The physics of systems

Copyright Graham Berrisford 2018. Last updated 07/12/2019 11:36

One of a hundred papers on the System Theory page at


 “Cybernetics depends in no essential way on the laws of physics.”

“In this discussion, questions of energy play almost no part; the energy is simply taken for granted.” Ashby

Nevertheless, though it barely concerns us most of time, this paper is about the physics of systems.


Classical mechanics and systems. 1

How does this relate to general system theory?. 2

Quantum mechanics and systems. 3

How does this relate to general system theory?. 4

Thermodynamics and systems. 5

How does this relate to general system theory?. 5

Time and systems. 6

How does this relate to general system theory?. 7

Energy optimisation and waste. 7

Evolutionary complexification. 8

How does this relate to general system theory?. 9


Classical mechanics and systems


Chunking space and time

Spacetime is a model that fuses the three dimensions of space and the dimension of time into a four-dimensional continuum.

The term “continuum” implies each dimension can be subdivided with no limit to size or duration.

Nevertheless, classical mechanics does divide the universe into discrete, measurable, chunks.


In computational physics/engineering the accuracy of solutions is improved by finer-grained chunking of space/time.

Sometimes, if the chunk size is incompatible with one set of physical laws, another set may be applied.


The object-motion duality

Physical reality is mysterious beyond human understanding.

Physicists observe and envisage that reality.

They tell us they describe it best in the mathematical equations of quantum mechanics.

Abstracting from the mathematic to ordinary words, they describe it in terms of particles or waves.

Quantum mechanics embraces classical mechanics as a special case - accurate enough in our everyday experience.

Classical (Newtonian) mechanics describes reality in terms of objects in that occupy space and move through time.

It begins with the quantitative variables of an object such as its mass, velocity and acceleration, and forces acting on it.


Classical mechanics

Objects and motions

<think in terms of>                  <represent>

Physicists          <observe and envisage >  Physical realities

How does this relate to general system theory?

We instinctively divide space and time into discrete entities and events

We usually divide space into entities that have physical phase boundaries (solid/liquid/gas).

And divide time into events that represent noticeable state changes, as when an object moves from place to place.


The structure-behavior duality in system description

The object-motion duality appears, in many ways, in the language we use to describe the world.

System describers typically classify system elements into structures and behaviors.


Structures in space

What the world is made of

Behaviors over time

What happens, what is done



















Stores (Stocks)

Streams (Flows)


Structures occupy addressable places in space.

Some structures are active, they act; others are passive, they are acted on.

Behaviors take place over time, and move or otherwise change structures.


The in-out duality in system description

Having boxed the universe into discrete objects, we can look at each object in two ways.

From the outside, we can see it as consuming and/or producing matter and energy.

Looking inside the box, we can describe what it contains and/or how it works.


The two dualities above are important in systems thinking.

System thinkers typically encapsulate a system behind an in-out boundary

And perceive and describe systems in terms of actors (structures) that occupy space and activities (behaviors) that occur over time.

This table maps one duality to the other.



Behavioral perspective

Structural perspective

External perspective

Discrete events

System interface

Internal perspective




Here is a different version of that same quadrant used in our system architect training.


Behavioural view

Structural view

External view

Service contract: an end-to-end process

defined as external entities see it.

Interface definition: a declaration of

available and accessible behaviours

Internal view

Process: a sequence of activities

performed by one or more components.

Component:  a subsystem capable of

performing one or more behaviours

Quantum mechanics and systems

Quantum mechanics has superseded classical mechanics as the foundation of physics.

It is needed to explain and predict microscopic processes at the molecular, atomic, and sub-atomic level.

It has a wider scope, and encompasses classical mechanics as a sub-discipline which applies under certain restricted circumstances.


Scientists believe our universe started with a big bang about 14,000 million years ago.

To begin with there was a lot of energy.

The temperatures were incredibly high, much higher than the interior of the sun.

Very soon after the big bang, some of the energy was converted into particles of matter called “quarks”.


After a tiny fraction of a second, quarks fall apart into other particles, such as the omega-minus

·       the omega-minus can decay into a neutral pion and xi-minus,

·       the pion decays into photons,

·       the xi-minus decays into a negative pion and a lambda.

·       the lambda decays into a negative pion and a proton.


Thus, energy was transformed into various kinds of matter.


Today, quarks don’t normally exist

Though they occur when cosmic rays (which are mostly protons) strike atomic nuclei in the earth’s atmosphere..

And they can be produced in a particle collision in an accelerator,


Today, at much lower temperatures, the matter around us is composed atoms, made of neutrons, protons and electrons
And you can’t normally see quarks directly, because they are permanently trapped inside other particles like neutrons and protons.


How to describe a quark?

In classical physics you could think of a quark as a point.

In quantum mechanics a quark is not exactly a point; it’s quite a flexible object.

Sometimes it behaves like a point, but it can be smeared out a little.

Sometimes it behaves like a wave.

