The 30th International Conference of the System Dynamics Society
A Critical Review of the Criticisms of System Dynamics
Charles R. Featherston and Matthew Doolan
Abstract
This paper presents a review of the criticisms of system dynamics and assesses the validity of these against recent findings in
the field. The authors survey the literature critical of system dynamics and review their criticisms using the current
understandings in the system dynamics field. This work suggests that there are some pertinent criticisms that have been aimed
at system dynamics. These include the apparent disagreements regarding the role of historical data in model confidence
building, system dynamics' reductionist perspective and how system dynamics addresses plurality and hierarchy. Overcoming
these criticisms require the ever present need for education, communication and theoretical work. It is hoped this paper will
strengthen the mandate of system dynamics in the eyes of its critics, assist and improve the field and its general acceptance as a
tool of analysis.
1 Introduction
It is important that a field of research address its
criticisms in order to understand, and help others to
understand, its limitations, to strengthen the field of research
and improve its general acceptance. Either by rebutting a
criticism or redefining it in light of a well founded criticism,
theories become stronger, more robust and it improves their
chances of being accepted by a more general audience.
This paper is a review of some criticisms that have been
levelled at the field of system dynamics and explores the field
with respect to these criticisms. The paper will take an
detailed look into several criticism, evaluate their validity and
evaluate the measures that have been taken by academics and
practitioners within the field to address the criticisms.
To build a critical analysis of the paradigm of system
dynamics, we must understand the context of the field.
System dynamics is a 'means of inquiring into the behavior of
part of the world in order to understand it and hence indicate
ways of improving its performance’. (Keys, 1990, p.480). The
paradigm is one of many fields that can be used to try to
understand the complex nature of the systems in which we
work. System dynamics has roots and relationships with a
number of diverse fields, including: systems thinking, servo-
mechanism theory, dynamical systems theory, cognitive
science and history (Richardson, 1991; Sterman, 2000;
Meadows, 2008; Newell, 2012).
This paper is specifically looking at criticisms of the
fundamentals of the system dynamics paradigm and not
specific criticisms of technique, specific content theories of
system dynamics or how these criticisms apply to other
approaches often used instead of system dynamics. For
example, Rouwette et al. (2011, p.1) claim there is ‘no clear
evidence for the effectiveness of group model building, and a
conceptual model linking elements of the modeling process
to goals is missing’. This is a flaw in a technique used in
system dynamics, not of the whole paradigm itself and as
such possible ‘gaps’ like this will not be included in the
analysis. The later exclusion, that of other approaches, could
prove a valuable extension to the body of systems research,
but is precluded from this paper.
This paper's main contribution to the field of system
dyanmics is to gather and review many of the criticisms that
have been made of system dynamics. During the life of the
paradigm, system dynamics has been the focus of a number
of criticisms. However, there have been few attempts to bring
a large number of the criticisms to bare together and assess
them as a collective. This paper aims to bring many of these
criticisms together and use the literature and an
understanding of the paradigm to address them, identify
which are valid and identify those that remain unaddressed.
This paper is also designed to stimulate discussion and
more constructive criticism of system dynamics. System
dynamics is a field that is not widely taught in schools and
colleges (Forrester, 2007). As a non-pervasive field in
education, it is possible that many critics of the paradigm
would ignore it rather than prepare and deliver structured
criticisms. It is hoped that this paper will stimulate critical
debate of the field to compensate for a possible lack of
enthusiasm from its external critics.
This paper will deal with five 'groups' of criticisms. These
are listed below. Many of the ‘areas’ of criticisms contain an
array of different criticisms that have been grouped to deal
with common elements simultaneously. There are many
common theoretical threads and challenges for the field that
link the groups of criticisms, which will be brought out in the
critical review and highlighted in the final discussion.
1. Applications of system dynamics
2. Mimicry of historical data and validation
3. Complexity
4. Determinism
5. Hierarchy
*Correspondences to: Research School of Engineering, College of Engineering and Computer Science, The Australian National University,
Canberra ACT 0200 Australia. Email: Charles.Featherston@ anu.edu.au
1
Charles R. Featherston and Matthew Doolan: A Critical Review of the Criticisms of System Dynamics
2 Applications of System Dynamics
One of the more prolific areas that generate criticism is
not of the paradigm itself per se, but are criticisms of how
system dynamics has been applied. These criticisms of the
application of system dynamics come from people from
within as well as outside the field. These criticisms range
from system dynamics being applied to the wrong situation to
criticisms of particular models' complexity, layout and size
(Forrester, 2007; Barlas, 2007).
It can be difficult to find published examples of poor
applications of system dynamics. This is generally for two
reasons. Firstly, many are not published. With the peer
review process in many journals, poor applications of system
dynamics, like poor journal articles, are rejected. The second
reason is that for a model to be bad, someone analysing the
model must know the system well and either analyse the
proposed model to find its flaws or be able to prove how the
assumptions or relationships within the model are fallacious.
2.1 Reasons for drawing criticisms
There seem to be four main reasons for these modellers
drawing criticisms, many of which are covered by Forrester
(2007) and Barlas (2007). Firstly, many of the examples of
system dynamics that generate criticism are because system
dynamics was applied to the wrong ‘type’ of problem
(Forrester, 2007; Barlas, 2007). System dynamics is designed
to explore ‘problematic behavior pattems caused primarily by
the feedback structure of the setting’ (Barlas, 2007, p.470).
Often however, system dynamics is applied to problems
where exogenous influences drive the system. In the words of
Barlas (2007, p.470) ‘many so-called SD [system dynamics]
modeling projects are about problems that simply do not have
SD [system dynamics] characteristics.'
Secondly, some modellers just apply system dynamics
incorrectly. System dynamics provides a set of tools and
techniques to apply to the appropriate problems (as outlined
above). However, some modellers misuse and mismanage the
tools of system dynamics. Forrester (2007) and Barlas (2007)
site the cause of this being the inherent difficulty of leaning
and applying the concepts of system dynamics. A claim
supported by Cronin et al. (2009) and Sterman (2010) and
their work with the understanding of the fundamental system
dynamics concept of accumulation. Forrester (2007) and
Barlas (2007, p.469) also site ‘no formal/clear accountability
for poor modeling’ as a possible cause of inexperienced
modellers publishing models that apply system dynamics
incorrectly and flout many of the paradigm's rules and
limitations.
