Documente Academic
Documente Profesional
Documente Cultură
ISSN 1400-1179
ISRN/KTH/MMK/R-05/10-SE
MMK
Technical Report
Stockholm
2005 Mechatronics Lab, Department of Machine Design
Royal Institute of Technology, KTH
S-100 44 STOCKHOLM
Technical Report
TRITA-MMK 2005:10
ISSN 1400-1179
ISRN/KTH/MMK/R-05/10-SE
MMK
Views on General System Theory
Machine Design
KTH
Mechatronics Lab Ola Larses and Jad El-khoury
{olal;jad@md.kth.se}
April 2005
Abstract
This report is the result of a literature study course work on the General System Theory (GST)
performed by Jad Elkhoury and Ola Larses at the Mechatronics Division of the Department of
Machine Design at the Royal Institute of Technology (KTH). The study was initially
performed in the fall of 2004 and concluded in the spring 2005.
The structure of the report consists of two major parts. The first part provides a general
overview of this broad field of science. It gives a short summary and overview of the General
System Theory is, as well as a reflection on how GST relates to current meta-modelling
efforts exemplified with the UML-MOF. The second major part of the report is a set of four
book reviews covering very different books about the area: One original work of von
Bertalanffy (1967), two books from authors providing their views on GST (Weinberg 2001,
Checkland 1999) and one text book covering several theories on the subject (Skyttner 2001).
Keywords
GST, Cybernetics, General System Theory, Skyttner, Bertalanffy, Checkland, Weinberg
Ola Larses and Jad El-khoury
Views on General System Theory
Contents
1 INTRODUCTION ....................................................................................................................................... 5
2 A SUMMARY OF THE FIELD OF GENERAL SYSTEM THEORY (GST)....................................... 5
2.1 HISTORY OF GENERAL SYSTEMS THEORY (GST) ................................................................................. 5
2.1.1 Current references, communities and courses ................................................................................ 7
2.2 CONCEPTS OF GST ............................................................................................................................... 9
2.3 A GENERAL SYSTEM MODEL - RELATING THE CONCEPTS OF GST ..................................................... 11
2.3.1 A Criticism of Bertalanffy ............................................................................................................. 12
2.3.2 The Analysis-Synthesis system method.......................................................................................... 12
2.3.3 The goal orientation of systems..................................................................................................... 13
3 GST AND THE MOF................................................................................................................................ 14
4 REFERENCES .......................................................................................................................................... 16
5 APPENDIX – BOOK SUMMARIES....................................................................................................... 17
5.1 BOOK REVIEWS ................................................................................................................................... 17
5.1.1 General Systems Theory – Lars Skyttner (2001) ........................................................................... 18
5.1.2 General System Theory – Ludwig Von Bertalanffy (1968)............................................................ 21
5.1.3 Systems Thinking, Systems Practice – Peter Checkland (1981/1999)........................................... 24
5.2 EXTENDED BOOK REVIEW: AN INTRODUCTION TO GENERAL SYSTEMS THINKING – GERALD M.
WEINBERG ........................................................................................................................................................ 28
5.2.1 The Problem .................................................................................................................................. 28
5.2.2 The Approach ................................................................................................................................ 30
5.2.3 System and Illusion........................................................................................................................ 33
5.2.4 Interpreting Observations ............................................................................................................. 37
5.2.5 Breaking down Observations ........................................................................................................ 41
5.2.6 Describing Behaviour ................................................................................................................... 45
5.2.7 Some Systems Questions................................................................................................................ 47
5.2.8 Further readings ........................................................................................................................... 48
3
Ola Larses and Jad El-khoury
Views on General System Theory
4
Ola Larses and Jad El-khoury
Views on General System Theory
1 Introduction
This report is the result of a literature study course work on the General System Theory (GST)
performed by Jad Elkhoury and Ola Larses at the Mechatronics Division of the Department of
Machine Design at the Royal Institute of Technology (KTH). The study was initially
performed in the fall of 2004 and concluded in the spring 2005.
The structure of the report consists of two major parts. The first part provides a general
overview of this broad field of science. It gives a short summary and overview of the General
System Theory is, as well as a reflection on how GST relates to current meta-modelling
efforts exemplified with the UML-MOF. The second major part of the report is a set of four
book reviews covering very different books about the area: One original work of von
Bertalanffy (1967), two books from authors providing their views on GST (Weinberg 2001,
Checkland 1999) and one text book covering several theories on the subject (Skyttner 2001).
3.2 Problem-solving application of systems thinking 3.1 Theoretical development of systems thinking
(formulation of GST)
5
Ola Larses and Jad El-khoury
Views on General System Theory
of GST is Ludwig von Bertalanffy, there are however a range of contemporary scientists who
contributed in the field. GST have strong bonds to Cybernetics, Information theory and
Control theory.
Bertalanffy founded the “Society for the Advancement of General Systems Theory” together
with Kenneth Boulding in 1954. The founders emphasized their desire to promote the unity of
science at the very first meeting which took place in December, 1954 in Berkeley, California.
In 1956 the organization was renewed as the “Society for General Systems Research”, with
the name later changing to the “International Society for General Systems Research”. The
organization is today known as the “International Society for the Systems Sciences” (ISSS)
and celebrated their fiftieth anniversary in 2004.
A related organization is the International Federation for Systems Research (IFSR) also
established through the Society for General Systems Research in 1980 together with the
Österreichische Studiengesellschaft für Kybernetik, and the Systeemgroep Nederland.
The more applied problem oriented systems thinking (box 3.2) is given three branches by
Checkland (shown in figure 1). However, in the introductory retrospective section of his book
he is content with separating hard and soft systems thinking. “Hard” methods like systems
engineering assumes that the system have a clear purpose and is optimized towards this
purpose, they are goal-oriented. This assumption holds for human-made systems in general
but breaks down for human activity systems, and also for some human-made systems where
there exist conflicting goals and purposes. For these problems Checkland proposes a “soft”
methodology.
The soft systems thinking (SST, box 4.3) is the approach developed by Checkland himself. In
this problem oriented approach model building (capturing abstract activities and issues) is in
focus. The actual details of the model are of less interest. According to Checkland SST should
be used iteratively, situation-driven and in interaction with the system, not sequential,
methodology-driven and intervening with the system (as hard systems thinking). SST is
related to social sciences. The target for Checkland is mainly human activity systems and
social systems but the ideas seem relevant also for the engineering of complex technical
systems with conflicting requirements. The basic ideas of SST are summarized by Flood
(2000).
The hard systems thinking approach (box 4.1) is known as systems engineering and tries to
arrange and describe the real world in a systemic manner in order to be able to perform proper
engineering. The exact details of the model describing the system are in focus. This branch
also has old roots. The first significant systems engineering was performed for telephone
systems to ensure that all the different parts of the phone system interoperated reliably. The
term systems engineering dates back to Bell Telephone Laboratories in the early 1940s, Bell
labs performed major applications of systems engineering during World War II. The first
attempt to teach systems engineering as we know it today came in 1950 at MIT by Mr.
Gilman, Director of Systems Engineering at Bell. (Buede 2000) Today, systems engineering
is promoted by the International Council on Systems Engineering (INCOSE) formed in 1990.
System theories are still a topic of high interest discussed in academia. Some discussions still
focus on the development of GST such as the journal of Systems Research and Behavioural
Science. Other communities have adopted the application of systems thinking, and still others
develop systems engineering, such as the rapidly growing INCOSE. A mapping of the
mentioned organizations and journals to the framework of Checkland is provided in figure 2.
6
Ola Larses and Jad El-khoury
Views on General System Theory
Journals:
Systems Engineering
7
Ola Larses and Jad El-khoury
Views on General System Theory
Site name Organization Link Brief descriptions from the sites
-homepage- American http://www.asc- The American Society for Cybernetics was founded in 1964
Society for cybernetics.org/ by a group of people in Washington, DC who were interested
Cybernetics in the then new field of cybernetics. The founding members
of the Society wanted to follow and to encourage the
development of this interdisciplinary field. The Society now
holds an annual conference, conducts seminars on the
fundamentals of cybernetics, manage a listserve, and
maintains contacts with cyberneticians in other countries.
-homepage- International http://www.ifsr. The International Federation For Systems Research (IFSR),
Federation for org/ founded 1981, is a a non-profit, scientific and educational
Systems agency, constituted of member organizations from various
Research countries. The overall purpose of the Federation is to advance
cybernetic and systems research and systems applications and
to serve the international systems community.
Project: A Harward Law http://h2o.law.h This course is a beta test for a subsequent course to be
General arvard.edu/View arranged for February 2006. Considerable arrangements need
Systems ProjectSyllabus. to be made first (with student input necessary). Pending that,
Theory do?projectId=35 the course text is von Bertalanffy, Ludwig (1969). General
8 System Theory, New York: George Braziller. (Chapter 2 in
General System Theory is the major read.) Editor comment:
The page provides a good overview of several communities,
journals and other GST resources.
