The philosophy of system theory
Copyright Graham Berrisford 2014. Last updated 16/02/2019 18:09
One of a hundred papers on the System Theory page at http://avancier.website.
Find that page for a link to the next System Theory for Architects Tutorial in London.
This paper develops a philosophical position statement summarized at the end
It does this by relating system theory to science and philosophy, with passing references to Nietzsche and Wittgenstein.
(If you are looking for the table of philosophical dichotomoties; it has been moved from the start to the end of this paper.)
Ashby, in his Introduction to Cybernetics, distinguished abstract systems from concrete systems.
Abstract systems are descriptions of roles, rules and variables that are exemplified in concrete systems.
The basis of system theory
<create and use> <idealise>
System theorists <observe and envisage > Concrete systems
These papers take Ashby’s view of system theory as axiomatic.
Some other viewpoints are challenged.
To perceive a thing (out there) is to form an internal representation of it.
Research (Anil Seth talk) suggests the brain does this by combining
· Observation: sensing information input from what is out there.
· Envisaging: making a best guess as to what has been sensed
We may hallucinate, we may perceive something where there is nothing.
But (despite Anil Seth’s talk title) to perceive what exists out is there is to model it, not to hallucinate something.
Our model may be a poor representation, it may fade to nothing, but still, there is something out there.
And our brains evolved to perceive what does exist out there.
Because our survival depends being able to do this reasonably well, most of the time.
To know a thing is to have access in our thoughts to a useful representation or description of it.
We can never know – perfectly - what a thing is; that is not even a meaningful suggestion.
We can only know how a thing is represented in some kind of description, model or theory.
However, we can also share our knowledge with others, and test that things turn out in the way our knowledge leads us to predict.
To deny that sharing and testing help to confirm our knowledge of the world would be to deny the history of mankind.
Relativism and perspectivism
Historical figures including Protagoras, Nietzsche and von Foerster have subscribed to a kind of relativism or perspectivism that can be misleading.
Friedrich Nietzsche (1844 to 1900) was a philosopher whose metaphysical ideas influenced many Western intellectuals.
“Nietzsche claimed the death of God would eventually lead to the loss of any universal perspective on things, along with any coherent sense of objective truth.
Nietzsche rejected the idea of objective reality, arguing that knowledge is contingent and conditional, relative to various fluid perspectives or interests.
This leads to constant reassessment of rules (i.e., those of philosophy, the scientific method, etc.) according to the circumstances of individual perspectives.
This view has acquired the name perspectivism.” Wikipedia December 2018
Protagoras, Nietzsche and von Foerster have a lot to answer for, as discussed in Postmodern Attacks on Science and Reality.
Some Marxists and postmodernists interpret perspectivism as meaning all descriptions of the world are equally valid.
At the extreme, this leads to the view that the “dialectic” is more important than evidence.
That any persuasively argued or widely believed assertion carries the same weight as science.
Or even that any personal opinion is as true as the facts the world’s best scientists agree.
Scientists are aware that our senses, perceptions, memories and communications are subjective and imperfect.
That doesn’t mean science is unreliable and should be discarded; the reverse is the case.
The scientific method is how we overcome our limitations as individual observers.
By repeated testing of results against predictions, logical analysis and peer group review - we incrementally improve our confidence that a model or theory is valid.
Communication involves sending and receiving messages that convey descriptions, directions and decisions.
Senders and receivers may use speech, writing, gestures and other kinds of signal.
Speakers create meanings in messages; hearers find meanings in messages.
Communication succeeds when created and found meanings are the same.
For successful communication, speakers and hearers must share the same language for encoding and decoding a message.
Communication between us only works when we share word definitions - well enough.
Advanced communication depends also on message senders and receivers having access to memories.
The Hermeneutic principle of communication?
The principle can be expressed as "The hearer, not the speaker, determines the meaning of a message."
This principle makes innocent speakers guilty of causing offence where none was intended.
The intention of a speaker does matter, practically, logically and morally.
There can be no absolute truth; in the sense that a description is never the same as the reality it describes.
However, there are degrees of truth - or confidence - in the accuracy and usefulness of descriptions.
The more we check a belief by testing and agreement with others, the more confidence we have in it.
Our survival as a species depends on that confidence having being justifiable in the past.
And the stunning success of hard science is proof that testing and peer review maximise the degree of truth.
What the world looks like is a description or model of what it is.
If we couldn’t form a decent model of the world, we wouldn’t survive.
As members of a social species, we necessarily see the world similarly.
