Enterprise Architecture Cybernetics and the Edge of Chaos:
Sustaining Enterprises as Complex Systems in Complex
Business Environments
Author
Kandjani, Hadi, Bernus, Peter, Nielsen, Sue
Published
2013
Conference Title
Proceedings of the 46th Hawaii International Conference on Systems Science, HICSS' 2013
DOI
https://doi.org/10.1109/HICSS.2013.199
Copyright Statement
© 2013 IEEE. Personal use of this material is permitted. Permission from IEEE must be
obtained for all other uses, in any current or future media, including reprinting/republishing this
material for advertising or promotional purposes, creating new collective works, for resale or
redistribution to servers or lists, or reuse of any copyrighted component of this work in other
works.
Downloaded from
http://hdl.handle.net/10072/53441
Griffith Research Online
https://research-repository.griffith.edu.au
Enterprise Architecture Cybernetics and the Edge of Chaos: Sustaining
Enterprises as Complex Systems in Complex Business Environments
Hadi Kandjani
Centre for Enterprise
Architecture Research and
Management (CEARM), School
of ICT, Griffith University,
Brisbane, Australia
H.Kandjani@grifith.edu.au
Peter Bernus
Centre for Enterprise
Architecture Research and
Management (CEARM), School
of ICT, Griffith University,
Brisbane, Australia
P.Bernus@griffith.edu.au
Sue Nielsen
Institute for Integrated and
Intelligent Systems (IIIS),
School of ICT,
Griffith University,
Brisbane, Australia
S.Nielsen@griffith.edu.au
Abstract
The purpose of this paper is primarily theoretical –
to propose and detail a model of system evolution, and
show its derivation from the fields of Enterprise
Architecture, cybernetics and systems theory.
Cybernetic thinking is used to develop a ‘Co-evolution
Path Model’ to explain how enterprises co-evolve with
their environments. The model is re-interpreting
Stafford Beer’s Viable System Model, and also uses
Conant and Ashby's theorem of the 'good regulator',
exemplifying how various complexity management
theories could be synthesised into a cybernetic theory
of Enterprise Architecture – informing management of
mechanisms to maintain harmony between the
evolution of the enterprise as a system and the
evolution of its environment.
1. Introduction
The increasing complexity of the IT and business
environment, and the need to ensure alignment of IT
with business goals and operations, have given rise to a
number of initiatives in information systems research
and practice [1]. Prominent amongst these is the
discipline of Enterprise Architecture which is now
widely accepted as a requirement for high level and
comprehensive management of the IT enterprise [2].
Despite this acceptance, the field of Enterprise
Architecture is still undergoing investigation into its
theoretical basis, with considerable work focused on
elaborating and harmonising the various frameworks
and models. This paper aims to contribute to this work
by exploring the application of cybernetic thinking to
explain how systems co-evolve with their
environments. A ‘Co-evolution Path Model’ is
developed which reinterprets ‘System 5’ of Beer’s
Viable System Model [3], i.e., the system which is
responsible for strategically steering the organisation.
The paper uses the GERAM framework [4, 34, 35]
as a basis for the model, because of it ‘agnostic’ nature
and its important concepts of Life History and Life
Cycle. GERAM defines a comprehensive set of
concepts to represent and explore enterprise evolution.
GERAM is a “toolkit of concepts for designing and
maintaining enterprises for their entire life history”
(ibid) and the objective of this framework is “to
systematise various contributions of the field that
address the creation and sustenance through life of the
enterprise as a complex system”. GERAM is different
from other frameworks (those developed for only
pragmatic purposes), as pragmatic frameworks do not
necessarily have to make certain theoretically
important differentiations, nor have to be complete
from every respect to be usable for some particular
practical EA project. However, fundamental
conceptual differentiations and domain completeness
are needed when it comes to the use of an EA
framework to interpret models of evolution and
management in general, so as to create a foundation of
what we call ‘EA Cybernetics’. While many
frameworks could be made complete by satisfying all
ISO15704 requirements [36, 37], we use GERAM’s
concepts as they already have the desired properties.
The purpose of this paper is primarily theoretical –
to propose and detail a model and show its derivation
from the fields of Enterprise Architecture, Cybernetics
and Systems Theory. Since the development of this
model is in its early stages, it has yet to be tested in
empirical studies. However, it proposes both “testable
propositions and causal explanations” [5] which may
be applied to real cases.