Also, courses are held at different deparments across universities throughout the world. To
provide a broad backround a selection of courses are referenced in table 2. The links have
been accessed April 2005.
Table 2 General Systems Courses
Course name Cts Department University Brief description
Generell 5p Matematiska Växjö Syftet med kursen är att de studerande ska:
Systemteori och Universitet • kunna använda ett generellt systemteorietiskt synsätt vid
7,5p systemtekniska
ECTS analyser av verksamheter och design av informationssystem
institutionen • känna till centrala begrepp inom systemteorin
• känna till olika systemklassifikationer
• kunna förstå och analysera konsekvenserna av att använda
olika systemindelningar.
http://w3.msi.vxu.se/~per/IVC742/IVC742.html
General 4p Business, Law University This subject explores a general framework for understanding
System Theory & Information of Canberra diverse kinds of systems, including 'hard' systems such as
Sciences those found in engineering applications, and 'soft' systems in
which the structure of the system is less well defined and
generally involving some form of human activity. The
practical application of systems theory is in advanced
problem-solving in systems. The main question which will be
pursued during the semester is whether the techniques and
ideas of so-called 'action research' - a way of finding out
about the world first described by Lewin and refined and
diversified by a number of authors and researchers including
Checkland.
http://www.canberra.edu.au/courses/index.cfm?action=detail
&subjectid=004601&year=2004
System Theory - Department of New In this course, we will contrast three schools of systems
Management Mexico theory: the naïve US school of systems theory, the General
State Systems Theory School, and the Language School of Systems
University Theory.
http://cbae.nmsu.edu/~dboje/655/655_overview.htm
8
Ola Larses and Jad El-khoury
Views on General System Theory
Course name Cts Department University Brief description
Systems 1p Department of Oregon The course is inspired by the increasing importance of
Theory Forest Science State interdisciplinary perspectives, especially as they relate to
University emerging issues such as ecosystem management and global
environmental change. The goal is to explore some of the old
and new ideas in this field and their possible application in
basic and applied research.
http://www.cof.orst.edu/cof/fs/gradprog/courses/turner/fs507-
02.htm
Seminar in - Information Ourso This course is designed to expand each student's ability to
General Systems and College Of analyze, understand and to conceptually model complex
Systems Decision Business systems. The modern manager faces exceedingly complex
Theory and Sciences Administrat systems that cannot be adequately analyzed using formal
Department ion
Strategic mathematical decision-making, intuiative or experiential
Modeling techniques. In this course, the system dynamics approach to
examining, conceptualizing and modeling macro systems is
emphasized. The informational feedback characteristics of
living systems are studied and methods to evaluate the
productivity and effectiveness of industrial and governmental
systems are addressed. Special attention is given to the
dynamic interaction between system structure, policy, time
delays and information structures in decisions and actions that
determine system behavior and performance. Attention is
given to the implications for information management that are
derived from the general systems analysis process.
http://emac3.ocs.lsu.edu/kathy/cct_courses/isds7920.html
General 5p - Czech In this class students will learn the general, broadly founded
System Theory Technical methodology of Systems Theory to gain a sufficient insight in
University general systems principles and their theoretical limits. The
in Prague course aims a wrapping up knowledge from other special
classes and giving a common frame for many special
engineering problems encountered in practice. It is dealt with
such concepts like identification, decomposition and self-
organization.
http://web.cvut.cz/ctu/international/web/en/prospectus/normal
/f300/subjXE33OTS.html
9
Ola Larses and Jad El-khoury
Views on General System Theory
Term Definition Examples
How many cars were produced? How many had to be
Information about some aspect of
recalled to correct errors? How many mistakes were made?
data or energy processing that can be
Why were mistakes made? HealthCareReportCard.com is
Feedback used to evaluate & monitor the
an example of how hospitals are doing with certain
system & to guide it to more
diagnoses. Accreditation reports are an example as are
effective performance.
patient satisfaction surveys, sales reports, and test results.
A system which is a part of a larger
The finance department, the information system, the
system. They can work parallel to
Subsystem managerial system, the renal system, the political system,
each other or in a series with each
the workflow system (such as the conveyor belt), etc.
other.
Neither system elements nor the
Static
system itself changes much over time A rock
system
in relation to the environment
A healthy young adult grows more independent,
The system constantly changes the
Dynamic interdependent, & self-sufficient & self-directed in response
environment & is changed by the
system to stimuli from peers, family, school, work, & recreational
environment
activities.
Fixed, automatic relationships among A rock is an example of the most closed system. We may
Closed
system components & no give or encounter families that are isolated from the community &
systems
take with the environment resistant to any outside influence.
Interacts with the environment
trading energy & raw materials for
Hospitals, families, people, body systems, banks,
Open goods & services produced by the
manufacturing plants, governmental bodies, associations,
systems system. They are self-regulating, &
businesses, etc.
capable of growth, development &
adaptation.
The line or point where a system or
subsystem can be differentiated from
its environment or from other The nursing unit, the occupational therapy department, the
Boundary subsystems. Can be rigid or elementary school, a person, an agency or business, a fence
permeable or some point in between. or wall, roles, ect
Systems or subsystems will engage
in boundary tending.
The overall purpose for existence or
the desired outcomes. The reason for To educate students, to support people during illness &
Goal being. Currently, many organizations restore them to health, to make money, to create social
put their goals into a mission order, etc.
statement.
Rules are made, policies & protocols are written, approved
The tendency for a system to develop & communicated to staff; laws are enacted & violators are
Entropy
order & energy over time. held accountable; a marathon runner in training gradually is
able to run farther.
The disorganization after a hurricane, a rigid, frightened
The tendency of a system to lose family produces a child who is unable to think
Negentropy
energy & dissolve into chaos independently or leave home, a new business has no forms
or protocols for handling consumer complaints.
Pilots use instrument panels & devices to constantly
evaluate & make course corrections; teachers grade papers
The activities & processes used to & give students grades on exams; parents measure their
Control or
evaluate input, throughput & output children's height & weight & may adjust the child's diet;
cybernation
in order to make corrections health care agencies use TQM or Quality Assurance
programs; employee health nurses review records to see
who needs immunization updates.
10
Ola Larses and Jad El-khoury
Views on General System Theory
Term Definition Examples
A nursing assistant assigned to empty catheter bags on a
unit could begin in the middle of the hall, on the right side,
on the left side, at the front or back of the hall & still end up
Objectives can be achieved with
Equifinality with all the bags emptied. A traveller could take the
varying inputs & in different ways.
interstate or back country roads & still arrive at their
destination. The traveller could go by train, plane, bus or car
& still arrive at desired location.
A set of objects together with
relationships between the objects and
between their attributes. The objects
System See subsystem
set (parts, elements, attribures, etc)
are the undefined primitives of
systems thinking.
An observer makes observations
such as sensations on the sense
organs, readings from instruments,
etc. An observation is the act of
choosing an element from a set of The programmer, user, owner or maintainer of an
Observer possible observations of that type for information system.
that observer.
The objects defining a system come
from the mind of an observer, and so
a system is relative to the point of
view of an observer.
A model of a system in which the
system can only be known through
observing its behaviour, without
looking inside of it. The observer can
Black box An astronomer studying the universe.
however decide on the scope and
range of observation, based on what
is believed to be the important
features of the system.
An approach to understanding the Analyzing a electronic circuit by understanding the
White box system in which the inside of the workings of its internal components such as resistors,
system is revealed. transistors, etc.
A particular situation of the system
State that the observer can recognize if it The on/off states of a lamp
occurs again.
Quality The quality of mass defined by the states in which masses
A way of grouping states of a system
(property) are the same or different.
11
Ola Larses and Jad El-khoury
Views on General System Theory
2.3.1 A Criticism of Bertalanffy
Beginning with the view of Bertalanffy, focus is placed on the properties of the whole.
Bertalanffy claims that there is no such thing as emergent properties of systems. The whole is
more than the sum of its parts, but equal to the sum of its parts and the relations between the
parts. (Guberman 2002) “If, however, we know the total of parts contained in a system and
the relations between them, the behavior of the system may be derived from the behavior of
the parts.” (Bertalanffy 1969)
System unity
Part Part
relationships
Part
12
Ola Larses and Jad El-khoury
Views on General System Theory
Suppose one drops a teacup (unity), so it breaks (‘analysis’) into pieces (parts). One then glues the
pieces together (‘synthesis’) in such a way that one can drink tea from it again (restored Unity). In this
metaphor, the glue symbolizes a new addition (Relationship) that was not present in the teacup before
it was broken, but had to be added in order to restore the cup.
This system definition, with unity as a complementary representation to parts and relations, is
illustrated in figure 4.