Some people also have a different concept of what system
dynamics is. As a group of theories and techniques, system
dynamics can be seen as just a name applied to techniques
and a process used to produce models. As a consequence, a
modeller can call a process system dynamics in situation
where others would not agree. Equally, someone observing a
model can call the process used to get there system dynamics,
even if it was not employed by the modeller. An example of
this is Hayden's (2006) criticism of a model by Boyer (2001)
that purports to define system dynamics’ views on
constitutional order, institutions, organisations and
conventions. Radzicki & Tauheed (2007), question whether
Boyer even proposed the model reflected the view system
dynamics took to these facets of a system or even if it
reflected anything of system dynamics at all.
Finally, the tendency to build unnecessarily large models
to address ‘big' problems is another aspect of modelling that
draws criticism (Forrester, 2007; Barlas, 2007). Barlas (2007,
p.470) explores several reasons why large models is an issue,
stating that large models ‘are not only difficult to build, they
are also nearly impossible to understand, test, and evaluate
critically’.
Both inexperienced and experienced modellers draw
criticisms for their applications of system dynamics. Many
instances where system dynamics has been applied poorly are
done by practitioners with little system dynamics experience
or those that are leaming; we, for instance, have several
examples that belong to that group. These tend to be of poor
quality and Barlas (2007, p.469) notes that there are ‘too
many system dynamics models - published or applied - that
do not meet our minimum standards of quality’.
However, some more experienced modellers also draw
criticisms for their applications. Solow (1972), Marxsen
(2003) and Simon (1981), for example, criticise Meadows et
al.'s (1972) results published in the book Limits To Growth
(also see Schmandt, 2010). Many of the criticisms of
Meadows et al. (1972) were aimed at the different
assumptions about the real world, some of which were
clarified in the subsequent updates of the study (Meadows et
al., 2004; Meadows et al., 1992). Others, such as Simon
(1981), had fundamental differences in assumptions, which
lead to the study drawing his criticism.
Other criticisms, founded as they may be, do not apply to
system dynamics because they miss some of the basic
theories and limitations of the methodology. For example,
criticisms of models' inability to perfectly simulate reality,
miss the point that models are often simplifications of reality
used to understand behaviour modes. As Meadows et al.
(1972, p.21) wrote “The model we have constructed is, like
every other model, imperfect, oversimplified, and
unfinished.” These simplified and often unfinished works are
therefore difficult to compare to historical data, a point
discussed at greater length in the next group of criticisms.
2.2 Regressing the Problem and its Implications
To regress the problem of why there are poor applications
of system dynamics only brings us to some well understood
ideas within the field. The reasons for system dynamics being
applied poorly and for people to fall victim to the reasons
mentioned above are few, succinct and commonly known: the
phenomena that system dynamics tries to explain are
counterintuitive and training is needed for people to master
the field (Forrester, 1961; 1971b; Comin & Gonzalez, 2007;
Cronin et al., 2009; Sterman, 2010). However, there is one
more aspect that this research does not appear to cover: they
demonstrate a poor understanding of the field of system
dynamics.
This analysis draws us to two points. Firstly, that there are
examples of and many reasons for poor applications of
system dynamics, some of which quiet rightly draw criticisms
2
The 30th International Conference of the System Dynamics Society
and some that should. Secondly, that there are misplaced
criticisms on models that meet high quality standards in the
field of system dynamics. These problems were the cause for
Forrester’s (2007) call for greater (more & better) systems
education in schools and universities and for greater
promotion of system dynamics in the public sphere.
Note will also be made here of a criticism put forward by
Hayden (2006, p.534) that to my understanding has not been
addressed in other literature. Hayden criticises the generality
and unclear terminology used in system dynamics and its
models. Further education of people outside of the field of
system dynamics in the terms used in the paradigm could
help to address this. Criticisms in this area help to illustrate
how important further promotion and education in system
dynamics is for the paradigm.
3 Mimicry, Validity, Comparison & Prediction
The inabilities of models to mimic reality and predict the
future are common criticisms levelled at system dynamics
(Solow, 1972; Simon, 1981; Keys, 1990; Hayden, 2006).
Meadows et al.'s (1972) Limits to Growth provides an
example of this. Many critics, such as Simon (1981) and
Solow (1972), picked up on attributes of the model that lead
them to believe the model did not reflect reality,
disenchanting them towards the entire model and the
conclusions that were drawn from it. What was lost however,
was that the basic dynamics of the model still appear correct
today, regardless of its inability to reflect exact points in
historical time or to reflect the material wealth of the world
today (Meadows et al., 2004). Furthermore, if reflecting
history is not necessarily a requirement of the field, then how
do people know if the model is an accurate explanation for
the underlying behaviour? This reflects a group of damaging
criticisms that have been levelled at the system dynamics
paradigm: models not mimicking reality, comparisons of
models and reality, model verification and the dependence of
the paradigm on data.
3.1 Mimicry
Many criticisms of system dynamics are aimed at the
inability of the paradigm's models to mimic reality (Solow,
1972; Simon, 1981; Keys, 1990; Hayden, 2006). However, it
is relatively widely accepted within the field of system
dynamics that models are not designed to and cannot
perfectly imitate the real world (Sterman, 2000; Forrester,
2003; Lane, 2000). As such, to get a model that reflects the
actual system perfectly is not the goal of system dynamics.
Instead, the goal of modelling in system dynamics is to
assist people to understand the internal systemic structure of a
system that drives behaviour (Forrester, 1961; Senge, 1990;
Sterman, 2000). Forrester (1985) and Radzicki and Tauheed.
(2007) even propose that the process of generating the model
and leaming about the system could even be of more benefit
than the model itself. Their justification is that the process of
modelling promotes greater learning about the internal causes
and effects of systemic structure than the model on its own
would.
Criticisms of system dynamic's tendency not to mimic
reality appear to come from one of two areas. Firstly, from
3
the point of view that system dynamics is a ‘hard' systems
thinking perspective (Keys, 1990). Hard systems thinking
approaches, such as systems engineering and systems
analysis, tend to
operate in environments of low complexity and high
problem visibility. As a consequence they are designed to
mimic historical data very closely. System dynamics
however, is applied in situations of high complexity (with
varying degrees of problem visibility), making it standout as
a field that ‘doesn't work' because of the often inability for
model's inability to mimic historical data. Another possible
reason for its apparent separation from other ‘hard’ systems
methodologies is that because of its endogenous focus it often
does not exhibit the behaviour caused by external shocks
without the cause for the shock being explicitly included in
the system.