In practical terms, such a model might enable
organisations to recognise the signs of dissonance
between system complexity and environment
complexity and as a result make deliberate decisions to
steer away both from system states that are the edge of
(perceived) chaos and trends to obsolescence. The
paper also proposes the development of an EA
cybernetics framework which can equally represent
evolutionary and deliberate/designed changes.
Note that there are two forces at work: a system
needs to be able to display complex enough behaviour
to respond to the needs of (or survive in) the
environment; at the same time excessive complexity
makes the system hard to control. Part of the question
is how complexity can be measured and how excessive
complexity can be eliminated. One important
complexity measure is the system’s information
content, as defined by the system’s Kolmogorov
complexity (KC) [44]. For example, the authors
previously demonstrated [45], how known
approximation methods of KC can be used to estimate
the information content of a system and how this can
be utilised to make architectural design decisions to
‘design out’ excessive complexity from a supply chain.
2. Complexity and the Cybernetics
Perspective
Enterprises are best understood as intrinsically
complex adaptive living systems: they can not purely
be considered as ‘designed systems’, because
deliberate design/control episodes and processes
(‘enterprise engineering’, using models) are intermixed
with emergent change episodes and processes (that
may perhaps be explained by models). The mix of
deliberate and emerging processes can create a
situation in which the enterprise as a system is in
dynamic equilibrium (for some stretch of time) – a
property studied in General Systems Theory [6].
The evolution of the enterprise (or enterprises,
networks, industries, the entire economy, society, etc)
includes the emergent as well as the deliberate aspects
of system change, therefore we believe that EA needs
to interpret previous research in both. This is
summarised as the main aim of the enterprise
architecture discipline and practice, i.e., to explain
change in enterprises as complex systems (through
theory, models and methodologies) [7].
In response to the problem of managing complexity
and fast change, many studies applied the cybernetic
perspective to Enterprise Architecture (application of
cybernetic concepts to EA management [8] and to EA
principles e.g. as embodied in TOGAF [9]).
Stafford Beer believed that the dynamics of
enterprises is about “the manipulation of men, material,
machinery and money: the four M’s”, but also about a
fundamental manipulation of systems, which we call
the “management of complexity” [10,3].
Norbert Wiener defined cybernetics as “the science
of control and communication in the animal and
machine” [11]. Ashby also calls it as the art of
“steermanship”, the study of coordination, regulation
and control of systems, and argued that “truths of
cybernetics are not conditional on their being derived
from some other branch of science” [12].
Therefore the field embraces a set of self-contained
groundings and foundations [12]. Ashby addressed the
complexity of a system as one of the peculiarities of
cybernetics and indicated that cybernetics must
prescribe a scientific method of dealing with
complexity as a critical attribute of a system.
Beer was the first person applying cybernetics to
management and defined cybernetics as “the science of
effective organisation” [13,14] . He was also first to
coin the word “Management Cybernetics” – a field
applying cybernetic laws and theories to all types of
socio-technical organisations / “enterprises” [15].
Beer elaborates on the relevance of cybernetics to
management in ‘Cybernetics and Management’ [13]
and describes his first discoveries and promises in the
management discipline. He also characterises
cybernetics as “the science of control” and
management as “the profession of control” [10].
Therefore EA research has acknowledged the
relevance of cybernetics for modern enterprises which
cannot expect to build ‘ideal’ and one-time systems but
must undertake continuous steering and control of their
evolving systems [16]. Such a perspective elaborates
on Beer’s ‘system 4 and 5’ to cope with the increasing
complexity of organisations and their environments.
3. Enterprise Architecture Cybernetics
One common topic of cybernetics is the treatment
of complexity (whether it is the complexity of the
structure, behaviour, control, management, or other
relevant view of the system of interest), raising the
question how systems can be managed, controlled,
changed, designed, or partially influenced for
producing emergent adaptive behaviour.
A distinct problem, characteristic of complex
systems, is (by definition of what constitutes a complex
system) that none of these tasks can be based on a
complete predictive model, therefore the involved
decisions must be based on incomplete information.