System unity
Part Part
Analysis
Synthesis
Part
Part
Figure 4 Logical relations among System constituent according to Kant (Dubrovsky 2004)
Further, three oppositions are defined as form-content, complex-simple and external-internal.
Based on these oppositions three system procedures are found. The first procedure concern
decomposition of an object into parts and is the opposite composing. The second is the
measuring of aspects of parts and wholes that is the opposite of configuration. The third is the
insertion of an element into the object’s structure, opposed by the extraction of an element out
of the structure.
13
Ola Larses and Jad El-khoury
Views on General System Theory
Transcendental systems are beyond knowledge, unknown to man and can be ignored for our
purposes. Natural systems can be analyzed, human activity systems can be analyzed and
influenced, and designed systems can be analyzed and redesigned. A designed system has a
function designed for a purpose. Checkland makes a clear distinction between activities (or
systems) that simply serve a purpose, and activities (or systems) which are the result of a
willed choice by human beings. Checkland chooses to label the first type serving a purpose as
purposive, and the latter, when conscious human action is involved purposeful.
* *
* Part
* contains
relation
(a) (b)
14
Ola Larses and Jad El-khoury
Views on General System Theory
15
Ola Larses and Jad El-khoury
Views on General System Theory
4 References
Ahari P. (2003) A Living Systems Approach to Product Design and Development. Doctoral thesis,
ISRN/KTH/MMK/R-03/41-SE, TRITA-MMK 2003:41, ISSN 1440-1179, Department of Machine
Design, KTH, dec 2003
Bertalanffy, L. von (1969) General System Theory. New York: Brazilier.
Buede, D. (2000) The Engineering Design of Systems: Models and Methods, John J. Wiley & Sons.
Checkland, P. (1999) Systems thinking, Systems practice: Includes a 30-Year Retrospective. John Wiley & Sons.
Checkland, P. (1981) Systems thinking, Systems practice. John Wiley & Sons.
Dubrovsky V. (2004) Toward System Principles: General System Theory and the Alternative Approach. Systems
Research and Behavioural Science. Vol 21. 2004. pp 109-122.
Flood R.L. (2000) A Brief Review of Peter B. Checkland’s Contribution to Systemic Thinking. Systemic
Practice and Action Research. Vol. 13, No. 6, 2000, p723-731.
Gillies, D. A. (1982). Nursing management a systems approach. Philadelphia: W. B. Saunders Company, 56-74.
Guberman S. (2002) Reflections on Ludwig von Bertalanfy’s “General System Theory: Foundations,
Development, Applications” Proceedings of the 5th European Systems Science Congress, Crete, October
2002
IEEE. (1998) IEEE Standard for Application and Management of the Systems Engineering Process. IEEE-Std
1220-1998., IEEE, New York, 1998.
Loureiro, G. Leany, P.G. & Hodgson M. (2004) A Systems Engineering Framework for Integrated Automotive
Development. Systems Engineering. Vol. 7, no. 2, p. 153-166.
OMG. 2002. OMG - Meta Object Facility, v1.4. April 2002.
Shchedrovitsky GP. (1966) Methodological problems of system research. General Systems 11.
Skyttner L. (2001) General Systems Theory. World Scientific Publishing, Singapore, ISBN 981-02-4175-5
Weinberg, G. (2001) An introduction to general systems thinking (silver anniversary ed.), Dorset House
Publishing Co., Inc., New York, NY, 2001
16
Ola Larses and Jad El-khoury
Views on General System Theory
17
Ola Larses and Jad El-khoury
Views on General System Theory
The book of Lars Skyttner provides an overview of the main original ideas related to general
systems theory as well as a summary of common ideas.
The first part of the book, titled “The theories and Why”, gives a good overview of related
thinkers and the historical development of more abstract and general theories. What is lacking
in this section is a proper conclusion, the contents are very descriptive and the author does
very little to generalize the described theories which would be expected after such a thorough
coverage of the topic. For my taste this would have been more valuable than a chapter with 14
different theories. It is however nice to read, especially for people collecting curiosities.
The second part of the book is astonishingly application specific with very few relevant
parallels, generalizations and conclusions related to any of the general theories. In the end a
summary of methodologies is given but the methodologies are too shallowly described in
order to provide further insight into general systems theory.
The book may be a nice introduction for people looking for an overview of system theories
and inspirational for further research; but it includes more detail in application specific topics
than necessary and less insightful generalization and details of general system theories than
expected. The concept of general systems theory remains unclear after reading the book.
Skyttner states that it is impossible to be efficient with one theory and that several views are
necessary, this may be true and also explains the general impression of the book as probing
much into the details of the applications, and avoiding generalizations. This lack of generality
seem a bit contradicting to the title and also the final chapter where a general systems theory
is seen as the next paradigm of science. The confusion may be attributed to the deficiency of a
clear conclusion in the book.
Summary of the contents
The first chapter of the book provides a historical summary of the view on science since the
15th century. We are placed in the system age where problems (in theory) are addressed in a
holistic and interdisciplinary way, instead of the prior reductionist method of breaking down
problems into parts. The emergence of the general systems theory (GST) in the 50’s is
mentioned, and Ludwig von Bertalanffy and Kenneth Boulding are specifically referenced in
this context.
In the next chapter, Skyttner summarizes some basic ideas in GST. He introduces a range of
definitions and concepts of systems that are used throughout the book. Some general laws,
principles, theorems and hypotheses are also posed. Much of the material here is immediately
derived from the work of von Bertalanffy.
The third chapter provides a summary of 14 existing system theories. Many of the theories
aim for a classification system of systems. What is interesting in this collection is that many
of the theories lack generality and instead apply specific layers of system types to describe the
world. Many of the described system theories fit into one proposed distinct classification
hierarchy defined by the respective authors, usually inspired/guided by the physical size of the
systems. Some of the classifications seem very adapted to popular notions of science, and
occasionally quite ad hoc, which make them feel less relevant and useful. However, some
interesting general ideas can be extracted, for example from the subsystem definition of
Millers Living Systems Theory, the typology of Checkland, the taxonomy of Jordan and the
recursive application of the Viable System model of Beer.
18
Ola Larses and Jad El-khoury
Views on General System Theory
After the overview one chapter is spent on describing the communication and information
theory. At first this dispositions feels strange, especially as the book lacks a good bridging
section in the text. The chapter brutally begins with defining the difference of communication
and information, but as the contents develop the chapter feels highly relevant and important
for a wider understanding of systems. The idea that information only exist in the eye of the
beholder is repeated and the nature of information as an abstract entity is discussed. The
general points of information are then elaborated in the next chapter in a human application
context discussing theories of brain and mind. This chapter concludes the first part of the
book titled “The theories and Why”.
The second part called “The applications and How” begins with a chapter about Artificial
Intelligence (AI) and Artificial Life that smoothly links to the last chapter of the first part
about theories of brain and mind. The chapter includes nice trivia on the definition of life and
futuristic visions including a recollection of current research topics in the field of AI.
The next, and seventh, chapter details the theory of organization and management. Here
another nice historical recollection of a theoretical field is given but the attempt to summary
of “a systems approach in ten points” is less clear. A management related topic is covered in
the next chapter, “decision-making and decision aids”. The chapter provides theories on what
decisions are, how they are conducted, and how computers can be used for supporting
decisions. Some useful concepts and categories are introduced. The process of decision
making is central both for management and engineering, however the role and process of
decisions is usually not equally clear in engineering compared to management. The inclusion
of the chapter seems highly relevant and provides some guidelines on how to cope with the
increasing uncertainty of increasingly complex systems in a fast changing environment.
The ninth chapter on informatics is again very application-specific and full of technical details
that add very little value for the informed reader. The only interesting content is a short
summary of a suggested lifecycle model for evolutionary development of information and
communication networks.
Chapter ten returns to the general theories and summarizes the application of a few system
methodologies. The distinction between soft and hard methodologies based on the ideas of
Checkland is established and the described methodologies are classified accordingly. Soft
methodologies are best applied to ill-structured problems with unclear objectives and
purposes. Hard methodologies are goal-oriented and solve well-defined structured problems.
The overview and points made are relevant, the methodologies are however too briefly
described to be well understood.
The final chapter is a political manifest that suggests systems thinking to be a new paradigm
struggling to make a break in a world of critics. In the end the system theories will prevail as
the only sustainable way to implement science.
Theoretical concepts and definitions
Even though the book contains much application specific information, chapter 2 “Basic Ideas
of General Systems Theory” provides a range of useful concepts and definitions.
The hallmarks of a general systems theory according to Skyttner, specifically referring to
Bertalanffy and Litterer are summarized in a list of 10 points.
Interrelationship and interdependence of objects and their attributes – Unrelated and independent
elements never constitute a system.
