Complex adaptive systems?
Copyright 2017 Graham Berrisford. One of about 300 papers at http://avancier.website. Last updated 04/01/2019 23:30
System Theory Tutorial in London Saturday March 2nd 2019
General system theory was supposed to unify the sciences.
Today, much systems thinking discussion is not general – it is unique to human society
And some gives science a bad name, especially in discussions that refer to “complex systems”.
This paper is about the use of the term “complex adaptive system” in much social systems thinking discussion.
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 commonly-referred-to chart of “complexity science” can be found here http://www.art-sciencefactory.com/complexity-map_feb09.html
The chart is misleading chronologically and conceptually.
Chronologically, for a different view, read papers before this one.
1. The science & philosophy of systems
2. Thinkers before systems thinking
3. System thinkers & their ideas.
Conceptually, the chart mixes up science, pseudo science and metaphysics.
It mixes up scientists and people who make assertions and classify things with no empirical validation.
It includes people (e.g. Parsons and Luhmann) whose ideas are metaphysical – cannot be verified or disproved.
And note one misleading implication: the line from first to second order cybernetics suggests a progression where there is rather a schismatic difference.
Systems thinkers often speak of a social entity or organisation as a complex adaptive system.
What does the term mean? Is it merely a pseudo-scientific label for a human society?
Or does it have a more general and scientific meaning?
In which case, what is true of a complex adaptive system that is not true of a simple activity system, or true in classical cybernetics?
MIT offer this definition.
"Complex Adaptive Systems are dynamic systems able to adapt in and evolve with a changing environment.
It is important to realize that there is no separation between a system and its environment in the idea that a system always adapts to a changing environment.
Rather, the concept to be examined is that of a system closely linked with all other related systems making up an ecosystem.
Within such a context, change needs to be seen in terms of co-evolution with all other related systems, rather than as adaptation to a separate and distinct environment."
This definition doesn’t tell us what a “system” is, how to distinguish it from any other kind of entity in the universe.
It uses the term “adaptive” in a way that may be interpreted to mean different things.
And it gives us no hint as to why or how any system, adaptive or not, should be considered “complex”.
"Complex Adaptive Systems [CAS] are dynamic systems”
The term dynamic may be read as changing or changeable.
In both CAS and classical cybernetics, systems are dynamic in the sense they change state in response to external events.
Also, they are dynamic in the very different sense that they can change via generational system mutation.
“able to adapt in and evolve with a changing environment.”
The terms adapt and evolve refer to changes inside the system.
In both CAS and classical cybernetics, a system is connected to its environment by an information feedback loop.
That loop enables the system and its environment (or external actors in it) to change state in step with each other.
Also, in a second and very different sense of change, a system may change via generational system mutation.
System mutation may happen in these three ways.
1. By self-replication with changes - as a biological system evolves with each new generation.
2. By redesign by actors outside the system - as a machine or software system is changed in each new version.
3. By redesign by actors who play roles inside the system - as a human social system may evolve in each new version.
Probably, the MIT definition is referring to the third kind of mutation.
However, human social systems might instead evolve in the first two ways above – are those to be ignored?
Moreover, it is unclear whether MIT mean generational change or continuous change, which is very different.
Because if a social group changes continually, stuff happens, but it can never be described and tested as a system.
So to call it a system is an empty assertion.
“there is no separation between a system and its environment”
The term separation might be misleading.
In both CAS and classical cybernetics, a system is connected to its environment.
They are connected logically in description, and physically in reality, by an information feedback loop.
At the same time, you must be able (at least logically) to separate what is inside the system from what is outside.
Else the system has no boundary, and you cannot distinguish it from its environment.
“a system always adapts to a changing environment.”
In both CAS and classical cybernetics, a system adapts by changing state in response to events it detects in its environment.
And, in a very different sense of adaptation, a system may evolve via generational system mutation.
Surely, what MIT mean to imply is that the system adapts, rather than the system is adapted?
In other words, the actors who play roles in the system are the ones who change its roles and rules?