Due to this character of complex systems we need
theories and methods, or structures, that produce such
self-control behaviour (either in deliberate or in
emergent way). Whichever way this control is
exercised, it should be able to be described by an
external observer as ‘partial control’ that nevertheless
achieves a set of valued system properties (such as
sustainability, viability, availability, and so on).
For the above reason, any controller (on any level
of a system that is characterised as complex on that
level) only has, or can only have, an incomplete model
of the system, and sees the system through this model
to make decisions to control that system. The
complexity of a model like this is the ‘apparent
complexity’ of the system from the given controller’s
(manager’s) point of view.
Checkland warns that theories, frameworks and
models with an excessive level of abstraction and
general systems principles of ‘wholeness’ could be in
danger of not being able to deal with real practice [17].
At the same time there exist very specific and context-
dependent theories, frameworks and models which
sacrifice generality and abstraction, on the other hand
often there is little guidance on the limits of their
applicability. The optimum degree of generality is
somewhere in-between with different levels of
abstraction for each purpose. For example, the aim of
General Systems Theory [6] is not achievable by a
single science discipline in isolation [18].
In order to develop a model which may explain
how systems co-evolve with their environments, we
have adopted fundamental concepts of the Generalised
Enterprise Reference Architecture and Methodology
(GERAM) framework [4].
EA frameworks such as GERAM acknowledge that
the optimum degree of generality is problem domain
dependent, therefore it is necessary to provide a
modelling framework that represents this continuum
from the most generic to the most specific.
GERAM defines a) Generic Enterprise Modelling
Concepts (GEMCs) [practically ontological theories],
b) Partial Enterprise Models (PEMs) [usually in form
of reference models] and c) Particular Enterprise
Models (EMs). GEMCs define and formalise the most
generic concepts of enterprise modelling, PEMs
capture characteristics common to categories of
enterprise entities within or across one or more
industry sectors, and particular enterprise Models
(EMs) that represent a particular enterprise entity [4].
EA Cybernetics must maintain an ‘optimum degree of
generality’ to provide the discipline and practice with
the ‘right level of abstraction’ for each purpose,
whereupon given the abstract theory and a concrete
system (and concrete problem), there should exist
methods that can be used to solve or explain the
problem, and achieve this within the limitations of
available resource- and time constraints.
Enterprise Architecture, as a developing discipline,
needs a model for theory development, testing and
knowledge creation. Anderton and Checkland [19]
developed a model of any developing discipline to
demonstrate the cyclic interaction between theory
development and formulation for a problem and theory
testing [19,20]. For EA to be a developing discipline
(Fig.1), we consider real world enterprise problem
domains as the source of a development process, a
source of issues to be addressed by theories, models
and methods in enterprise related disciplines.
EA
CYBERNETI
C CASE
RECORDS
EA CYBERNETIC
METHODOLOGY
UNIFIED CYBERNETIC
THEORIES OF EA
FORMALISED ENTERPRISE
PROBLEMS DOMAINS
EA CYBERNETICS
Theories, Models &
Methods
Gives rise to
Are harmonised, formalised,
synthesized and systematised
by
Which may be used
to develop
Which may be represented
using
Which produce
Contribute to
ENTE
RPRI
SE
RELA
TED
DISCI
PLIN
ES
REAL-
WORLD
ENTERP
RISE
PROBLE
M
DOMAIN
S
Addresses individually by
To be used in
EA Practice
(intervention,
influence,
observation)
Which support
criticism
of
ISSUES
Provide
Which support
criticism
of
Figure 1. Enterprise Architecture as a Developing
Discipline based on the model of activities and
results of developing disciplines [19,20]
These will shape ideas by which two types of
theories could be developed [20]:
a) substantive theories derived from related
disciplines to apply relevant models, theories and
methods in the problem domains, and
b) methodological theories about how to
individually apply enterprise related disciplines in the
problem domains.
Once we developed such theories, it is possible to
state problems – not only existing problems in concrete
enterprise problem domains, but also formalised,
harmonised and synthesised problem statements by EA
cybernetics within a new theory.
As a new theory, EA cybernetics produces
formalised enterprise problem domains which may be
represented using the unified cybernetic theory of EA.
These unified theories may be used to develop a
methodology(ies) to be used in EA practice.
Results of such synthesis must be tested in practice
(through intervention, influence, or observation) to
create ‘case records’, which in turn provide the source
of criticism allowing better theories to be formulated
(and better models, techniques, and methodologies).