Holism – Some properties exist only at system level and cannot be detected by analysis of the
components of the system.
19
Ola Larses and Jad El-khoury
Views on General System Theory
Goal seeking – Systems strive for a final state or equilibrium.
Transformation process – Systems transform inputs into outputs.
Inputs and outputs – Open or closed to the environment of the system.
Entropy and negentropy – All systems tend toward disorder. A living system can for a finite time use
energy to create order (negentropy).
Regulation and feedback – All systems have regulatory mechanism, feedback is a requisite for
effective control.
Hierarchy – Systems exist of subsystems, that in turn are systems of their own. A hierarchy of systems
exist.
Differentiation – Specialized units in the system performs specialized functions.
Equifinality and multifinality – systems have alternative ways to achieve the same goals
(convergence), and can obtain different mutually exclusive goals from a similar initial state
(divergence).
Besides the summary of the general systems theory a set of useful classification frameworks
and concepts are provided. For example, systems can be:
Concrete (living, non-living), conceptual or abstract – Depending on the tangibility of the system.
Open, closed or isolated – Depending on the relation to the environment.
Decomposable, near-decomposable or non-decomposable – Based on the dependence of subsystems.
Static or dynamic – Based on the activity of the system.
Black, grey or white box – Depending on the knowledge of the internals of the system.
Chapter 2 provides a filtered summary of the collected work in systems theory and provides a
toolbox for system reasoning. Unfortunately the presentation of the material could have been
more structured. The concepts are highlighted in the text, which is very good, but a good
structuring or overview of the concepts are lacking.
20
Ola Larses and Jad El-khoury
Views on General System Theory
Von Bertalanffy is considered one of the founders of the general system theory and the book,
also titled “General System Theory” is a collection of papers and book excerpts by his hand.
The core of the theory is given in chapter 3 based on a paper authored in 1945, first published
in German with the translated title “An Outline of the General Systems Theory”.
The ideas of Bertalanffy are surprisingly focused on mathematics, but he also strongly
acknowledges the use of soft, qualitative models which is seen as an intermediate step in the
theory building process.
Bertalanffy brings out a range of concepts related to general systems, but the system
definition and ontology with related general system principles is not that clear and crisp. The
models and concepts are rather soft, which Bertalanffy acknowledges and motivates by
claiming that this is the first steps towards a more rigorously defined theory.
The book has good value as a collection of papers from one of the named “fathers” of GST. It
contains some of the original formulations of some basic system concepts. However, in order
to be useful, the theories presented need more elaborate contents. A harsh interpretation of the
work is that is only says that entities and relationships should be represented by mathematics,
and only a few hints of what these mathematics should look like is given.
Summary of the contents
The book, being a revised edition, is introduced by a preface where some additions are made
on the background and at the time current flows of the theory. The first chapter, also an
introduction, repeats some of the historical background and the motivation of the field.
Mathematics is acknowledged as a general theory as it can be applied to a variety of
problems. With some previous background on the subject the introduction becomes somewhat
long and tedious. In addition, the references to trends and people in academia in the 50’s and
60’s also make the section hard to read.
The second chapter details the shortcomings of disciplinary science and introduces some basic
concepts. It is noted that contemporary science only works with closed systems and theories
must be developed also for open systems. The constant increase of entropy is contradicting
the possibility to build organized complex systems; the idea of an open system consuming
energy remedies this conceptual problem. Further, the distinction of information as a
complement to energy is necessary for some theoretical constructs. Also, goal orientation and
purpose are concepts that must be added, the contemporary science is referred to as
mechanistic and only working with causality. A mathematically based GST is expected to be
developed, however the use of soft and verbal models should not be underestimated,
especially at higher levels of abstraction, by von Bertalanffy referred to as organization.
The third chapter expands the mathematical content of the theory. First an important
distinction between summative and constitutive characteristics of elements is introduced. The
summative characteristics concern properties independent of relations. For summative
characteristics the system is no more than the sum of its parts, a simple example is the total
weight of a mechanical system. Summative properties are independent of other entities. The
constitutive characteristics concern properties that depend on specific relations to other
entities. In the expression “the whole is more than the sum of its parts” the difference refers to
the constitutive characteristics. The usefulness of differential equations for this purpose is
developed and some basics of mathematics and control theory are given.
21
Ola Larses and Jad El-khoury
Views on General System Theory
Then a range of system concepts are established. A discussion on dependent and independent
properties of entities is given, where the system properties are the dependent properties that
make the whole more than the sum of its parts. This is followed by an interesting note that in
biological, psychological and social systems interactions among elements decrease over time,
from wholeness to independent, specialized causal chains. The machine-like behaviour of the
sum of the independent chains is referred to as progressive mechanization. Another related
interesting note is that systems often have a leading entity. This leading entity defines the
individuality (from indivisible) of the system. The individuality and the leading role of the
leading entity, increases as specialization of entities progress in the system. By this
specialization the dependencies among the entities increase and over time a system becomes
more bound together. The process is referred to as progressive centralization. Systems will be
arranged in a hierarchical order of centralized systems.
Then the concept of finality and types of finality is introduced. The first type is static
teleology or fitness, which means that a given system seems to be useful for a given purpose.
Then there is dynamic teleology indicating a directiveness of processes. This is a direction of
events toward a final state, possibly based upon the structure of the system. There is also true
finality or purposiveness, for example in man made systems where it is fitness and structured
working of machines due to a planning intelligence. It is also mentioned that the same final
state can be reached from different initial conditions and through different paths, this is
labelled equifinality.
The fourth chapter expands the advances in GST and exemplifies how several fields of theory
introduce system concepts like organismic analogies, and interdependencies of entities rather
than causality. The last section of the chapter lists a set of theories and relates them to the
previously described concepts of general system theory. Bertalanffy recognizes two methods
of general systems research: the empirico-intuitive followed by himself, and the deductive
developed by Ashby. The emprico-intuitive approach looks for laws in each discipline and
then compares the results to find laws that hold across systems in general. The deductive
approach begins with a definition of system and from this definition general laws are deduced.
In the fifth chapter the mathematical parts are expanded with definitions of equifinality and
concepts related to open systems. A few applications in biology and their maturity in
developing mathematical models are referenced.
The next chapter, called “the model of open systems” discusses how most (contemporary)
theories are based on closed systems without import and export of matter. It quotes the laws
that entropy is always increasing; but recognizes that for an open system, entropy may be
reduced by import of matter. Negative entropy (or negentropy) is identified as information. A
decrease in entropy of a system means that more information is available in the system in the
sense that a complete system description requires a more elaborate content.
Chapter seven is a case study of general systems theory in biology. Open systems, feedback
and mathematical relations are suggested to be complementary and equally important
theoretical contributions. Bertalanffy also make a warning for oversimplification in the
models. Mathematical models are shown to be useful but not conclusive, all models are said
to be approximations of what they describe. Equations are representing a theory if all
parameters of the equation can be confirmed by independent experiment; and if predictions of
yet unobserved facts can be derived from the theory. Then a case showing the development of
a mathematical description for metabolism and growth exemplifies the reasoning.
Chapters eight and nine continues to exemplify applications of the general systems concepts
through the sciences of man. Chapter eight details the benefits of using a system model in the
social sciences. First it is established that the contemporary theories of man as a robot is
22
Ola Larses and Jad El-khoury
Views on General System Theory
incomplete, and must be replaced with a view of man as an active open system interacting at
the social system level. This does not imply that mathematical laws are useless, for example,
historical events can often be explained by applying a law of social behaviour; however,
Bertalanffy again warns about oversimplification and interpreting models to strictly.
In chapter nine, system theory in psychology and psychiatry is discussed. The model of man
as an active organism is further extended and exemplified. The ideas of progressive
mechanization and the development of leading entities are repeated. It is also implied that the
brain may perform such changes, adding higher layers of behaviour with each layer including
a leading entity. This is in line with the levels of Maslow, and is exemplified by the evolution
of creatures in nature.
The final chapter widens the scope by posting that most of the published theories are based on
the mindset of the western world and our notions of organizational structures, time and space.
Examples of cultural differences in the number of existing words for a given phenomenon,
and even in how time is treated are given, specifically from the Eskimo Hopi culture and
American Indian cultures. Any model only captures a few aspects of reality and the categories
and models of our experience and thinking are determined by biological and cultural factors.
Also, different backgrounds and domain knowledge also influences our view of a system. An
example is given with a table, seen as a system of atoms by the physicist, a system of wood by
the biologist and a unit of capital by the economist.
Theoretical concepts and definitions
Being an overview of the early theories of GST several concepts and definitions are
introduced. Below are a few of the core ideas:
Isomorphism – Along the disciplines of science there exist common ground, possible to cover by a
common theory. This property is referred to as isomorphism.
