually modified, there is no describable or testable system.

So how to maintain the integrity of the system concept?

The answer is to insist that actors make incremental (generation-by-generation) rather than continual changes to system roles and rules.


Thus, it is a social network (not a social system) that has self-organizing dynamics.

And a “resilient adaptive self-organising system” might better be called a “resilient continuously evolving self-organising social network”.


Principles for optimising system mutation

Beware the spiral to inefficiency

  1. The more cheaply you can buy resources and readily you give them to a system
  2. The more designers will design resource-hungry systems
  3. The more resources you will buy.
  4. The cheaper the resources will become
  5. Return to 1 until you have exhausted the resources.


Inefficiency arises from

Optimisation arises through

Having only one design option

Having competing designs

Freely expanding resources

Limiting resources

Preventing change to a design

Enabling change to a design

Long generation/change cycles

Short generation/change cycles

Pricing based on desire to have the system

Pricing based on cost of making the system

Evolutionary complexification

Enterprise architecture has traditionally emphasised making step changes to standardise and integrate business processes.

The presumption is that this simplifies and optimises business systems.

The variety of behaviour is reduced, duplications are removed, the business is simpler


There is a different view of the effect this has on the enterprise.

The newly-created business system is larger and more complex.

Because standardisation increases the population of actors who play roles in one system.

And integration joins formerly-independent and simpler systems into a more elaborate system.

Also, integration of systems into one system creates “emergent” properties and issues.


Complexification - incremental elaboration of one system – can be good or bad.

Unconstrained complexification can lead to a bloated wasteful structure.

Consider how successive releases of a software application grow much larger than their additional functions would seem to justify.


So, how to constrain complexifcation?

One way is to pitch a variety of different complexifications against each other in a competition for resources or for survival.



Kenneth Lloyd has pointed me to a paper in the Journal of Artificial Intelligence Research.

The paper asks: how to discover and improve solutions to complex problems?

Experiments with robots suggest that complexifying evolution discovers significantly more sophisticated strategies than evolution of networks with fixed structure.

The experimental results suggested three trends:

(1)   As evolution progresses, complexity of solutions increases,

(2)   Evolution uses complexification to elaborate on existing strategies

(3)   Complexifying coevolution is significantly more successful in finding highly sophisticated strategies than non-complexifying coevolution.


The suggestion is that, to discover and improve complex solutions, evolution should be allowed to complexify as well as optimize.

On stability and change in living entities

Following the principles above, we may distinguish:

·         a biological organism: a social network of cells

·         a biological system: the roles played by cells and the rules they follow.


And distinguish two kinds of system change:

·         biological system state change, which occurs by homeostasis (as in classical cybernetics and general system theory).

·         biological system mutation, which occurs by reproduction and evolution.


The biologists Maturana and Varela characterised living entities as autopoietic (think, self sustaining).

An autopoietic organism manufactures its own body parts from primitive edible chemicals.

The processes make the structures, which perform the processes, which make the structures, which perform the processes, and so on.


There is no one definitive list of what characterises a life form, but there are similarities between lists in different sources.

And various system theory concepts can be found in the lists from two sources below.


In this source

living entities have

7 characteristics

In this source

living entities have

8 characteristics

System Theory concepts



System description/definition


Adaptation through evolution

System mutation



System mutation


Growth and development

System state change


Cellular organization

System composition


System input (material)


System input (information)


System output (material)



System processing (material)


Response to stimuli

System processing (information)



System processing (information)


Neither source mentions decay and death as characteristics that define a living entity.



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..