The second reason arises from poor understandings of the
goal of system dynamics: not to mimic or mirror the real
world but to use models to understand why certain behaviour
is occurring (Forrester, 2007; Radzicki & Tauheed, 2007).
This again arises from system dynamics not being understood
more widely. System dynamics is one of many fields that
tries to make sense of a complex environment. It does not
propose to uncover all there is to know about a system and
like the other techniques it has its own goals, limitations and
expected outcomes.
3.2 Model Verification and confidence building
Despite much of the learning coming from the modelling
process rather than 'the' model, models are still an important
part of communicating conclusions and testing their ‘validity’.
A model itself however, cannot be tested for validity. In fact,
the idea of verifying a model is fallacious (Sterman, 2000).
As Sterman (2000) points out, 'no model can be verified.
Why? Because all models are wrong...... all models, mental
or formal, are limited, simplified representations of the real
world’.
Many researchers believe that building confidence in
models is the central method of verifying a model (Radzicki
& Tauheed, 2007; Sterman, 2000; Forrester & Senge, 1980).
Confidence building in system dynamics is a method of
verifying a model ‘along multiple dimensions' (Radzicki &
Tauheed, 2007). Sterman (2000) points out that Popplerian
philosophy tells us while we can't establish if a model is
correct, we can establish that a model is false. We can then
alter the model to form a modelling version of an auxiliary
hypothesis which we can then test. By subjecting models to a
series of tests we can slowly build confidence in it: the more
tests it passes the more confidence we have that the model
reflects the correct behaviour. Peterson (1975, Appendix B)
provides thirty-five informal tests that models can be
subjected to build confidence. Similarly, Sterman (2000)
provides ten such tests.
However, testing a model against historical data is still
important for some modellers. For some, consultants in
particular, comparing a model to historical data is the most
important test of the model (Homer, 1997). For others,
comparing the model to historical data is still considered one
of the tests for building confidence (Sharp & Price, 1984;
Sterman, 2000). One of Sterman's (2000) ten tests is the
Charles R. Featherston and Matthew Doolan: A Critical Review of the Criticisms of System Dynamics
Behavior Reproduction Test, which compares the model's
numerical behaviour to past data (while at one point
qualitative behavioural testing is proposed, Sterman does not
discuss this point any further). However, Sterman (2000)
does state that fitting the data does not mean validation and
that the Behavior Reproduction Test is to uncover flaws and
structural issues with the model.
Sterman's (2000) focus on historical data (a decent
portion of the section dedicated to model testing) seems to
differ somewhat with some other system dynamicists.
Forrester (2003, p.5) claims that ‘the dynamical character of
past behaviour is very important, but the specific values at
exact points in historical time are not’. Barlas (2007) supports
this by purporting that ‘proper measures of historical fit
would stress fitting past dynamical patterns, such as periods,
amplitudes and trends' (p.471). Keys (1990, p.488), after
some discussion concludes that, 'model validity should be
assessed relative to the purpose and not to a universal
measure of correctness’. All of these appear to contrast with
Sterman's (2000) Behavior Reproduction Test, which mostly
espouses ‘fit’. Even when there is only a variation in the bias
equation (U™) of Theil's Inequality Statistic, Sterman (2000,
p.876) still claims that a systematic error should he ‘corrected
by parameter adjustment’.
The disagreement over the role of historical data in model
validation makes Forrester's (2001) claim that more work
needs to be done in the field to establish methods of model
validity still pertinent. Ultimately, Keys (1990, p.488) claim
that models ‘should be assessed relative to the purpose’ is
appropriate. It also seems generally accepted in the field that
comparing a model to historical data is one of the least
powerful methods for building confidence in the model
(Forrester & Senge, 1980; Saeed, 1992; Radzicki, 2004;
Radzicki & Tauheed, 2007). However, there must be a use
for historical data in to building confidence in a model that is
supposed to emulate it.
Perhaps one of the more significant and overlooked tests
is the qualitative assessment and comparison of a model's
behaviour with historical data. As stated earlier, Sterman
(2000) recommends qualitative comparison, but does not
draw out the point any further. Peterson (1975, Appendix B)
offers several tests that use historical data, but only
qualitatively compares the relevant behaviours (although
some could progress towards quantitative measures, Peterson
does not explicitly include this extension in his test).
Qualitative tests such as these could be used to address
Forrester’s (2001) and Barlas' (2007) assessments of model
data with general pattems of behaviour.
Some of the criticisms, mostly originating from the ‘hard’
systems thinking approaches, claim that system dynamics
generates models that have trouble matching historical data
(Keys, 1990). As has been shown, researchers in the field still
disagree about the role of historical data in building
confidence in a model. Many researchers appear to agree that
matching historical data exactly is not the aim of system
dynamics, but this seems to be a source of much criticism. It
seems researchers in the field need to ensure their critics are
better informed about the theories they espouse.
3.3. Prediction and Prophecy: Determinism
When grouping system dynamics with other ‘hard’
systems thinking techniques, a common accusation of the
field is its determinism and the accompanying tenet that it
can predict or prophesises the future (Ansoff & Slevin, 1968;
Sharp & Price, 1984; Jackson, 1991; Lane, 2000; Forrester,
2001). Many people disagree with this proposed capability;
indeed it makes many others feel uncomfortable.
The complexity in the argument comes from the partial
adherence of system dynamics to ‘hard’ system
methodologies. Determinism is often considered a
characteristic of ‘hard' systems thinking approaches
(Checkland, 1978; Lane, 2000). For situations with low
uncertainty and relatively low complexity, hard
methodologies, such as cybemetics, have been employed to
mathematically model the system to predict behaviour. As
system dynamics adopts some of the characteristics of ‘hard’
systems thinking, many see this as also taking a deterministic
view of the world. Many however, believe that system
dynamics cannot be as deterministic as other ‘hard’
approaches because of the complexity of the systems it
attempts to deal with. As Hayden (2006, p.539) states, ‘Social
systems are much too open, irregular, and dynamic for a
mechanistic theory to apply’ (this statement contains a variety
of criticisms which will be decomposed gradually throughout
the paper).
