Change:
adaptation and evolution - analysis
Copyright 2017 Graham Berrisford. One of about
300 papers at http://avancier.website. Last updated 06/01/2019
15:44
THIS PAPER IS
AN OLD ANALYSIS
I DON’T
REJECT IT.
BUT I NOW
PREFER THE DIFFERENT PERSPECTIVE IN THIS
PAPER.
This paper presents enterprise architecture as a meta system for system evolution.
Contents
Evolution of a system from one generation
to the next (2nd order cybernetics)
Discrete step
evolutionary change
A
discussion of “structural change”
An individual can, following given rules,
can adapt to fluctuations in the state of its environment.
By contrast, a species can evolve to fit more fundamental changes in its environment.
Evolution (as opposed to adaptation)
changes the very nature or rules of a system.
It can change not only how it behaves
(processes) but even what it is made of (components).
Environment |
Friends, enemies and resources v
^ |
Evolution |
Parents Children v
^ |
Reproduction according
to “fitness” |
Biological organisms evolve by accident, without direction or design.
Human activity systems do the same, but that accidental change is outside our scope for now.
Second order cybernetics introduced the “observer” into systems thinking.
An observer (inside or outside a system) can envisage changes to the system, and change it.
Environment |
Business strategy and operations v
^ |
Enterprise
architecture |
Change requests New systems v
^ |
System design processes performed
by architects |
|
according
to business objectives principles,
policies, patterns and road
maps |
You cannot describe the future systems
that continual, undirected, evolution will bring.
You can however describe the meta system that changes one generation to
the next.
Darwin described the process of natural
selection by which biological organisms evolve.
Note that natural selection is not a
property of any individual organism, of any parent or child.
Each
system generation has a life time from birth to death.
An enterprise architecture framework is a meta system for changing business systems.
E.g.
TOGAF encourages architects to describe the baseline system, the target system,
and transition system states in between.
Each system generation has a life time from when it starts working to when it disappears or stops working.
TOGAF is recursive in that it modifies itself (the meta system) in a Preliminary Phase, before enterprise architects get to work.
The
meta system is not part of the baseline or target
operational systems that are described in the course of any particular
enterprise architecture cycle.
The term discrete is defined in our system theory dictionary thus:
Discrete: |
not
continuous, dividing a whole or a continuum into distinct or separate things,
entities, events or states. |
The universe is an ever unfolding process in which systems are discrete, if transient, islands of stability.
Where a system changes in discrete steps, then each system generation can be described as per general system theory.
But where changes to the structure
or behaviour of things are seen as continuous, there is never a system to be
described.
The table below outlines continuous and discrete evolution, and hybrid called semi-controlled evolution.
Social group type |
Evolution types classified
according to rules |
|
Ordered |
Discrete evolution: (e.g. computer activity system) all rules are fixed (even though they may be configurable) Every rule change requires a new generation. |
|
Ordered/chaotic hybrid |
Semi-controlled: (e.g. a human activity system) Some rule changes require a new system generation. System actors can change other rules. |
Goal-driven: system actors are
constrained by fixed goals Unguided: system actors please themselves. |
Chaotic |
Continuous evolution: no rules are fixed and there are no system generations |
What is "structural change", what is the motivation for it, and how is it relevant to EA?
Reader: I take ‘structural change’ to mean organisational changes to formalise roles, responsibilities or activities.
Graham: The term "structure" is used ambiguously in systems thinking and system theory.
A business has a human authority/communication structure (usually a hierarchy).
Business processes are performed by actors connected in a network structure.
The two structures may change independently of each other.
Surely it would be better to distinguish sociology from system theory?
A systems thinker has to choose their viewpoint: actor-oriented or activity-oriented?
Social hiearchology and EA are different things; EA teams do not define human management structures.
And surely it would better to distinguish structure from behaviour?
Holistically, it doesn't matter how the structure or roles of a system change, provided the overall behaviour is the same.
Always, the operation of a system costs money; and changing the structure or behaviour of a system consumes time and money.
"Refactoring" - changing a structure without changing behaviour - is relatively cheap kinds of change.
And shuffling the management deck chairs is peripheral to EA.
The universe or a business is a continuum, with no separation, until separations are observed or envisaged by actors.
Separations appear only descriptions or models that carve the continuum up into discrete things.
Separations are only realised in so far as working actors follow the rules in those system descriptions.
Given that divisions have been drawn and established between systems, EA is concerned to address cross-system impacts.
Reader: It is the interaction of structures in the operation of the business that determine the architecture.
Graham: That is true at every level of granularity.
Within a system, its subsystems cooperate in the system's architecture.
More widely, systems cooperate (now as subsystems) in a wider system's architecture.
Reader: EA should be managing the communication boundaries of structures.
Graham: Again, “structures” is ambiguous.
EA does tend to focus on the interfaces of coarser-grained systems rather than finer-grained systems.
Before that, those systems, those boundaries, have to be defined somewhere, by somebody.
Reorganising a human management hierarchy is one thing.
Reorganising how actors cooperate in the processes of a system is another.
The two structures may change independently of each other.