(Edited from a discussion with Murray Gell-Mann.)


The Schrödinger equation represents the possibilities of an electron's characteristics as a wave of chance.

The particle is everywhere at a bunch of speeds … some more likely than others.

A split second after the particle starts moving, you can be fairly sure it's still near its starting point.

Over time, the range of its possible positions and speeds expands.

However, Schrödinger's equation is reversible

Which means a 'smeared' particle can localise back into a small region of space over the same time period.


Notice there are two ways of looking at subatomic entities, as particles or waves.


                      Quantum mechanics

Particles or waves?

<think in terms of>               <represent>

Physicists  <observe and envisage> Quantum entities


Neither description is “right”; but both are true to the extent that they are useful for particular purposes.

How does this relate to general system theory?

Ashby, Ackoff, Checkland and other systems thinkers distinguish abstract systems from concrete systems.

This triangular graphic shows the abstraction of system description from the physical world.


System theory

Abstract systems (descriptions)

<think in terms of>                              <represent>

System theorists   <observe and envisage >  Concrete systems (realisations)


These papers take this triangular view of system theory as axiomatic.

An abstract system is a description of how some part of the word behaves, or should behave.


Ashby spoke of what is called a Discrete Event-Driven System (DEDS).

In his kind a system, there is:

·       An abstract system of state variables whose values are changed by events.

·       A concrete system of physical entities that realises the abstract system.



Event and state variable types

<think in terms of>                     <represent>

Cyberneticians   <observe and envisage >  Physical entities changing


The abstract system does not have to be a perfect model of what is described; it only has to be accurate enough to be useful

An exception is the code of a software system, which is expected to be a perfect representation of the run-time system.

Thermodynamics and systems

Scientists discuss thermodynamics with reference to a system and its surroundings.


First law of thermodynamics: Energy cannot be created or destroyed inside an isolated system.

Energy is the capacity for doing work; it may exist in various forms, and be transferred from one body to another.

It can be input or output.


Second law of thermodynamics: The entropy of an isolated system always increases.

Entropy is the measure of a system’s thermal energy per unit of temperature that is not available for doing useful work.

Energy transforms and spreads out from areas where it's most intense.


The second law of thermodynamics is a principle more than a rule.

In classical physics, it explains why the balls on a pool table don’t reform the starting triangle.

If you saw pool balls reform their starting triangle, it would be a sobering experience.

It is so incredibly unlikely you would be shocked.

You'd probably need to stare at billions of pool tables forever to see it happen once.

In quantum mechanics, such strange things do happen.

How does this relate to general system theory?


Negentropy as a kind of order – created by consuming energy

The opposite of entropy is called negative entropy or “negentropy”.

Systems thinkers speak of systems maintaining negentropy - or maintaining order.


To maintain the order inside a system requires the input of energy from the rest of the universe.

The maintenance of order inside the system is more than paid for by an increase in disorder outside - by the output of heat.


Erwin Schrödinger (1887 –1961) discussed the thermodynamic processes by which organisms maintains themselves in an orderly state.

Ludwig von Bertalanffy (1901-1972) considered an organism as a thermodynamic system in which homeostatic processes keep entropy at bay.

“By importing complex molecules high in free energy, an organism can maintain its state, avoid increasing entropy…." Bertalanffy

“Nature's many complex systems - physical, biological, and cultural - are islands of low-entropy order within increasingly disordered seas of surrounding, high-entropy chaos.” Cornell University web site.


Simply put, the plants in the biosphere maintain its order by consuming high-level energy from the sun.

Each animal maintain its internal order by consuming high-level energy in the form of complex molecule).

It increases the disorder of its surroundings by producing lower-level waste products (simpler molecules) and heat energy.


However, thermodynamics is irrelevant to most systems thinking, certainly at the level social systems. 

In Ashby's cybernetics and Forrester's System Dynamics, the provision of sufficient energy to drive the system is taken for granted.

Introduction to cybernetics, 1956: 

Page 3: "In this discussion, questions of energy play almost no part—the energy is simply taken for granted.

Even whether the system is closed to energy or open is often irrelevant."

Page 136: "Sometimes the second law of thermodynamics is appealed to, but this is often irrelevant to the systems discussed here."


Information as a kind of order – created by coding and decoding

There is an analogy to be drawn.

In thermodynamics, think of energy as either stored in a structure, or flowing between structures (e.g. light, heat, force and sound).

In cybernetics, think of information as either stored in a memory structure, or flowing between structures in messages.


But it seems to me some confusion arises from the ambiguity of "information". 

In the domain of information theory, information exchange implies a sender and a receiver who code and decode a structure with two or more states. 

In the domain of thermodynamics, negentropy = order = information, which I'd rather call "information potential".


 “Living matter evades the decay to thermodynamical equilibrium by homeostatically maintaining negative entropy (today this quantity is called information) in an open system.” Cornell University web site.