Forrester (1968) and Coyle (1986) counter determinism
by arguing that system dynamics is concemed with the
structure of the system under examination and the structural
reasons for the broad behaviour of the system. These
observations in isolation excludes the use of dynamic models
to observe the implications of actions on systems, for
example, in the Beer Game. This would encourage only
setting a model in action and seeing where it tended to find
equilibrium and not using it to test pulses or shocks to the
system.
Perhaps Lane's (2000) invocation of Popper's (1945a;
1945b; 1957) view of determinism is more appropriate:
‘prophecy is sharply distinct from that of
technological/scientific prediction" (p.7). Popper (1945a;
1945b; 1957) defines technological/scientific (here on
scientific) prediction as being conditionally dependent on the
assumptions; if one of the assumptions changes then the
prediction becomes invalid. From a system dynamics
perspective, models and any of their predications have similar
limitations. Lane (2000) even draws attention to early writing
in system dynamics, (such as Forrester, 1961) to demonstrate
that a deterministic view was never really espoused by the
field. Perhaps such a description may not even apply to many
of the so-called ‘hard' systems thinking approaches.
Forrester’s (1968) and Coyle's (1986) views are not
necessarily divergent to Popper (1945a; 1945b; 1957) and
Lane's (2000). Understanding the system is one of the main
goals of system dynamics, key to Forrester's (1968) and
Coyle's (1986) counterarguments. It is beyond understanding
the current structure and cause of behaviour and into
scientific prediction, where the dynamic aspect of
assumptions and conclusions are more uncertain, when
Popper's (1945a; 1945b; 1957) and Lane's (2000)
counterarguments becomes important.
The 30th International Conference of the System Dynamics Society
3.4 Tipping points
A note can be made here about the idea of tipping points.
Both Sterman (2000) in his explanation of the Susceptible-
Infectious-Recovery (SIR) model and Morecroft (2007) in his
explanation of models of fisheries, use the term tipping point.
Tipping point in these instances refers to the shift from one
feedback loop being dominant to another and a shift in the
behaviour of the overall system and Sterman's (2000) and
Morecroft's (2007) examples demonstrate that the techniques
used in system dynamics are capable of simulating these
shifts.
However, there are other shifts that system dynamics may
not be able to simulate as well. As shown above, system
dynamics is only a methodology of scientific prediction,
limited by the assumptions made in ‘foreseeing' the
implications of action taken on the system. Sometimes an
event or change may occur in the system that results in new
feedback loops forming and becoming dominant; a shift in
the structure of the entire system. Often, revisiting a model,
like one does with their mental models, is the only way to
understand a change in system structure, though this often
occurs in retrospect of the structural change.
Perhaps scenarios could be helpful to simulate these
structural shifts. 'Scenarios' have been used to simulate shifts
in parameters and observe their effect on the resulting
systemic behaviour (Zagonel et al., 2011). These can
simulate possible shifts in feedback loop dominance within a
system before the event occurs (Morecroft, 2007; Sterman,
2000). However, the situation is different when addressing a
shift in the structure of the system. Maybe scenario planning,
that envisions entirely different system structures and not just
scenarios in the form of parameter variations, could be used
to build a methodology that considers these potential
structural shifts and uses system dynamics to help understand
potential systemic behaviour.
3.5 Implications
These criticisms are damaging for system dynamics as
they negatively affect general opinion of the field and its
validity. By demonstrating that the model upon which
conclusions are drawn does not reflect historical data, people
who aren't versed in the particular theories and limitations of
system dynamics begin to believe that the field offers little
value. That is, if people believe that system dynamics tries to
mimic and even predict the real world, and are shown that
model that are proposed as system dynamics models do not
do this, then they may tend to believe it is the paradigm at
fault and not the criticism of prediction itself.
4 Complexity: Richness, Reductionism, Pluralism &
Social Systems
From a ‘soft’ systems perspective, it is argued that the
dependence of system dynamics on quantitative data and
explicit relationships does not allow system dynamics to deal
with the complexity of the real world and reduces the
richness of analysis it can conduct. (Keys, 1990). ‘Soft’
systems thinking and approaches often deal with more
5
complex environments than ‘hard' systems thinking
(Checkland, 1978; Jackson & Keys, 1984). To deal with this
complexity ‘soft’ systems thinking relies on qualitative
information and linguistic terms to describe complexity
(Checkland, 1978; Keys, 1990). Many believe that this level
of detail cannot be caught by the hard data and mathematical
models that are used in system dynamics.
In reply to this criticism, Keys (1990, p.489) argues that
[tlhe use of causal loop models is a movement towards the
soft systems type of model but the reliance upon a single
model remains a basic difference between [system dynamics]
and soft systems methodologies’. This counterargument only
goes some of the way to answering to the criticism.
Perhaps a better approach to addressing this criticism is
by emphasising that the model itself is only a portion of what
the system dynamics proposes it can do. As stated earlier,
system dynamics is essentially a leaning tool and the
‘process' of modelling is often seen as more important than
the model itself (Forrester, 1985). The process of modelling
involves information transfer of a rich linguistic and
qualitative nature that ‘soft’ systems proponents believe
system dynamics lacks. Moreover, it assists the field to deal
with increasingly complex situations, similar to those
addressed by ‘soft’ systems thinking.
4.1 Reductionism
System dynamics has often been accused of being
reductionist (Keys, 1990). Reductionism - the act of
describing a system through only discussing the interactions
of its parts - has generally been deemed inappropriate for
social messes (Ackoff, 1974). 'Hard' systems theories use
reductionism to create laws and rules that define how the
system operates (Keys, 1990). 'Soft' systems methodologies
however, which are designed to deal with social messes, do
not take a reductionist perspective (Jackson, 1982). As
system dynamics appears to break down the system to
understand how its components interact - essentially using
reductionism - many perceive system dynamics as unable to
deal social messes (Keys, 1990).
Rather than these criticisms taking issue with the
reductionist nature of system dynamics - the breaking down
of the system into nodes in a model - the issue seems to be
with using these nodes to construct the system that simulates
the behaviour under examination. However, the ‘reductionist’
criticism assumes that a reductionist hypothesis implies a
constructionist hypothesis, an assumption disproved by
Anderson (1972).