Since our ability to perceive and communicate about things in the world evolved represent those things.
At least, represent them well enough we can determine our actions, cooperate socially, and survive.
Truth does not exist in a world without description.
Truth is a measure we apply to our descriptions of the world.
The truth of a model = the degree to which the model proves accurate and useful.
There are three ways to test the truth of a theory, description or model:
It is empirically true - supported by evidence from test cases.
It can help you recognise and predict what exists and happens in reality.
E.g. If you stay on the railway track, your belief that the train will strike you will be confirmed.
It is logically true - can be deduced from other concepts within a body of knowledge.
It follows logically from axioms (presumed truths) that body of knowledge is based on.
E.g. The force on a body struck by a moving train can be calculated from its mass and speed.
Last and least convincingly, it is socially true - widely believed in your social network.
Biological evolution has favoured social animals who usually communicate what is empirically true.
In the absence of empirical and logical evidence, you may retreat to the Nietzsche-like presumption that “shared perception is reality”.
But the flexibility of human speech has made the communication of fanciful and misconceived ideas commonplace.
Scientists take care to ensure social truth does not over-ride the first two tests above.
Naturally, we observe the world and describe it to others.
We test what we are told against our experience of the world.
The scientific method formalises our natural approach to describing and testing reality.
The basis of science
<create and use> <exemplified in>
Scientists <observe and envisage > Realities
Inevitably, our descriptions of reality - by hiding its infinite complexity - tidy it up.
And any description of description does the same.
But our everyday concern is not whether verbal descriptions are perfect, or circular.
It is rather, are they useful; do they help us to understand, test and predict things in reality?
“Modern physics strongly suggests ... reality is very much like what was inferred by some remarkable thinkers in the ancient world:
a universe composed of elementary objects that move around in an otherwise empty void.” Postmodern Attacks on Science and Reality
Physicists have invented ways of describing the universe in terms objects that move.
The basis of physics
Object and Motion Types
<classify and use> <idealise>
Physicists <observe and envisage > The universe
In classical physics, space and time are presumed to be continuous.
But still, to describe the world, physicists divide to into discretely measurable objects and motions.
And in quantum mechanics, tiny atoms, particles and changes are described as discrete objects in space and jumps in time.
Neither classical physics nor quantum mechanics is only a speculative description of reality.
Tests show that, in particular circumstances, they each represent or symbolise reality well enough.
Which is to say, both descriptions are true, or at least, true enough to be useful.
Most human descriptions of reality are digital in the sense they divide it into discrete structures (entities, actors) and discrete behaviors (events, actions).
And like physicists, system theorists perceive and describe the universe in that same very general way.
They see it as composed of actors or objects that occupy space at a moment in time and act or change over time.
Where do numbers come from?
That's easier for a psycho-biologist than a mathematician or physicist to answer.
The evolution of concepts and numbers
Animals evolved to perceive the universe in terms of discrete things in space.
Animals evolved to recognise family resemblances between things. E.g. food items, friends and enemies.
An earthworm knows enough to recognise another worm of the same type – for mating purposes.
Certainly, a worm can recognise others members of the worm set, though it may not remember them, and surely cannot count them.
Experiments show babies, before they have words, can recognise when a small group of things gains or loses a member.
And dolphins can recognise which of two boards has, say, five dots rather than six.
Animals evolved to communicate facts about things that resemble each other.
The members of a species must have a similar idea of what food, friends and enemies have in common.
The survival of a social group depends on its members sharing ideas, like where food can be found.
Many animals can communicate facts about things of interest by gestures and/or noises.
Honey bees can communicate the direction and distance of a pollen source.
Astonishingly, experiments suggest honey bees can count up to four and communicate that amount to other bees.
Humans evolved to create words, to suggest and discuss similarities between things.
We formalised family resemblances into a type, an intensional definition (e.g. “sister”).
So, where do numbers come from?
They come from when animals started counting the members of a family, kind or type.
Since, as soon as we have a family or type in mind, we can count the members of the family or type.
And then, numbers are types.
Since, as soon as we can count, we find many families and types have something in common.
That is, they share the total number of their members.
We use numbers as types, to describe what groups of the same size have in common.
· “oneness” is the property shared by all groups with one member
· “twoness” is the property shared by any one-thing group to which we have added one.
· “empty (zeroness)” is the property of any group from which we have removed all members.
The human ability and urge to generalise across groups and types of different things led to numbers.
And numbers are the basis of mathematics, hard science and the types used in software.