The application of the latter methodologies should be
documented in case records which provide feedback to
improve the individual- and the unified theories.
The EA discipline not only embraces the models,
methods and theories of management and control – it
also uses the same of systems engineering, linguistics,
cognitive science, environmental science, biology,
social science and artificial intelligence.
What cybernetic thinking is able to do is to provide
a method of unifying (and relating) the contribution of
these disciplines: cybernetic thinking can be used to
represent the essence of multiple theories using
abstract functions and processes (and meta-processes)
and their relationships, rules and axioms (likely to be
expressed in suitably selected logics).
Fig.1., shows the pathway through which the apport
of these disciplines is formalised, synthesised,
harmonised, systematised and eventually represented
as a unified Cybernetic theory of EA. The Co-
evolution Path Model introduced in the next sections is
an example of a cybernetic model of the control and
management of viable complex systems that operate in
complex environments.
4. The Co-evolution Path Model: Dynamic
Homeostasis vs. Dynamic Hetereostasis:
An Example of an EA Cybernetics Model
A key property of a viable system and a “measure
of its submission to the control mechanism” is its
ability to maintain homoeostasis, defined by Beer [10]
as “constancy of some critical variable (output)”.
In our model of co-evolution, we demonstrate the
dynamic sustenance of requisite variety based on
Ashby’s law: "only variety can destroy variety” [12],
paraphrased by Beer [21] as "variety absorbs variety",
where ‘variety’ is the number of possible states of a
system [22], or as recently clarified as the number of
relevant states of a system [23].
In order for a system to dynamically achieve and
maintain requisite variety and to be in dynamic
equilibrium, the system requires communication
channels and feedback loops. These communication
channels serve as self-perpetuating mechanism and
include both attenuation and amplification
mechanisms. (Note that for the discussion below what
we call a ‘system’ includes the system’s controller.)
Considering the system and its environment as two
coupled entities, if one component is perturbed, the
effect of that perturbation on the other component is
either amplified through positive feedback, or may be
reversed (attenuated) through negative feedback.
Dynamic Homeostasis:
Sustaining Requisite Variety
Dynamic Heterostasis:
Oscillating Requisite Variety
Co-evolution of the System with its Environment
through First and Second Feedback loops
C
S
= C
E
Static
Homeostasis
C
S
> C
E
Amplification
Mechanism
C
S’
= C
E’
Co-evolution
C
S’
< C
E’
Attenuation
Mechanism
C
S”
= C
E”
Co-evolution
C
S’
> C
E’
Amplification
Mechanism
C
S
< C
E
Attenuation
Mechanism
Complexity of the Environment (C
E
)
Complexity of the System (C
S
)
Figure 2. The Co-evolution Path Model
The role of a negative feedback loop is to reverse
the effect of the initial perturbation and restore system
homeostasis (in which critical variables are stable)
while positive feedback can create unstable states [24].
We observe that both a system and its environment
(including systems in that environment) evolve, and
this can create imbalance between the requisite variety
(maintained by the controller) of our system of interest
and the variety that would be required for it to maintain
homeostasis. In other words, systems that want to live
long must co-evolve with their environment.
Formally: we consider the environment an entity
with a possible set of observable states and if two such
states require different response from the system then
the system must be able to differentiate between them
(thus they are two different relevant states). (Note: we
may not be able to describe the environment as a
system, although it may contain one or more systems.)
Consequently, in Fig. 2, the complexity of a system
(CS) is defined to be the complexity of the model the
controller of the system maintains (appears to be
maintaining) so as to manage the system’s operations –
Including the need to interact with the environment.
The complexity of the system’s environment (CE)
is a relative notion and is defined to be the complexity
of the model of the environment that the controller of
the system would need in order to maintain the
system’s homeostasis; – although it is sufficient if in
the eyes of an external observer the system’s controller
appears to be maintaining such model.
Specifically, an ‘environment model’ must have
predictive capability, so that the system, while
interacting with its environment, can maintain a
![Figure 1. Enterprise Architecture as a Developing Discipline based on the model of activities and results of developing disciplines [19,20]](/figures/figure-1-enterprise-architecture-as-a-developing-discipline-24emfdkn.webp)