The constructionist criticism is however, harder to
disprove. Anderson (1972, p.393), states that the
‘constructionist hypothesis breaks down when confronted
with the twin difficulties of scale and complexity’. However,
it is sensible to conclude that in many situations a limited
increase in the scale and complexity can allow one to
construct the system out of the basic theories, just as
Newton's theories of mechanics can be used to construct
Kepler's laws of motion. As Anderson (1972) points out, to
do this, one must have an understanding of the functionality,
structure and goals of the ‘higher hierarchical’ system (or
theory, as is the case in the above example).
Charles R. Featherston and Matthew Doolan: A Critical Review of the Criticisms of System Dynamics
System dynamics essentially does just this. It uses
complex basic principles to describe complex systems.
However, constructing the system, or rather a model of the
system, from these principles requires an understanding of
the structure of the system in question, precisely where the
paradigm proposes to start. Furthermore, as outlined above,
the techniques involved in system dynamics can only be
applied to the ‘right’ type of problem. While it seems clear
that more work needs to be done on the theoretical
foundations of system dynamics in this area, it is apparent
that part of the notion of ‘the’ right problem is one that can be
analysed using the right amount or types of reduction.
4.2. Pluralism
The notion of pluralism affects two distinct acts in system
dynamics. Firstly, in the form of multiple perspectives of the
complex decision problem, the goals of system intervention
and different perspectives of the system itself. Secondly,
pluralism in the way individuals behave differently within a
system.
System dynamics, as Forrester (1961; 1969; 1971a)
defined it and its approaches (otherwise known as
Forrestarian system dynamics - the system dynamics of
MLLT. in the 1960's and 1970s), is accused of not dealing
with pluralism (Keys, 1990). What ‘type’ pluralism Keys
(1990) is referring to is not immediately evident. However,
when referring to ‘hard' systems thinking, Jackson & Keys
(1984, p.476) state that '[a] set of decision makers is pluralist
if they cannot agree on a common set of goals and make
decisions which are in accordance with differing objectives.'
That is, 'Hard' systems methodologies do not consider a range
of perspectives on goals, ‘the’ problem or 'the' system offered
by the relevant people. The idea that multiple perspective
should be considered is known as weltanschauung, a term
used frequently by Checkland (1981; 1987).
If this is the ‘type’ of pluralism Keys (1990, p.485) is
referring to then it seems he believes that Forrestarian system
dynamics makes no greater attempt to deal with plurality of
perspectives as any other ‘hard’ systems thinking approach.
However, Keys (1990) sees the introduction of influence
diagrams into system diagrams in the 1980's, by a group at
Bradford University, as a significant step towards dealing
with pluralism (see Wolstenholme, 1982; 1983; Coyle, 1983;
for a brief history of causal mapping see Sharp & Price;
1984). Keys (1990) believes these diagrams allow system
dynamics to accommodate many perspectives on ‘the’ goals
of intervention, the problem and the system in question.
More recently techniques, such as Collaborative
Conceptual Modelling (CCM) developed by Newell et al.
(2008), Newell & Proust (2009) and Newell et al. (2011)
further allows system dynamics to accommodate pluralism.
CCM uses a collaborative approach whereby people map out
their own perspectives of the systems and then slowly build
on this with other stakeholders or 'relevant' people to come to
a broad agreement on the structure of the system. Processes
such as these can however, be politically charged and it is the
responsibility of the system dynamicist to negate any of the
possible negative consequences that such processes can
entail.
Pluralism seems to be slowly being addressed by the
system dynamics paradigm. However, it seems that more
work needs to be done to ensure plurality is considered. In
situations where problems and system structures are hard to
define (wicked problems, Churchman, 1967), it is important
that people do not become subject to group think or narrow
avenues of thought in order to properly identify the goals and
system structures and assist the adoption of any
recommendations offered (GroRler, 2007).
From the perspective of a plurality of actors within the
system, system dynamics has the ability to model at an
aggregated level right down to the individual level. Osgood
(2009) states that while many studies are aggregated (gives
many example), some dynamics models need to model
individual behaviour. The level of detail depends on the
purpose of the model and the implications of individual
behaviour. Osgood (2009) proposes a model to assist with
modelling individual's behaviour in a more effective way,
bridging the gap between aggregation and individual
modelling and assisting system dynamics to consider
plurality. However, as simpler models are easier to
understand it is important that the correct level of aggregation
is selected.
4.3 Social Systems: 'Open, irregular & dynamic’
One of Hayden's (2006) central claims against system
dynamics is that the social systems that system dynamics
attempts to explore are ‘much too open, irregular, and
dynamic' and states that cybernetics - a very mechanistic
approach of analysing systems - is far to structured to deal
with such complexities (p.530). To explore this criticism, it
will be broken down into its components, beginning with the
proposed link between system dynamics and cybernetics.
Cybernetics
In their critique of Hayden's criticisms, Radzicki and
Tauheed (2007) refer Hayden to the work of Richardson
(1991). Richardson (1991) takes an in depth look into the
history of feedback thought in both the social sciences and
systems thinking. He identifies ‘two main lines of
development of the feedback idea.... the servomechanisms
thread and the cybernetics thread' (Richardson, 1991, p.1).
Richardson (1991) explicitly places the system dynamics
‘tradition’ in the servomechanism thread (see Richardson
1991 for more on the servomechanism theory and system
dynamics).
Open
Many social systems are open and subject to outside
influences. Hayden (2006) claims that system dynamics
cannot model this apparent openness in these systems and on
this point it appears theory in the field supports this claim.
System dynamics does not attempt to model the effect of
extemal behaviour on a system, instead it addresses the
behaviour generated intemally by a system; that is, it takes an
endogenous view of behaviour. It is the internal behaviour of
a system that often drives the system (Forrester, 1961; Senge,
1990). Richardson (2011) believes that it is this endogenous
perspective that is the fields greatest contribution to the study
6
The 30th International Conference of the System Dynamics Society
of systems. This endogenous view was behind one of the first
applications of servomechanism theory to economics,
conducted by Goodwin (1951).