The basis of mathematics
<create and use> <represent quantities of>
Mathematicians <observe and envisage > Instances of Types
Realities are composed of matter and energy that exists; meaning it can be located in space and time.
A description is also physical matter/energy structure.
But is intentionally created by a described to represent something else – be it observed or envisaged.
The basis of description
<create and use> <represent>
Describers <observe and envisage > Realities
There was no description of reality before life.
Description is a side effect of biological evolution.
Descriptions appear in both internal mental phenomena and in external representations
The external representations can take the form of speech, writing and models of other kinds.
Internal and external memories are different in many ways.
But similar in the most important way; they are created with the intent they should be retrieved and used.
Meaning (or information) is the intent of a description creator or the interpretation of a description user.
Many more or less accurate copies of a description can be made.
There is no ethereal description aside from what exists in one or more copies of it.
Delete all copies of a description and it disappears from the universe.
Descriptions are created when actors encode them in some form of matter and/or energy.
Descriptions are used when actors decode them from those forms.
Communication between actors succeeds when the encoded and decoded meanings are the same.
Or rather, near enough the same to be useful, because there can be degrees of truth.
Since individual rocks share some common qualities, they are readily perceived as instances of the same kind.
The ability to describe things and ideas using words (and graphical symbols of them) dramatically extended human descriptive/typification ability.
The basis of verbal description
<create and use> <symbolise>
Humans <observe and envisage > Realities
The types we symbolise using the words “rock”, “plant” and “circle” have been created, remembered and communicated countless times.
That does not mean any type exists independently of its appearance in a description, in a more ethereal form.
It only means that many observers have generalised the same or similar type from a set of similar concrete realities.
Surely when all “rock”, “plant” or “circle” descriptions are destroyed, then that concept or type will disappear from the universe?
There is no type or concept outside of a description encoded in a matter and/or energy structure?
Many (perhaps most) mathematicians are reluctant to believe that there were no numbers before mankind, or life.
But surely, you cannot have numbers until you have types, of which instances can be counted?
And you cannot typify things until there is some kind of intelligence.
So, numbers only existed the form of types when life forms started to create, remember and communicate types
On the other hand, there were always things that can (in retrospect) be regarded as similar.
In the history of the universe, this was first true at the level of atomic particles, then stars and planets.
So, numbers always existed in the sense that numerous similar instances of (what we now choose to describe as) a type have existed.
The more general question is whether there were any types before life.
The premise here is that there were no types, no descriptions, before life.
The idea of an ethereal type is useless, redundant, and better cut out using Occam’s razor.
We describe particular situations, entities and events by attributing universals to them.
A universal is a property, quality, characteristic, attribute or type such as “tall”, “yellow”, “circular” or “dangerous”.
Sometimes we attach a measure or degree to the property, like 2 metres tall, or very dangerous.
The basis of verbal description
<create and use> <typify>
Describers <observe and envisage > Particulars
The “problem of universals is the question of whether universal properties exist - or what it means to “exist”.
Traditionally, “realist” and “idealist” philosophers disagree on whether universals exist independently of thought and speech (and records of those).
Philosophers, mathematicians and scientists have debated this for millennia, at least since Plato and Aristotle.
Another paper - The problem of universals - concludes that to a psycho-biologist, the problem of universals is something of a fake problem.
Or at least, the distinction between realist and idealist philosophical positions is a false distinction.
Over centuries, idealism and idealism have evolved into a confusingly diverse and overlapping mess of different positions.
And today, what that other paper calls Scientific Idealism and Scientific Realism are much the same.
Ludwig Wittgenstein (1889-1951) influenced the “Vienna circle” of logical empiricists (aka logical positivists).
He argued philosophical disagreements and confusions can be resolved by analysing the use and abuse of language.
In his “Tractatus Logico-Philosophicus” he set out seven propositions.
The propositions are famous for being a tough read, and have been interpreted in various ways.
That doesn’t matter here, because Wittgenstein later realised his tractatus was self-contradictory.
In “Philosophical Investigations”, published after his death, he developed an entirely different linguistics.
He turned his focus from the precision of language to the fluidity of language.
He dropped the metaphor of language “picturing” reality and replaced it with language as a tool.
In so far as philosophy is about language, knowledge and truth, it seems to have been overtaken by biological and software sciences.
This new attempt at a “tractacus” is written from the perspective of a psycho-biologist rather than a linguist or mathematician.
-1- Reality is what exists in matter and energy, in space and time.
The context is the universe we live in, as observed and described by physicists.
Whether the same system theory would apply in alternative universes is out of scope here.