In 2009, Mahulikar & Herwig re-defined the negentropy of a dynamically ordered sub-system.

Negentropy = the entropy deficit of an ordered system relative to its surrounding chaos.

“Negentropy might be equated with “free energy” in physics or with “order”; some equate it with "information".


In cybernetics and systems thinking "information" usually has a more specific meaning.

Information is the meaning created or found by an actor in a message or memory – using energy.

To encode meaning in a message or memory is to create a very specific kind of order.

To decode meaning from a message or memory is transform order from one form into another.

Time and systems

Remember the second law above: The entropy of an isolated system always increases.

Energy transforms and spreads out from areas where it's most intense.

This is a one directional law; whereas most other of laws of physics can be reversed and still make sense.


Time is change, or the processes of change.

We all presume time inexorably moves in one direction - forwards.

Suppose you stood outside time, looked at the world, and saw people frozen in motion.

You would assume, for them, time had stopped.

The people are growing neither older or younger.


Physicists say time can flow backwards as well as forwards.

“Physicists Just Reversed Time on The Smallest Scale by Using a Quantum Computer”

Surely these physicists did not reverse time; they only simulated reversing time using a computer?

If they did reverse time outside of the computer, they would not only undo the experiment, but forget they intended to do it!


At the macro scale at which our brains work, the second law of thermodynamics applies.

As time moves forward, the process of physics, chemistry and biology run in one direction.

You create memories by using energy to create orderly patterns in your brain – and grow older


As time moves backward, the process of physics, chemistry and biology run in the opposite direction.

You forget things and grow younger.

So, the only time direction you can appreciate is the forward one.

How does this relate to general system theory?

In describing systems, we take it for granted that processes run forward from start to end.

An asteroid is formed in space, then flies around for eons until it crashes into the moon

The current structure (or state) of the system can be seen as a memory of all that has happened to it.

And to create or maintain order in its internal structure, a system must consume energy.

Energy optimisation and waste

Energy is the quantitative property that must be transferred to an object in order to perform work on, or to heat, the object.

Energy is a conserved quantity; the law of conservation of energy states that energy can be converted in form, but not created or destroyed. (Wikipedia 12/02/2019)


Biological systems find energy in sunlight and food.

Many man-made systems find their energy in oil or electricity

Software systems consume electricity of course.


To encode or decode information requires energy.

Usually, we take it for granted that information process consumes relatively little energy compared with mining, machines.and material processing

However, the energy needed to do information processing is becoming an issue.


Principles: systems in competition tend to optimise their use of energy

This “optimal use of energy” principle means that systems which consume less energy are more likely to survive.


“Nature's many complex systems--physical, biological, and cultural--are islands of low-entropy order within increasingly disordered seas of surrounding, high-entropy chaos.

Energy is a principal facilitator of the rising complexity of all such systems in the expanding Universe, including galaxies, stars, planets, life, society, and machines.

Energy flows are as centrally important to life and society as they are to stars and galaxies.

Operationally, those systems able to utilize optimal amounts of energy tend to survive and those that cannot are non-randomly eliminated.” Cornell University web site.


This “optimal use of energy” principle has been at work in the evolution of biological systems.

But where minimising energy consumption is of little or no advantage, evolution proceeds in a suboptimal way.


Principle: where resources are cheap, systems tend to sub-optimise use of energy

Human society uses much energy in keep the internal temperature of buildings in a comfortable range.

The highest energy consumption per head is found in countries that are too cold (Iceland, Canada).

And too hot (Trinidad and Tobago, Qatar, Kuwait, Brunei Darussalam, United Arab Emirates, Bahrain, Oman).


But in not countries in Africa or South America, because another factor is money.

The highest energy consumption per head is also found in countries too rich to care about the cost (Luxembourg, and the United States).


The energy consumed by software is becoming a problem for society – not least in global warming.

The fear is that many modern software systems are over vastly over complex and suboptimal.

Because we have been careless in their design and given them as much memory space and electricity as their design needs.


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

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.

How does this relate to general system theory?

Suppose EA leads me to integrate my CRM and Billing systems.

Think thermodynamics: he integration increases the order and complexity of the enterprise-as-a-digital-system.

Think trade offs; it reduces the energy spent the human activity system.

But increases the energy required in the computer activity system, and the maintenance thereof.


Enterprise architecture encourages standardising and integrating business roles and processes.

The intention is to reduce the variety of behaviours and remove duplications between systems.

In the baseline state, the enterprise has many small and disintegrated systems.

In the target state, the enterprise has fewer and more integrated systems.


This reduces the overall size and complexity of smaller, silo, business systems.

At the same time, it increases the overall size and complexity of the business-as-a-system.

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

This has pros and cons.

A limiting factor is the ability of humans to manage the business-as-a-system.


Complexification - incremental extension and elaboration of a system – has pros and cons.

Through successive releases, it seems a software application grows larger than its additional features justify.

How to constrain complexifcation?

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