The endogenous perspective may be confused somewhat
by comments such as ‘in reality flows are determined by so
many external things' (Hayden, 2006, p.534). When taken out
of context, Sterman's (2000, p.95) claim that 'the focus in
system dynamics on endogenous explanations does not mean
you should never include any exogenous variables' could add
to the confusion. However, as Sterman (2000, p.95-6)
clarifies, 'the number of exogenous inputs should be small,
and..... carefully scrutinized to consider whether there are in
fact any important feed feedbacks [involving the exogenous
input]'. This is how Sterman (2000) justifies the ability to use
historical data in some sense to test a model: without any
extemal inputs, the model's behaviour may be completely
different to the behaviour of the actual system.
Irregular
Social systems appear irregular. However, driving
humans' behaviour is a system of rules, obligations, controls,
regulations and limitations that is defined by them. This
appears to be a deterministic view, but as outlined earlier and
by Lane (2000), this view is not deterministic as it integrates
the ‘agency and structure’ that is common in many
contemporary social theories.
When taking this view is seems many of the irregularities
are removed. However, external factors still play a large role
in determining the behaviour of the system. Take, for
example, the Beer Game (Senge, 1990); the Beer Game
begins with a shock to the system, without which the
behaviour commonly observed in the game would never take
place. This is where the internal focus of system dynamics is
important. Once the intemal system architecture is
understood and the shock that caused the real behaviour is
understood, in an appropriate application of system
dynamics, the behaviour of the system is dictated internally
after the shock.
This perspective is supported by much of the thinking on
mental models - the mental constructions we have of the real
world. Real systems can be very complex and humans often
have difficulty accurately identifying the causes of certain
systemic behaviours due to factors such as time and spatial
separation of cause and effect and incorrect or limited
information (Piaget, 1928; 1930; Sterman, 2000; Sosna et al.,
2010). Consequently, what may appear irregular is actually
not, it is just the inability of humans to properly attribute
cause, the appearance of cycles and apparent inconsistencies.
Dynamic
Hayden (2006) also believes that system dynamics cannot
deal with the dynamic nature of social systems, but does not
elaborate further on this point. Much of a social system's
dynamics stem from the irregularities and openness of the
system. Both of these characteristics can cause a system to
fluctuate so much that would be difficult to understand the
underlying causes and pattems in the behaviour. However, as
has been shown, system dynamics uses limited openness to
understand the exogenous nature of the system. By limiting
7
its scope, it is providing a perspective that could help explain
the dynamic characteristics of the system. Furthermore,
imegularities in the system are often only apparent
imegularities with many suggesting that there is an
underlying order, as has been shown in the Beer Game
(Senge, 1990).
A different perspective on the criticism is that Hayden
(2006) believes that system dynamics cannot handle
dynamics or explore dynamic behaviour. As the goal of the
paradigm is to understand dynamic behaviour and as
dynamicists have provided many examples that the tools of
the field do handle and explore system dynamics, this seems
unlikely (Forrester, 1961; for examples of system dynamics
see references). However, without any further clarification of
Hayden's (2006) criticism, further discussion on this point is
likely to yield little.
5 Determinism: Dehumanising, 'Grand' Theory and
Austere
Somewhat linked to the previous discussion of
determinism from the point of view of prediction and
prophesy, system dynamics has also been accused of being
deterministic in the sense of dehumanising, aspiring to be
some ‘grand’ theory of systems and operationally austere
(Jackson, 1991; Lane, 2000). These differ from determinism
as it was previously discussed as the criticisms relate to an
accused imposition of system dynamics on humans and
theories. These are still considered as deterministic attributes
however, because these criticisms still imply the human
aspect of systems research be somewhat removed. Grand
theory is placed here because of its apparent disregard for the
variation between systems, particularly from the human
perspective.
5.1 Dehumanising
Jackson (1991) believes system dynamics is deterministic
and that it relegates people to ‘cogs in a system' and
disregards free will. Many instances of such criticisms view
system dynamics making the assumption that laws operate
outside of human subjectivity and of dehumanising its topic
(Lane, 2000). Lane (2000) sites Forrester (1961) and Bowen
(1994) to demonstrate that system dynamics is not as
deterministic as Jackson (1991) believes. As a detailed
analysis is given by Lane, only a brief account of the
counterargument will be given.
Forrester’s view is that system dynamics takes the
perspective that ‘decisions are not entirely "free will" but are
strongly conditioned by the environment’ (Forrester, 1961,
p.17). Furthermore, system dynamics is designed to allow
people to recraft a system advantageously and promote
different behaviour, thereby acknowledging the actual
environment that ‘conditions' people's behaviour (Lane,
2000). Bowen's (1994) take on the topic is somewhat similar,
believing the ability to change the system structure and the
conditions of decisions places system dynamics on a middle
point of human determinism. This is somewhat supported by
Bloomfield (1982), who demonstrated that system dynamics
is described by neither complete determinism or complete
free-will.
Charles R. Featherston and Matthew Doolan: A Critical Review of the Criticisms of System Dynamics
However, Lane (2000) believes that this 'mid-point'
between deterministic and complete ‘free will' is an
unsatisfactory conclusion. Lane (2000, p.10) recommends
that more contemporary ‘social theories which integrate
agency and structure by giving an account of the process that
mutually shape them both' is a more appropriate lens through
which to observe the paradigm's treatment of human action.
5.2 System Dynamics as a 'Grand' Theory
A second aspect to the determinism criticism of system
dynamics is that it is proposing a form of a ‘Grand’ theory.
The idea of a ‘grand’ theory is not unknown in science, for
example Von Bertalanffy's (1968) General Systems Theory
(GST). These ‘grand’ theories proposed to bring together
‘models, principles, and laws that apply to generalized
systems... It seems legitimate to ask for a theory, not of
systems of a more or less special kind, but of universal
principles applying to systems in general' (Von Bertalanffy,
1968, p.32).
However, system dynamics is only a methodology applied
to different situations and does not promote a single 'Grand
theory' of systems (Lane, 2000). While this does fit, for
example, within one of Von Bertalanffy's (1968) domains, it
still removes the notion of universally applied concepts and
principles of General Systems Theory and other ‘grand’
theories.
5.3 Austere
Another perspective on the deterministic nature of system
dynamics is given by Jackson (1991), when he groups the
paradigm with that of systems engineering and systems
analysis. Jacksons (1991, p.80) criticism of the group is that
‘people are treated as components to be engineered just like
other mechanical parts of the system. The fact that human
beings possess understanding and are only motivated to
support change and perform well if they attach favourable
meaning to the situation in which they find themselves is
ignored’.