-2- A description is created by an actor (internally or externally) to represent a reality that is observed or envisaged.
A description is a representation of a reality – real or imagined.
Other kinds of information (directions and decisions) are mostly out of scope here.
-3- A description is also a part of reality
Description is not an ethereal concept.
-4- There are degree of truth in description creation and use, which makes them subjective.
The words “true” and “false” may be read as “true enough” and “false enough”.
Because truth and falsehood are judgements made by description creators and users at a moment in time.
And those judgements may be different on different occasions.
E.g. “Standing on that railway track is dangerous” may be judged as:
· True to an actor who creates the description with the intent to represent a reality - well enough to be useful.
· True to an actor who finds (in empirical or logical tests) that it does represent a reality - well enough to be useful.
E.g. “This horse has five legs” may be judged as:
· False to an actor who creates the description with the intent to misrepresent a reality.
· False to an actor who finds (in empirical or logical tests) that it does misrepresent a reality.
-5- A description (e.g. of a unicorn) is fanciful to an actor who believes it represents an imaginary reality
However, it might later turn out to true.
-6- Communication succeeds when the meanings/information in a description are the same when encoded and decoded - near enough.
Communication is a process that conveys a description (and/or other information) from a creator to a user.
The description and communication processes are performed by actors that may be animals or machines.
“Thinking” and “intelligence” includes the ability to create and use descriptions, and more that is out of scope here.
-7- Communication requires that speakers and listeners share the same language for encoding and decoding a description.
A language contains a set of symbols used in the process of creating and using descriptions.
Mostly, we are talking about languages with verbal or graphical symbols, but symbols can also be gestures or even smells.
-8- To share a language, human speakers and listeners must share a lot more.
They largely share same biology, psychology, experience of the world and education.
-9- A description typifies what is described; it attributes general properties or qualities to particular things.
Every description could, potentially, be realised in several realities.
-10- Natural language types are loose, fuzzy and flexible (as Wittgenstein observed).
However, the process of forming a system description involves formalising descriptive types - as follows.
-11- A description may be a singular type (e.g. tasty) or a compound type (hot, tasty, liquid).
A system description, however large and complex, can be seen as a compound type.
-12- A singular description/type is explained in a circular fashion in terms of other descriptive types.
E.g. A “rock” might be described/typified as “dry”, “perceptibly discrete entity”, “solid body” and “mineral material”.
-13- To create a consistent and coherent domain-specific language we must break the circularity by agreeing some basic axioms or base types.
To describe a system, we must create a domain-specific language.
-14- In languages for describing systems, the base types divide along the lines of space and time.
This table illustrates how system elements are typically classified into two types.
Structures in space
Behaviors over time
Space = the volume that things occupy, the addressable places that structural system elements can be found in.
Some structural elements are active, they act; others are passive, they are acted on.
Some elements are ambiguous: transient messages may be persisted in memories.
Time = change, the state changes that system elements experience.
Modern physics suggests time is an illusion, since it can equally well run forwards or backwards.
But we can only understand and communicate about the universe in the forward direction.
Since running forward, brains remember the past and software system state changes record events or their effects.
And running backwards, brains erase memories and software systems undo or remove memories.
The table below is an attempt to help me and readers compare and contrast the terms and concepts therein.
The second and third columns were edited from the three sources below.
· The philosophy book. ISBN 978-1-4053-5329-8
· http://www.hbcse.tifr.res.in/jrmcont/notespart1/node9.html (this may be a dead link)
The first column contains my view, distilled from history of life on earth in this paper The science of system theory.
Since posting the table in 2014 I’ve had many reservations about it.
Some terms are defined differently in other sources and/or have multiple meanings.
Some terms presented as different are arguably not opposites.
Some definitions depend on other terms, such as “existence”, whose meaning is debatable.
And some philosophical positions seem like meaningless babble to me.
In so far as philosophy is about language, knowledge and truth, it seems to have been overtaken by biological and software sciences.
Some philosophical positions
Some different philosophical positions
Matter and energy exists, but is deeply mysterious, beyond our full comprehension.
Our perceptions, descriptions and mental models of material reality also exist in material form – both in mental phenomena and in external representations of them in speech, writing and models of various kinds. (See notes below.)
Idealism: existence is mental or spiritual.
Foerster’s Constructivist Postulate: "Experience is the cause, the world is the consequence."
Materialism: existence is material.
Foerster’s Realist Postulate: "The World is the cause, experience is the consequence."