“the concept to be examined is that of a system closely linked with all other related systems making up an ecosystem.”
In both CAS and classical cybernetics, the environment outside any one system of interest may be divided into separately describable systems.
And all those systems may be seen as subsystems of a wider system (an eco system if you like).
Is this not as true of simple systems as complex adaptive systems?
“change needs to be seen in terms of co-evolution with all other related systems, rather than as adaptation to a separate and distinct environment."
In both CAS and classical cybernetics, if one system changes state, then related systems may be triggered to change state.
Or if one system mutates, then related systems may be have to mutate also, if they are to survive.
Is this not as true of simple systems as complex adaptive systems?
So what differentiates MIT’s complex adaptive system from other systems?
Surely, primarily, that it mutates of its own accord, from within?
And isn’t that true of every group of human actors who communicate with each other?
Every family, social club, business (say IBM) or other social network – whether it is describable and testable as a system or not?
IBM might well be called a complex adaptive social network.
After all, the network of actors employed by IBM is reasonably called large and complex.
And those actors adapt what IBM does in response to changes in its environment – and systems outside of IBM.
But what does it mean to assert that IBM is “a system”?
How measure its adaptiveness? And to how measure its complexity?
Regarding complexity, this definition from another source introduces the concept of non-linear or chaotic behavior.
“A complex adaptive system exhibits both linear and non-linear behavior.
As per chaos theory, incremental changes to one of its state variables can turn linear quantitative change into qualitative change.
There comes a point where the system behaves in a qualitatively different way.
As per catastrophe theory, the topological shape of the system state space may change dramatically.” Source lost
The trouble is, simple systems fit this definition also.
As you turn a simple tap, the stream of water running out changes from orderly to chaotic.
The term system is widely used, often loosely, and with various meanings.
At its most vacuous, it means only "some things that are in some way connected”.
Or in a sociological context, it may mean only “some people who talk to each other, or cooperate in some way”.
However, classical system theory is used ubiquitously in the design of everyday systems.
And the more classical and general concept of a system was defined in previous papers.
A system exhibits regular or repeatable activities, it behaves in orderly or deterministic way.
An abstract system is a description or model that conforms to the principles of cybernetics or system dynamics.
A concrete system is a network in which actors realise the roles and rules of an abstract system.
A network is a structure in which actors are related and communicate with each other.
This paper recasts these terms for discussion of societies.
Ashby defined a system as an abstraction from the infinite complexity of any entity that realises it.
A system may be characterised as a set of roles and rules (e.g. the mating ritual of a pair of sticklebacks).
When those abstract roles and rules are realised by a concrete entity, which behaves as described, we have a concrete system.
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
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.
On complex/concrete system versus complicated/abstract system
In Ashby’s view, a system is abstraction from, and infinitely simpler than, any concrete entity that realises the system.
When systems thinkers contrast complicated systems with complex systems, they are often really comparing abstract and concrete systems
E.g. they are contrasting a simple social system (say exchanging Christmas cards) with the complexity of a human social network that realises the system.
Or contrasting a simple business system (say order > invoice > payment) with the complexity of a real world business that realises the system.
On social network versus social system
In Ashby’s view, a social network is only a system in so far as it performs the behaviors in a given system description.
When systems thinkers speak of a complex adaptive system, they are usually thinking of a named human organization or institution.
They are thinking of social network, perhaps as defined in an organisation structure, or as derived from observations of inter-actor communications.
But such a social network can be described from different viewpoints as realising countless different, even conflicting, systems.
E.g. IBM may reasonably be described as a complex, adaptive, self-organising social network.
At the same time, IBM realises (instantiates, manifests) countless describable and testable systems.
Of those countless systems, some are social, some are technological and some are socio-technical.
Some are complex, others are simple; some are adaptable, others are inflexible; some are cooperative, others are in conflict.
Any social network might adapt its behavior by changing its activities and aims, and changing the roles and rules of any system it realises.
But to adapt a system would be to change its roles and rules, and so change into another system.
And if its actors may change a system continually (rather than incrementally), this undermines the very concept of a system.