Goal orientation (active systems) – Systems may actively strive for a goal or final state. They are not
necessarily mere causal machines.
Open system – Systems where material flows in and out are open systems. In an open system entropy
may increase locally due to the inflow of material.
Homeostasis (feedback) – Systems strive for a desirable steady-state through regulating itself based on
feedback of information.
Equifinality – systems have alternative ways to achieve the same goals (convergence), and can obtain
different mutually exclusive goals from a similar initial state (divergence).
Progressive mechanization – Systems strive towards specialized entities.
Progressive centralization – In the process of mechanization each system is organized around a
leading entity.
Hierarchical Layers – Each leading entity creates a system layer, higher layers can be introduced with
new leading entities and subsystems also have leading entities.
23
Ola Larses and Jad El-khoury
Views on General System Theory
The book was first published in 1981 and in the 1999 edition it includes “a 30-year
retrospective” a paper also published elsewhere. This section is one of the most interesting
parts of the book, giving a brief outline of the soft systems methodology (SSM) comparing it
to hard systems engineering.
SSM advocated by Checkland is seen as complementary to systems engineering to deal with
complex situations, exploring the underlying process. The target for Checkland is mainly
human activity systems and social systems but the ideas seem relevant also for the
engineering of complex technical systems with conflicting requirements.
Studying the ideas reveals a loosely defined methodology that occasionally seem arbitrary and
non-systematic. However, this is maintained as one of the strengths of the methodology as it
enables a discovery of the actual underlying processes. The general idea is to draw simple,
graphical models to visualize parts of the system and create a common ground for discussions.
The actual notation in the models is very simple but elaboration of modelling is suggested if it
improves understanding of the system, the selection of extended language is left as an open
issue.
The book gives a nice perspective on the benefits and limitations of system ideas in practical
application for undefined problems. For technical systems with clear requirements a hard
systems engineering approach is suggested, but the book places nice bounds on the domain of
systems engineering.
Summary of the contents
The 30-year retrospective, which is the first part of the volume, begins by discussing the
notion of hard and soft systems. Checkland points out that hard systems thinking have been
successful in engineering technical systems with clear purposes and objectives. However, in
fuzzy situations with unclear entities and where the objectives are uncertain and contradicting
soft systems thinking can be useful to clarify the picture. (This can be applied to social
systems but also architecture design and similar situations.)
After an historical overview of proposed methodologies from the author, the current
modelling methodology is outlined. Checkland proposes that soft problems should be
explored by modelling the transformation in focus, guidelines for performing this modelling is
given. The purpose of the model is either to implement change or to understand a complex
process.
Checkland underlines the methodology aspect of soft system methodology (SSM). A
methodology is a collection of methods from which you choose the appropriate ones for a
given situation. Ultimately SSM should be used iteratively, situation-driven and in interaction
with the system, not sequential, methodology-driven and intervening with the system. It is
concluded that SSM does not replace but rather extends the traditional and existing system
engineering approach. Then the first chapter in the main book begins with an introduction that
places systems theory side-by-side to science. Systems theory is not a science, it is a meta-
theory that can be applied in the same way as science.
The main text is divided in two parts similar to the title, a systems thinking part and a systems
practice part. Chapter two is the first of the systems thinking part and gives a thorough and
well told walkthrough of the development of science, from the Greeks (as always) to Einstein.
A scientific approach breaks down a problem, formulates it mathematically, and makes
24
Ola Larses and Jad El-khoury
Views on General System Theory
repeatable experiments to verify the validity of predictions. The value and usefulness of this
approach are appreciated as monumental. This background serves as a foundation for the
discussions of the systems approach covered in the next chapter.
The third chapter introduces some problems for science. The scientific approach has problems
to cope with complex, large, multi-variable problems where it is impossible to isolate a small
set of variables for analysis. Further, for higher level systems like social systems it is difficult,
not to say impossible to properly measure and repeat experiments. Monitoring a system
changes a system which means that results depend on the method of measuring (compare the
uncertainty of Heisenberg). A third problem is the issue of management where decision-
making is an instant interaction and the situation is rarely repeated. The one shot nature of
monitored events inhibits a scientific approach. These problems call for a different approach
to cope with some human activities.
First Checkland introduces the concepts of Emergence and Hierarchy. Systems are
hierarchical, layered arrangements where some properties emerge at a given layer of
abstraction and each layer contains laws that need to be studied separately. Each layer
contains systems or ‘holons’ that are organized and linked. Second, Communication and
Control are introduced as an important systems concept pair. In a system not only energy but
also information is flowing. The information is a view of the system that allows entities
within it to react. Some entities exert control on others with respect to given control variables,
defined by the information content. In technical systems this control is designed by the control
engineer, for biological systems this control exists and can be modelled by a systems
approach.
Next, Checkland produces and overview of the systems movement where he recognizes three
problem solving applications related to the theoretical development of systems thinking.
Applied systems thinking is performed in (1) ‘hard’ systems, (2) decision-making problems
and (3) ‘soft’ systems.
The fourth chapter “Some Systems Thinking” begins with describing some basic ideas that
are recurring in different theories. A systems description is always related to an observer. The
description itself contains entities, relationships/coherence, a boundary, a control mechanism
(defining the entity’s identity) and is part of a hierarchy.
The theory of Boulding’s hierarchy and Jordans’ taxonomy are referenced and the importance
of the observer in a system description, lacking in Jordans model, is underlined. Then the
typology of Checkland himself is described. Systems belong to one of five types: natural
systems, human-made physical systems, human-made abstract systems, human activity
systems and unknown transcendental systems. Natural systems are evolutionary made, while
the human-made systems are designed for a purpose. Human activity systems are, according
to Checkland, distinguished by the free will which makes humans unobservable and
unpredictable. Human activity systems are made up of purposeful activities, occasionally
using purposive designed systems as tools. These definitions together with a short conclusion
on basic systems thinking summarize the first part.
The second part “Systems Practice” begins with an overview of “Hard” systems thinking, and
its’ limitations. “Hard” methods like systems engineering assumes that the system have a
clear purpose and is optimized towards this purpose, they are goal-oriented. This assumption
holds for human-made systems in general but breaks down for human activity systems, and
also for some human-made systems where there exist conflicting goals and purposes. For
these problems Checkland proposes a “soft” methodology which is elaborated in the next
chapter.
25
Ola Larses and Jad El-khoury
Views on General System Theory
The soft systems methodology has seven stages. Stages 1 and 2 concerns expression of the
problem; as much information as possible is collected from a variety of sources to make the
richest possible picture available. In stage 3 a root definition of relevant system(s) are
formulated. Based on the root definition a conceptual model is developed. The root definition
is an account of what the system is; the conceptual model is an account of the activities the
system must do in order to be the system of the root definition. In phase 5 the derived
conceptual model is compared to reality. Do the activities of the conceptual model fit the
existing system or what are the differences? Based on the results, stage 6 contemplates
feasible and desirable changes and stage 7 implements changes through action. The
methodology is illustrated by an example, and further examples are given in chapter seven.
After the examples, chapter seven introduces some conclusions from the research. First the
importance of understanding the Weltanschauung (viewpoint) of the root definition is
discussed. Depending on the viewpoint of the person formulating the root definition the
contents of it will be different; this is one of the purposes of the root definition, to capture
diverging views of the system. Further, the root definition can be focused on a given primary
task or a more general issue of a system. The mnemonic CATWOE elaborated below is given
to support the construction of root definitions. The importance of elaborating the initial study
is underlined, and understanding of the power and politics derived from roles, norms and
values are rated as highly important for the success of a study and the development of a useful
conceptual model.
The importance of understanding the difference between what a system does and how the
system does it is also important. The conceptual model derived from the root definition is an
abstract formulation of what the system does, the comparison with reality reveals how the
desired function is performed. In the stage of mapping between the two the link must be
understood. Related to the logical hierarchy is the law of conceptualization that states that a
system which serves another cannot be defined or modelled until a definition and model of the
system served are available. Models, according to Checkland, consist of verbs specifying
activities which actors carry out.
The eight and final chapter is presented as a third part of the book containing conclusions. In
this chapter Checkland places his theory and methodology in the context of social science and
related work of other authors. At the end of the book are two appendices with some hands-on
advice on the topic, collected under the headings “building conceptual models” and “a
workbook for starting system studies”.
Theoretical concepts and definitions
In the typology of Checkland there exist five types of systems:
Natural systems – Found in nature and developed by evolution.
Human-made physical systems – Tools and machinery existing in the real world.
Human-made abstract systems – Conceptual systems like mathematics.
Human activity systems – A purposive system which expresses some purposeful human activity.
Trancendental systems – Systems yet unknown to man.
In the soft systems methodology a root definition must be formulated, the root definition is a
formulation of the function of a given system (a ‘what’).