To address this criticism, Lane (2000) explores the idea
that ‘system dynamics has an austere view of what should be
in a model and coercive view of how users should respond to
such a model’ (Lane, 2000, p.15). He explores the multitude
of views considered by system dynamics, the multiple
possible objectives of system dynamics, relationship between
modeller and problem owner and the fallacious idea of using
system dynamics to search for an ‘optimal solution’ (Lane,
2000).
The idea that Jackson is referring to the operational
austerity of the process seems only part of the criticism made
by Jackson. Lane (2000) demonstrates that in the execution
of the techniques of system dynamics, understanding and
motivation are considered. It seems just as likely though, that
the criticism is aimed at the understanding and motivation of
people within the system being explored. These notions are
addressed by system dynamics in such elements as the
‘modes of behaviour’ the paradigm wishes to explore and goal
seeking activity (Forrester; 1961; 1969).
6 Hierarchy
Hayden (2006) makes three direct criticisms of system
dynamics all with regards to a figure purported by Boyer
(2001) (see Figure 1). One of the criticisms was with regards
to system dynamics and its consideration of hierarchy.
Radzicki and Tauheed (2007, p.1) address this criticisms by
demonstrating that Boyer’s figure is not intended to describe
how system dynamics considers hierarchy. Radzicki and
Tauheed (2007, p.1) go further by explaining that the non-
linear nature of system dynamics puts limitations on the
systems that the methodology deals with and that these
limitations describe are the systems hierarchy. This
explanation appears to be only part of how system dynamics
considers hierarchy, but before we go further, a brief
exposition of hierarchy is needed.
Checkland (1981) sees ‘emergence and _ hierarchy,
communication and control' as ‘basic’ systems ideas and
central to understanding a system's behaviour. Hierarchy, in
Checkland's sense, is the relationship sub-systems have to
each other (Checkland, 1987). The hierarchy that forms in a
system defines a set of rules, obligations, controls,
regulations and limitations that exist within the system
(Hayden, 2006). To address is criticisms, system dynamics
should have a clear picture of how it's techniques consider
and inform users of these features of hierarchy.
Techniques in system dynamics, such as influence
diagrams and causal loop diagrams address the mules,
obligations, controls, regulations and limitations to a certain
degree by outlining the structure of the system and linking
relationships between elements of the system. However, they
do not describe much about the relationship between the
elements of the system as, because these are simplified
reflections of peoples' mental models, they are not designed
to.
It is not until a dynamic model is built that the rules,
obligations, controls, regulations and limitations are
formalised and crystallised to reflect the system. These
models define the mules, obligations, controls, regulations and
limitations within the model by mathematically defining the
relationships between elements.
These arguments only scratch the surface of the question
of hierarchy in system dynamics. Radzicki and Tauheed's
(2007) contribution is that the nonlinear nature of system
dynamics places some of the limiting factors on systems
explored by system dynamics. Here it is argued that the
modelling of systems and the relationships outlined in those
models contain many more of the ‘rules, obligations, controls,
regulations and limitations' in the system and that modelling
plays a greater role in addressing hierarchy. Perhaps
hierarchy, and maybe Checkland's (1981) other 'basic' system
ideas, need to be discussed application by application to
ensure their consideration in modelling, or perhaps system
dynamics needs to crystallise its thinking in this area and
construct formal theories around system hierarchy.
The 30th International Conference of the System Dynamics Society
OK
Constitutional Order
incentives
Vv
constraints
Vv
Institutional Forms
incentives
Vv
Organizations
constraints
Vv
+
incentives
\ Individuals /
constraints
Figure 1: Hierarchy: Rules and Relationships between Constitutional Order,
Institutions, Organisations and Conventions (Source: Boyer, 2001)
7 Discussion
7.1 Communication
As was shown in the above review, while many of the
criticisms that have been aimed at system dynamics are
theoretically founded, many are invalid in the context of
system dynamics. These demonstrate more a poor
understanding of system dynamics rather than failures of
system dynamics.
However, the concepts behind and encapsulated in system
dynamics and can be difficult to understand and leam
(Forrester, 1961; 1971b; Comin & Gonzalez, 2007; Cronin et
al., 2009; Sterman, 2010). Furthermore, the conclusions
reached through a system dynamics process and its models
are often difficult to communicate to people not directly
involved in their generation (GroRler, 2007; Barlas, 2007).
These pose great challenges for the field of system dynamics.
These errable criticisms exemplify the problem for system
dynamics. It demonstrates that many people, including those
willing to criticise it, have a poor understanding of the
underlying theories of systems dynamics; in particular its
aims, techniques and limitations. Overcoming this through
communication and developing a greater understanding of
system dynamics in a general audience is important for the
field (Forrester, 2007).
Despite this level of aptitude in communication, some
believe that system dynamics does have some intuitiveness in
its communication. Sharp and Price (1984, p.5) believe that ‘it
would seem unlikely that policy prescriptions generated via
SD [system dynamics] models would enjoy even their present
success, if the reason for them working could not be
appreciated by the decision maker in a straight forward way’.
However, this view is concerned with the communication of
results, rather than communication of the theories
underpinning system dynamics, a different discussion that
itself needs development (see for example GroRler, 2007).
9
7.2 Adoption
Forrester (2007, p.361) stated that the ‘failure of system
dynamics to penetrate lies directly with the system dynamics
profession and not with those in govemment’. System
dynamics has been developed over fifty years and has been
applied successfully to many situations: one must question
why is it not in great use. Forrester (2007) later describes
some of the exigent needs for the field, including education,
increasing public awareness and promoting system dynamics
as a tool that can be used to help solve some of their
problems. The above review demonstrates that this call is still
pertinent. Perhaps Ulli-Beer et al.'s (2010) model of
acceptance (adoption) and rejection (abandonment) dynamics
could provide more insight into how this could be achieved.
7.3. Critical review
The review has uncovered some exigent theories in
system dynamics that need to be developed and consolidated.