Contrary to Cartesian dualism, the modern view (cognitive embodiment) sees the mind as a part of the body rather than separable from it.
Cognitive embodiment: mental states and activities are bodily states; the mind is inseparable from the body.
Dualism: views something as made of two parts
Cartesian Dualism: views the mind as standing apart from the body; the mind controls, interacts with and reacts to the body. (After Descartes)
Wisdom is the ability to respond effectively to knowledge.
Knowledge is information that is accurate or true enough to be useful.
Knowledge represents what exists well enough to help us manipulate what exists, and predict its behavior.
Knowledge is acquired in various ways.
The members of a social species necessarily see the world similarly.
Since their ability to perceive and communicate about the world evolved represent it, well enough that they can determine their actions, cooperate socially, and survive..
We humans learn from a mix of
1. perception and experience of real world entities and events
2. logical deduction
3. educational social interaction
Perception, deduction and education are abilities humans acquired though biological evolution, which have enabled our species to understand reality and manipulate it.
Perspectivism, radical constructivism and post-modernism are dangerous ideas in that they tend undermine science and its importance to society.
Empiricism: knowledge is acquired from information obtained from the senses rather from reasoning.
Interpretative: we understand things by perceiving them.
Functionalism: we build mental structures through maturation and interaction with the world.
Cognitive constructivism: knowledge is acquired by creating mental structures in response to experiences. (Piaget)
Social constructivism: knowledge is acquired from social interaction and language usage, and is a shared rather than individual (Prawatt & Floden).
Epistemological Postulate: "He who organises his experience organises the world". The world is unique to each individual.
Radical constructivism: knowledge is acquired from experience, but is not, in any discernible way, an accurate representation of the external world or reality (von Glasersfeld).
Perspectivism: There is no objective truth; knowledge is conditional upon personal perspectives or interests. (Nietzsche)
Rationalism: knowledge is acquired by reason and logical analysis.
Formalism: we understand things by manipulating symbols. E.g. Mathematics does not require the existence of objects or properties.
Information is meaning created or found by an actor in a structure or behavior.
What matters, what must be investigated, is whether speakers and hearers share the same language for encoding and decoding a message. (See notes below.)
The Hermeneutic Principle: "The hearer, not the speaker determines the meaning of an utterance."
The communication principle: Speakers create meanings in utterances; hearers find meanings in utterances; communication succeeds when the created and found meanings are the same.
Data is a structure of matter/energy in which information has been created or found.
Facts are encoded in the data structure by a sender and can be decoded from it by a receiver.
Whether there is some truth in structuralism or not, the human mind is plastic and language is infinitely flexible.
To describe a testable system, an artificial domain-specific language is needed.
Structuralism: we are born with structures that determine how perceptions (phenomena) of concrete things (noumena or a priori objects) are brought together and organised in the mind.
Structuralism in linguistics: language consists of rules that enable speakers to produce an infinite number of sentences. (Wilhelm Wundt (1832-1920) and Chomsky).
At a micro level, the world as we experience it is deterministic.
We can predict the next discernible event - at least in theory.
At a macro level, the world we experience appears indeterminate.
The long term outcomes of events are unpredictable (aka chaotic).
At a psychological and sociological level we have no reasonable or acceptable option but to treat people of sound mind as having free will.
Deterministic: every state and event is the consequence of antecedent states and events. This implies that prediction is possible in theory.
Deterministic automaton: a machine in state Si,
when it receives input Ij,
will go into state Sk and
produce output Ol
(for a finite number of states, inputs and outputs).
Self-determination: choices arise from reasons or desires (regardless of how the processes of choice work).
Indeterministic: a state or event is not wholly the consequence of antecedent states or events. This seems to imply some kind of randomness in state transitions.
Random: haphazard, not-predetermined. In maths it is a measure of how unpredictable a future state or event is.
Chaotic: disorderly. In maths it means behavior in which small differences in an initial state or event yield widely diverging outcomes (even though the system is deterministic, with no random elements). This makes long-term prediction impossible.
Both holist and reductionist views of a system are important and helpful different times. Enterprise architecture is deprecated by some “systems thinkers” as being reductionist. The implication is that other kinds of “systems thinking” are better for being purely holistic. In practice, both enterprise architects and systems thinkers take both views of systems.
Holism: treats a system’s parts as inseparable. The properties of the whole system are not the properties of any part. These “emergent properties” emerge only from the interaction between parts
Reductionism: explains the properties of one thing by the properties of another (lower level) thing. Or else, ignores the higher thing in favour of discussing the lower thing(s).