Root definition – A position in a means end hierarchy of Why(R), What(P), How(Q). It is always
possible to go up or down in this hierarchy.
A root definition should meet the requirements of including the six elements of CATWOE:
Customers – The beneficiaries or victims of the system.
26
Ola Larses and Jad El-khoury
Views on General System Theory
Actors – Agents who carry out the activities of the system, especially its main transformation.
Transformation – The core process of the root definition.
Weltanschauung – The viewpoint of the person formulating the definition.
Ownership – The guarantee of the existence of the system.
Environmental constraints – Impositions that the system takes for given.
27
Ola Larses and Jad El-khoury
Views on General System Theory
1
GS thinker produces a set of tools of simplifications that are common enough to be useful for all scientists.
That is, they build a common model that can be applicable by all sciences. Isn’t this what we are trying to do in
AIDA2? Build a common model that all specialists understand and can base their specific models on? So, why
not use GST?
28
Ola Larses and Jad El-khoury
Views on General System Theory
the methods of simplification that have succeeded and failed in the past, we can handle
complexity.2
III
Randomness
Complexity
I – organised simplicity – machines
II – unorganised complexity – populations
III – organised complexity – systems
III is too complex for analysis and too organised for statistics, hence we resort to systems
theory. System theory came about as knowledge moved from the mechanical view (I) to the
organised complexity world (III)3
2
This is also what we are interested in! To produce some simplification techniques that allow the designers to
handle the complexity of the systems to be built
3
Modern machinery is moving from I to III as computer technology is introduced. Maybe that is why we can use
GST in our engineering work?
29
Ola Larses and Jad El-khoury
Views on General System Theory
simplifications. Many ills of society came from a too good an application of these fruits. Page
20: Science is unable to cope with MNS, though its success with systems of its own choosing mislead
many into thinking of science as a way of dealing with ALL systems.
One method of simplification applied in technology, is to focus on the parts of the system,
while paying less attention to the connections between these parts and to the rest of the
system. The problem is pushed from the parts to the connections. This may be carried to great
extremes. From time to time, it is recognised that a system is not merely a collection of parts,
but a collection together with the relationships between them. Then, a new level of technology
is reached. This new technology becomes in its turn a ‘component’ in the new way of thinking
and the connections to it become the weakest part of the system. This separation of function is
useful, but it should not be carried to extreme. Revolutionary movements recognise the
importance of the connections and synthesise them into a new field of knowledge (new part)
such as electromagnetism, physical chemistry, etc. GST is not going to yield the kind of
control expected over MNS, its contribution is to be in limiting the excesses of other
approaches to complexity. Page 22: Perhaps we are reaching the useful limits of science and
technology whose philosophical underpinnings are techniques restricted to systems of small and large
numbers.4
4
Isn’t that what we are also experiencing when designing a truck? The parts (disciplines) are very clean but the
connections are getting weak. Need to synthesise, leading to multidisciplinary engineering (Mechatronics).
30
Ola Larses and Jad El-khoury
Views on General System Theory
schemes. When he does, he will identify the ‘foreign’ language of auto mechanics as the
source of difficulty (ethnocentrism).5
Scientific disciplines, like social groups, have category schemes to facilitate internal
communication. These categories may change while ‘normal scientists’ work within a given
scheme or paradigm. ‘Revolutionists’ create new ones and destroy old ones. This same
revolution may be performed by ‘interdisciplinarians’ on many different disciplines by
carrying the change intact from one to the other. ‘Interdisciplinarians’ differ from ‘generalists’
in that the former knows one thing that they apply over and over again, while the latter knows
many things. He adapts to the other paradigms instead of applying his paradigm on the new
discipline. Like the anthropologist that adopts to live with many different cultures, as opposed
to the colonist who imposes his paradigm on the cultures he needs to live with. How is that
done? They too have a single paradigm, but it is taken from a much higher vantage point,
from which all the disciplines are seen to be alike, but obscured by different languages. For
this to work, a belief in the unity of these disciplines is needed.6
Page 35: To be a good generalist, one should not have faith in anything. Every article of faith is a
restriction on the free movement of the generalist among the disciplines. One should be careful in
assuming that one paradigm is more ‘real’ that another. It may be essential for a scientist to
have faith in the truth of his discipline, but this only diminishes his chances of making a
revolution or moving to another discipline.
31
Ola Larses and Jad El-khoury
Views on General System Theory
more familiar territories. If we are mistaken, we can readily correct. If we insist on reading
every house number, we may miss dinner.)
8
We will use these laws in that spirit also. Don’t have to follow laws strictly.
32
Ola Larses and Jad El-khoury
Views on General System Theory
becoming one. While it was originally intended to overcome overspecialisation, it is
becoming a specialisation itself.
33
Ola Larses and Jad El-khoury
Views on General System Theory
observer. The appearance of absolute meaning in certain statements comes because there is an
almost universal agreement on the meaning.
A system has ‘no purpose’, for purpose is a relation that depends on the observer. General
Motors to a user exists to put out cars, yet to the junk dealer it is there to put out scrap metal.
A system does not have a reason to exist, but more or less an official public reason, just like
the public agreement on the meaning of a word. We just do not need to highlight this all the
time.
It is more forceful to speak in absolute terms. Most of the time, absolute speech will not get us
in trouble, though we may learn something if we examine the relative nature of some
seemingly absolute statements. A simple example of absolute thinking is seen in answers to
the question: ‘What happens to the reading on a thermometer if we suddenly plunge it into hot
water?’ A simple answer is: ‘the reading must rise because the reading measures the
expansion of mercury and the mercury expands when heated.’ There are 2 concealed
absolutisms in this statement. One has to do with the time scale of the observation, since it
seems to imply instantaneous expansion. The second lies in the ‘expansion of mercury’
statement. The reading actually measures the difference in expansion between mercury and
glass (relative expansion, not absolute.) and because the glass, being on the outside, expands
first, the thermometer actually drops first before it starts rising. A thermometer, like a
language, is an instrument for understanding the world. When we use it for simple things, we
can use simple language to describe what it does. But, for more advanced applications, we
may need to refine the view of the thermometer.
System writers speak of ‘emergent’ properties of a system, properties that did not exist in the
parts, but are found in the whole. Others attack this idea, saying that these properties are but
another name for vital essence. They support their arguments with examples of emergent
properties that turned out to be perfectly predictable. Both arguments are right, but they are in
trouble because they speak in absolute terms, as if the ‘emergence’ were stuff in the system,
rather than relationships between a system and an observer. Properties emerge for a particular
observer when he did not or could not predict their appearance. We can find cases where a
property is emergent to one observer and predictable to another. By recognising emergence as
a relationship between the observer and whatever he observes, we understand that properties
will emerge when we put together complex systems. The property of ‘emergence’ no longer
emerges for us, though it surprises those who take the absolute view.
How can we avoid fallacies of absolute thought? Always remember the human origin of our
models, words, instruments and techniques. Absolute thought is a simplification that serves
well at certain times, on a certain scale of observation and for certain purposes. It works as
long as we work following conventional patterns, in conventional situations. Page 62: … any
system is the point of view of one or several observers. Whether our view – or their view – is good or
bad can only be judged according to the purposes which the system is designed to satisfy.9
A System is a Set
Even though any arbitrary way of looking a the world can be a system, we could not say
anything about truly arbitrary sytems. So, we narrow our attention to non-arbitrary systems,
forcing attention to the reaons for the non-arbitrariness. These reasons are the source of order
that makes systems thinking possible. Non-arbitrariness has two sources. It could be ‘out
there’ in the real physical world, or in the observer. We focus on the observer for the moment.
We note that any way of looking at the system do not form an arbitrary system, for the way
9
In AIDA2, there are several types of observers designing a system. A system will consist of several points of
view. The good/bad is based on the purpose of each of the types of observers.
34
Ola Larses and Jad El-khoury
Views on General System Theory
belongs to the mind of the observer. Arbitray systems are hard to find since as soon as we
think of one, it becomes non-arbitrary.
The role of the observer is ignored in systems writing by, for example, moving into a
mathematical representation of the system – without saying how that representation was
chosen. For example, Hall and Fagen give the following definition: ‘A system is a set of
objects together with relationships between the objects and between their attributes.’ No clue
is given to where did the objects come from. We know that they come from the mind of the
observer. While they emphasise the relationships as essential parts of a system concept, they
fail to note that the system itself is relative to the view point of an observer.10
If systems are sets of things, set theory and its notations would be of great convenience. The
mathematics of sets (set theory) tells much about the properties of sets, but nothing about how
to choose them.
10
Each observer type would choose its own set of object types and hence we need different points of view for
different observers. The choice is based on the observers’ own expertise.
11
In Aida2, language designers do not talk about what the elements are, we just talk in general.
12
In aida2, we will use math once we have the intuitive feeling of things and the maths will be used to simplify.