The term complex widely used, often loosely, and with various meanings.
There is no agreed way measure to complexity.
Which is more complex, IBM or a hen’s egg?
To measure the complexity of one thing, you need a description of its parts.
And to compare the complexity of two things, those parts must described comparably, at the same level of abstraction.
At the top most level, the organisation structure of IBM is simple.
A description of IBM that included every human actor in it would be complex beyond imagination.
And a description of a hen’s egg that included every atomic particle would also be complex beyond imagination.
Often, an entity is called a “complex adaptive system” where one or more of the following are true.
· No measure of complexity has been agreed
· No level of abstraction has been agreed
· No quantifiable properties are described, which makes any measure of complexity impossible.
· No description of the entity as a system has been agreed, or even made.
· No description is possible, because the entity changes continually, rather than generationally.
Some naively presume that the complexity of an organisation lies in its management structure.
Or (after Ashby and Beer) a system’s complexity is defined by the number of variables it maintains.
Our paper on complexity measures points out that one ought to measure the complexity of behaviors as well as structures.
Complex = unpredictable or chaotic behavior?
The Plexus Institute glossary at https://plexusinstitute.org/ says:
"complexity is found in systems when there are unpredictable interactions of multiple participants and components across many levels of the system."
Yet chaos theory taught us even the simplest of systems can be unpredictable.
And unfortunately, the glossary contains no definition of "system", let alone what the "levels" of a system are.
By complex, systems thinkers often mean the system’s behavior changes the state of world in a non-linear or chaotic fashion.
But the concepts of complexity and chaos ought to be distinguished.
The roles and rules of a system may be very simple or very complex
A complex system can behave in a linear or predictable way over time.
A simple system can behave in a non-linear of chaotic way over time. (See the papers on System Dynamics.)
A system’s behavior may switch from linear change to non-linear change or vice-versa.
The term adapt is widely used with two different meanings.
Remember Ashby insisted we should on no account confuse these two kinds of change.
System state change
the value of at least one state variable
homeostatic regulation of values to stay within a desired range
the type of at least one variable or behavior.
re-organization changing the variables or the rules that update their values.
Classical cybernetics is about maintaining the values of defined state variables.
Ashby used the term adaptation in the context of homeostatic state change.
Systems thinkers often use the term adaptive in the second – mutation – sense
And moreover, they mean the system is self-organising.
Second-order cybernetics was developed around 1970 by Margaret Mead, Heinz von Foerster and others.
It is about self-organising systems; it is the recursive application of cybernetics to itself.
It allows systems actors to be system thinkers, who re-organise themselves.
It allows actors in a system to study the system and change it.
Actors not only play roles in a system, but also observe and change the roles, rules and state variables of that system.
What if actors may change a system continually, rather than incrementally, generation by generation?
This undermines the very concept of a system.
It does appear the term complex adaptive system is widely used as a pseudo-scientific label for a human society or organisation.
Many social systems thinkers use words that are drawn from harder science, or sound as though they are.
Some speak of self-organising systems.
If the actors in a social network change the roles or rules of a system they realise, that may well be called self-organization
But if the actors make changes continually (rather than incrementally) then there is never any describable or testable “organization”.
The term complexity theory is odd, given there is no widely agreed measure of complexity.
By complex, people often mean a system has non-linear dynamics, or changes the state of world in a chaotic fashion.
But the equation of complex and chaotic is misleading.
Even a very simple orderly system can produce chaotic results over time - as may be revealed by System Dynamics.
Some systems thinkers’ terms relate to patterns revealed by System Dynamics - like strange attractors and fractal geometry.
Other terms often used more poetically than scientifically include emergent properties and entropy.
And some thinkers, deliberately, radically alter the meanings of terms used in science.
E.g. Luhmann’s metaphysical autopoietic social system radically differs from an autopoietic biological organism.
Read Social networks versus social systems for an answer to that question.
How to extend system theory to embrace “self-organisation”?
Read System stability and change for an answer to that question.
All free-to-read materials at http://avancier.website 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 avancier.website in whichever social media you use..