It has to make sense first.
35
Ola Larses and Jad El-khoury
Views on General System Theory
We may be either not aware or interested in certain possible observations or the resolution
levels.13
A complete observation by an observer consists of one selection from each set in his scope.
How many possible observations can be made? The set of all possible observations is the
product set (Cartesian product) of the observer’s range sets. The product set may be too broad
a model for the observer, where even though he can make each of the component
discriminations, he may not be able to make all combinations. Certain elements in the product
set may need to be excluded for a more precise description of the observer. By including such
a broad characterisation of the observer, we are committing an error of assuming that the
observer can observe things he may not be able to do. On the other hand, if we properly
characterise his scope and the grain of each component, then the cross product model would
at least not exclude any observations he might make. In this model of the observer, we shall
remind ourselves how much computational capacity our model requires (Square law of
computation).
We have no requirement that the observer be able to make individual observations ‘correctly’,
because the observations are our primitive, undefined elements and the word ‘correct’ applied
to them is meaningless. All he must do is to recognise 2 sensations as being ‘the same’ and he
is the final arbiter.
13
In aida2, an observer is replaced by a designer. An observation is equivalent to a particular design decision he
can perform. This decision comes from the set of all possible decisions for that designer. Each decision type is
concerned with a particular type of decisions.
36
Ola Larses and Jad El-khoury
Views on General System Theory
not how close it is to them and each can tell if the penny is on or off the table. A and B can
agree if the penny is on/off the table. If the penny is on the table, we cannot predict what A
says from that of B, and vice versa. If we use their observations properly, each will make a
contribution to our understanding of where the penny lies.14
We have been assuming a special position for ourselves, our point of view. It is easy to slip
into imagining that we can get ‘above the table’ when talking about other people’s
viewpoints, but we really have no reason to believe that we have such super powers of
observations. For simple cases, we can talk about different points of view if we are willing to
introduce an explicit fiction – the superobserver. The superobserver needs to have enough
viewing capacity which covers the abilities of the other observers (but not more). He needs to
be able to dominate all other involved observers. A superobserver’s view must dominate the
view of every other observer present. This dominance can be assured if the superobserver’s
set of observation states is the Cartesian product of all of the others (It is all possible
combinations of observations.). But, in many cases, only a subset of these combinations is
needed. The superobserver powers, though finite, grow much faster that the other observers.
Combinatorial growth is a critical flaw in any discussion of multiple points of view, for
though we can imagine that a superobserver might exist, there is little chance of having one in
complex situations. We must particularly refrain from imagining that WE are the
superobserver, capable of seeing what ordinary mortals cannot.15
14
In Aida2, we generalise the concept to involve types of observers. Each type of observer/designer/discipline
makes his set of observations and there may be overlap in common observations/properties. Common
observations need to be consistent. However, each observer type dominates the other in one way but is
dominated in other ways. If we use these observations properly, we can learn more about the system. Combining
these observations, we can reach further understanding of the system (emergent properties) that neither observer
could have contributed to. This is the role of x-disciplinary analysis.
15
This is a good argument against having a single super-model that has all properties as opposed to having
multiple-models for each observer/discipline. In aida2, we refrain from thinking or having a super discipline.
Every thing is always distributed. An analysis view might combine parts from different views but is not seen as a
superobserver of these views, since it only takes a small set of properties from each view.
37
Ola Larses and Jad El-khoury
Views on General System Theory
24x24 members. Therefore, there are 256 possible sequences of length 2, or (24 power n)
sequences of length n. A superobserver would need a super memory if he’s to remember
everything he sees, or else be lucky and see highly constrained sequences. Being constraint,
one can use compact means of recording the observations as a mapping from one observation
to the one that follows it. If the mapping is not one-to-many, there are no ambiguities, and the
behaviour of the box is predictable after just one observation.
So far, you have been a passive observer. Although you are omniscient (all seeing), you are
not omnipotent (all powerful). As a superobserver, you have no power at all (impotent). You
have been playing a game called ‘black box’, where the observer is impotent to manipulate
the box by looking ‘inside’. The black box models an observer who cannot or will not
influence the system to be investigated, such as an astronomer studying the universe. Human
beings often interact with the systems they observe. We may believe the world to be
independent of the percipient observer, but we definetly feel it depends on the participant
observer.
Let us suspend the black-box rules, and endow you with limited powers of interaction. You
touch a spring and a little door opens. Inside the door, a sign says ‘KICK ME’. So, you give
the box a tap. Instantly, the pattern of light and sound changes and we see … g m d f g m d f
… You then give the box a bolder kick and get … b j r c q h p l o e b j r … Further kicking
fails to produce any other behaviours than the three already produced.
The box is further examined independently by two other observers. A friend, upon observing
the box, concludes that it only changes between two sequences of states, and that it is not
determinate in its behaviour. A stranger, the inventor of the ‘music box’, agrees that there are
three sequences, but that the last sequence is 5 states long.
To settle the disagreements, all observers walk into the room together. There, the inventor
explains that the music box plays three national anthems, and to get the music box to change
tunes, all you have to do is to ‘kick it’, which actually means ’yell at it’. When he
demonstrates this, everyone declares in unison: ‘See, it works just the way I said’.
38
Ola Larses and Jad El-khoury
Views on General System Theory
see an A or a D state after state A. But, if the inventor can remember the previous two states,
he can predict the next state. The substitution of mental capacity for observing power is an
illustration of a general law about observers, called the Eye-Brain Law: ‘To a certain extent,
mental power can compensate for observational weakness.’ Through symmetry, the Brain-
Eye Law is: ‘To a certain extent, observational power can compensate for mental weakness.’
As examples of the Eye-Brain Law, the experienced doctor needs fewer laboratory tests than
the intern to make the same diagnosis, but the intern can substitute a good laboratory for the
years of experience lacking. This Law will not work if there are no constraints at all on the
observations. Memory is of no use unless the future is like the past.
Although the superobserver sees the fine details, he may miss the big picture. See-it-all does
not mean know-it-all, for knowing means knowing how to ignore certain details.
A state is a situation that the observer can recognise if it occurs again. Discriminating too
many states has been defined as undergeneralisation. Scientists are envisioned making the
most precise measurements as a basis of theories, but in practice, they are lucky that the
measurements are not overly precise. Newton based his law of universal gravitation on certain
observations, which if they were made more precise (as precise as can be made today), his
work would have been much more difficult. The balance between ‘eye power’ and ‘brain
power’ cannot be pushed too far in either direction. The problem of science is to find the
appropriate compromise.
39
Ola Larses and Jad El-khoury
Views on General System Theory
f(M, m, r) before he could give the exact form of the gravitational forces between two masses.
Functional notation can be mixed with explicit formulas to show ‘intermediate stages of
knowledge’, for example F = g(m, M)/r2. Similarly, incomplete knowledge can be denoted as
F = h(m, M, …)/r2.
Given a function T = f(W, I, t, D), we might desire to further refine the model by expressing
D as a function of other quantities, even if we do not know what these quantities are. We can
use functional notation to show our intentions to do so, by writing D = g(…).The form T =
f(W, I, t, D) implies that the quantities in parenthesis are independent, in the sense that for the
given discussion, we are not interested in the functional dependencies of these quantities, even
thought they can themselves be dependant on other quantities. On the other hand, functional
composition such as T = f(W, I, t, g(…)) shows the deeper levels of dependencies in a
compact form.
The notation of decomposition of functions is appealing, since science explains by reducing
one phenomenon to the terms of other phenomena. There are two main fallacies that can be
committed during this reduction. A scientist may commit a Fallacy of Incompletness by
omitting some quantity from one of the functional relationships, such as stating z = f(x, y)
when it really is z = f(x, y, …). Secondly, even if the view is complete, the reduction must
eventually stop either because of the limited capacity of the observer, or because the situation
will not admit further reduction. Such a limit leads to a situation called ‘complementarity’ of
observation.
40
Ola Larses and Jad El-khoury
Views on General System Theory
Two models that fit all observed data are said to be ‘isomorphic’. (Mathematically, the two
models would have to fit all ‘possible’ data, but we use the term in a more limited sense.)
Black box observation, once it no longer yields new observations cannot resolve this
isomorphism unless the box is opened. And opening the box means decomposing/reducing
one step further (Until one cannot reduce no further and the problem of complementarity
between the given isomorphs arises).
At any level of observation, the choice of isomorph is strictly up to us depending on our
memory capabilities, previous knowledge, etc. This arbitrariness ensures that different
observers will have a multitude of ways in which to interpret their observations, not just
because of the different choice of isomorph, but because of what is to be observed as of
primary importance.
41
Ola Larses and Jad El-khoury
Views on General System Theory
that the system has a 20 state cycle. However, having been drugged by the inventor, you are
only capable to remember the last 10 state transitions of the system, and hence are no longer
able to see the cycle. In order to deal with this problem, you decide to narrow your view and
only observer the two lights, since that reduces the number of states to 4 and hence increases
your chances to remember the complete cycle. You now succeed in seeing the state
determined, but smaller system. You try also to only focus on the tones and succeed in
observing the complete cycle of 6 states. You have actually invented a new way of looking at
the world, by decomposing the system in two independent parts, each of which is state
determined, at the expense of having to deal with 2 smaller systems instead of one. But given
that you have limited mental powers, the decomposition into independent parts enabled you to
predict the system behaviour better. Page 134: This is the method of science, which would be
unnecessary has it not been for our limited brains.
Can a system’s behaviour always be decomposed into independent parts? This depends on the
set of qualities being observed. A more appropriate set of qualities of the same system, might
allow a better decomposition, and hence a more simplified view, of a system. A physicist
recognises entropy and density; the chemist, valence and PH; or the economist profit and
marginal utility. Depending on the choice of properties, your understanding and grasp of the
system will hence vary.
17
In Aida2, we will use the tools described later in this chapter, as common experiences for communication
between engineers.
42
Ola Larses and Jad El-khoury
Views on General System Theory
assumed known and left undefined. These ultimate reductions must be rooted in observation
of the world. Page 143: by examining the metaphors of science, we can learn about the limitations of
the brains that do science.
43
Ola Larses and Jad El-khoury
Views on General System Theory
Qualities have a mental function for observers with limited memory. We may think that
certain qualities are more ‘natural’ than others, but this actually means that we are more
accustomed to observing in those terms, since it has been found more useful to observe in
those terms. As we work in less familiar situations, our learned capacities become less
effective. A quality is a way of grouping states of a system. For example, the quality of mass
is defined by the states in which masses are the same or different. The ‘sameness’ and
‘difference’ operations allow us to identify and explain the quality in question. If we want to
‘measure’ the mass quality, we have to introduce another operation besides ‘sameness’ and
‘difference’ for states, such as ‘greater than’.
Physical scientists differentiate between ‘extensive’ and ‘intensive’ qualities depending on
what happens to the quality when the system is divided into parts. An intensive quality is one
which maintains the same quantity after the system is divided such as density, while an
extensive quality depends on maintaining the full extent of the system, such as mass. These
concepts are defined ‘relative to some act of breaking’ of the system. For example, the density
quality of a chocolate block is extensive when related to the act of cutting the block in half.
However, if the block is divided into its chocolate and peanut parts, then the densities would
be different. The definitions of intensive and extensive qualities can be turned around to give
a definition of ‘breaking into parts’: ‘If the intensive properties remain the same, then you
have probably broken the system.’.
More generally, if the chocolate block is divided into the qualities of flavour and consistency,
then neither part have any density at all. ‘Breaking into parts’ can be generalised into
‘transformations’, and we can hence derive the more general Invariance Principle: ‘With
respect to any given property, there are those transformations that preserve it and those that do
not preserve it’, or restated in terms of transformations: ‘With respect to a given
transformation, there are those properties that are preserved by it and those that are not’.
A quality may be characterised by the transformations that preserve it, or a transformation
may be characterised by the properties it preserves. In general, we cannot say precisely what
we mean by a certain quality because there are an infinite number of possible transformations
that can be performed. The Principle of Invariance can be restated: ‘We understand change
only by observing what remains invariant, and permanence only by what is transformed.’
Partitions
As an example of the division of a system into parts, consider the act of dividing the system’s
behaviour into qualities. A partition is defined by a set of ordered pairs of the parts (states), in
which the relation between each pair describes the relation ‘has the same value of the quality’.
Hence, the set consists of the Cartesian product of the set of states that has the given value of
that quality. In order to define a ‘sharp’ partition, there exists three mathematical conditions
that must be satisfied.
Clearly, a quality will not satisfy our idea of a quality if we cannot consistently identify it
with a particular state. If a state x does not have the same value of the quality as the last time
that state was observed – the pair (x, x) is not in the set – then the whole idea of quality breaks
down. This leads to the mathematical condition for describing a partition, and thus describing
a quality: ‘For every state x, the pair (x, x) must be in the relation. This is the ‘reflexive’
condition. When the partition describes a quality, it means that the quality cannot shift back
and forth with time while the state remains the same. If we start with the idea of qualities, we
identify states by the shifting of quality values. The reflexivity condition prevents us from the
erroneous absolute thinking, of taking a relative quality as an absolute one. For example, if we
attempt to divide a village into groups of ‘cousins’, the attempt is faulty because this property
44
Ola Larses and Jad El-khoury
Views on General System Theory
is a relation between two parts, and not an absolute property. This error is exposed when we
notice that one is not a cousin of oneself.
The second property a relation must have to fit our intuitive notion of a quality is ‘symmetry’.
If state x has the same quality value as state y, the it should also be the case that state y has the
same quality value as state x. Consider trying to partition a village into groups of ‘friends’,
and assuming that reflexivity is satisfy by assuming that one is one’s own friend. In certain
understanding of the ‘friend’ quality, even if A considers B as a friend, it is not necessarily the
case that B considers A as a friend.
The third condition is that of ‘transitivity’. Considering a symmetric definition of the ‘friend’
property. Even if A is a friend of B, and B is a friend of C, then for transitivity to hold, then A
must be a friend of C, which is not necessarily the case. Transitivity may not hold with
qualities involving graininess, that is qualities in which the sensing device has a minimum
resolution level under which it cannot detect differences between two values. For example, if
A is slightly more blue than B, and B is slightly more blue that C, such that a human cannot
detect any of these differences, while the difference between A and C is noticeable, then an
observer may classify A as the same colour as B as well as B as the same colour as C, but A
as not the same colour as C. Resolution levels are part of any measuring process. With
graininess transitivity may not hold, and hence there is no complete partition, no clear
division of system into subsystems, no clear separation of system and environment.
State Spaces
When dealing with systems with a large number of states, new tools, other than drawing the
different states and the movements between them, are needed to represent them. If the system
can be composed of two qualities, we can draw the states using Cartesian coordinates, where
there is a place for every state, and every state has its place. Note that the assignment of state
of a two-variable system to points in a physical plane is arbitrary. Certain arrangements may
45
Ola Larses and Jad El-khoury
Views on General System Theory
appear to yield a continuous line of behaviour, but we should remember that this appearance
is a consequence of our assignment of numbers to quality values. Of course, we may want to
find such an assignment so as to reduce our effort at describing the behaviour. Generally, a
system with n qualities can also be mapped onto an n-dimensional space, called a ‘state
space’. Operations such as ‘projection’, ‘sectioning’, ‘dimensional reduction’ can be
performed on such a space producing a ‘section’, ‘project’ or ‘subspace’ respectively. Such
operations may be useful in order to reduce the system complexity or the mental power
needed, at the expense of loosing certain information.
The opposite of projection is that of ‘expansion’, in which a new dimension is added. This
may be necessary when studying a system, after discovering that a certain variable is missing.
Old work needs not be thrown away, since it becomes a projection of the new set of
dimensions. The need for a new dimension may be discovered when realising that the line of
behaviour of the system seems to cross itself, which indicates that something is wrong with
our point of view if the system is to be state determined. A crossing represents two different
paths emanating from the same point. Note however that in a projection, a crossing poses no
problem. For behaviour represented in a state space, we have the Diachronic Principle: ‘If a
line of behaviour crosses itself, then either the system is not state determined, or we are
viewing a projection – an incomplete view.’.
Projections and other transformations help us overcome the limitation of our brains to handle
many dimensions. Another way of handling the surplus of dimensions is by introducing the
dimension of time. Time has the property of always moving in one direction, and hence
ensures that no cycles or crossings occur. The addition of time also allows us to project each
other dimension onto a two dimensional graph as a function of time. Page 196: If you cannot
think of three ways of abusing a tool, you do not understand how to use it.
46
Ola Larses and Jad El-khoury
Views on General System Theory
Given that the behaviour of an open system is influenced by inputs from the environment, as
well from its own behaviour, an observer cannot tell the reason for certain behaviour, unless
he has white box knowledge of the system. This leads to the principle of Indeterminability:
‘We cannot with certainty attribute observed constraint either to system or environment’
We prefer to think and create our systems to be as closed as possible. Openness complicated
prediction and observation, yet at the same time, it lets us gain predictability by allowing us to
act on the system.
18
In aida2, there should be really no precedence or favouritism between the structural and behavioural models of
our systems.
47
Ola Larses and Jad El-khoury
Views on General System Theory
on an observer need not be changed’ or ‘A way of looking at the world may be changed to
reduce the stress on an observer.’19
19
In Aida2, we adopt the principle that we want to reduce the stress on the observer, by choosing view points
that best fit their view of the world.
48