How to build confidence in models is one such area that was
identified. There are several different theories in the field
regarding confidence building that appear to conflict with
each other. Contrasting theories that allow people to build
confidence in different ways might be good for system
dynamics if they are mutually accommodating. However, the
different perspectives on confidence building, in particular
the importance of historical data and how it can be used, can
be damaging for the field as it gives external observers the
impression that the paradigm is in its adolescence, with major
theories still in contention. This could be a partial explanation
for the low adoption rate of system dynamics.
System dynamics also has some criticisms that remain
unaddressed. While individually system dynamicists can
somewhat deal with the endogenous nature of system
dynamics and the as yet unsettled debate on the use of
Charles R. Featherston and Matthew Doolan: A Critical Review of the Criticisms of System Dynamics
historical data in confidence building, together they create
some unease and do not appear to coexists. If historical data
is important then the study of purely endogenous behaviour is
diluted as external noise and pulses are required to mould the
model to simulate historical data. If endogenous behaviour is
the central aim of any application of system dynamics, as has
been shown by Richardson (2011) and above, then
comparison to historical data is likely to be one of the less
useful tests for a dynamic model. A clearer understanding of
this apparent inability to coexist or proof that it is not a
dichotomy at all would be an valuable contribution to the
field.
Hierarchy is another area that seems to have received
little attention in the system dynamics literature. The
argument above is that system dynamics builds into its
models the rules, obligations, controls, regulations and
limitations to which the term hierarchy refers. However, if
hierarchy is one of the four ‘ideas' central to understanding a
system's behaviour as Checkland (1981) purports, then
perhaps more developed and detailed work in this area is
required. By establishing and addressing these more
commonly accepted 'systems ideas' perhaps greater dialogue
and exchange can begin to occur between the field of system
dynamics and the larger field of systems research, something
that others have observed have been lacking in the field (see
Forrester, 2007).
The review has also uncovered theories that have been
somewhat addressed, but a shortage of literature and research
in the area suggests more work needs to be done to establish
a strong theoretical position. For example, work is being done
on pluralism from the perspective of differing points of views
on the problem at hand, goals of intervention and the
structure of the system itself (for example Newell et al.,
2008; Newell & Proust, 2009; Newell et al., 2011). However,
systems dynamics could draw on other pluralistic activities,
such as those suggested by Sibbet (2010; 2011) or those used
in Technology Roadmapping (Phaal et al., 2010), to improve
its ability to build consensus and consider multiple
perspectives. Bounded rationality limits one's decision
making abilities (Simon, 1953). By including more
perspectives in the process and increasing the ‘bounds' that
limit the problem and potential adaptations (actions &
reactions to the problem), the rational used to make decisions
- and in the case of system dynamics learn about systems and
test and develop mental models - can be improved and
perceived limitations can be removed.
Another example of partially addressed theories is that of
pluralism from the perspective of aggregation within the
system. This tends to be addressed on a contingent,
application-by-application basis. It can be difficult for people
learning system dynamics to achieve an appropriate level of
aggregation and often trial and error is required to do so. It
appears the field could benefit from work being done to
generate more formal and teachable rules for aggregation in
system dynamic models.
One obvious conclusion that became clear in our search
and review of the criticisms of system dynamics is that there
are few criticisms aimed at the mathematics behind system
dynamics. It is likely that this is because of the field's strong
mathematical foundation in Dynamical Systems Theory, a
branch of mathematics that has been around for hundreds of
years and has its origins in the work of Isaac Newton
(Beltrami, 1987). This strength of the field's is important and
should not be lost in the process of addressing and
developing the more qualitative and philosophical issues
raised in this review.
7.4 Overcoming the criticisms
Calls for education (Forrester, 2007; Barlas, 2007) and
calls for more theoretical work (Forrester, 2007; Lane, 2000)
are two often cited areas that the field of system dynamics
could develop to help overcome its criticisms. Developing the
theories underpinning the field and informing people of its
goals, techniques and limitations would help reduce invalid
criticisms of system dynamics and build its theoretical
resilience.
In addition to education and theoretical work, the review
has identified an image issue for system dynamics. Bourne
from poor education in the field, partially explained by its
complexity, many seem to believe the field unready or unable
to assist in dealing with the complex world for which it was
designed. Whether it is because of its seeming unsettled
validation techniques, its implicit handling of hierarchy,
unconventional take on determinism and free-will, its
adaptable take on pluralism or some other yet unidentified
issue, applications of system dynamics seems to incur
scepticism among many individuals. Despite positive reviews
in many fields of the benefits of applying system dynamics
and optimism of its potential, it is important system dynamics
works to address the scepticism that it imbues.
8 Conclusion
System dynamics has two central problems that lead to
many of the claims being made against it. Firstly, is that it is
often misapplied (Richardson, 2011). Secondly, as has been
shown in this paper, that people are often misinformed as to
its goals, expected outcomes and limitations.
Central to the second point is the endogenous perspective,
that many critics apparently do not understand or know, the
encouragement of learning during the process of forming a
model, and not the emphasis on ‘the’ model, and the
contingent nature of its application. A possible cause of much
of this miscommunication is people applying system
dynamics incorrectly, not qualifying their applications
appropriately, or overstating the outcomes from the process.
Systems dynamics also needs to address more formally
some of the more considered criticisms of the paradigm. A
formal understanding of how system dynamics considers
hierarchy is one such example. Addressing the concem for
complex applications and a more formal method of
quantising intangible variables. Many of these criticisms
could be the cause for the poor take-up of system dynamics in
strategy and policy development (Grofler, 2007; Forester,
2007). As a consequence it is important system dynamics
address these issues, and all other criticisms, to strengthen the
field, both theoretically and in the eyes of potential users.
Finally, system dynamics needs to have greater
communication with the public, with other technical people
and with other systems research areas (Forrester, 2007;
10
The 30th International Conference of the System Dynamics Society
Barlas, 2007). Communication with the broader public could
increase its adoption and help it to be utilised more broadly.
Communication with technical people, other academics and
other fields of systems thinking could help to develop the
field's basic theoretical framework. Finally, greater
communication is essential in educating people about system
dynamics, which could help to avoid the errable and
misapplied criticisms that have been aimed at system
dynamics.
Many of the criticisms of system dynamics, such as its
determinism and human austerity have been addressed.
However, theoretical work still needs to be done on some of
the field's criticisms, including the role of historical data in
building confidence in models, the field's reductionist
perspective and how system dynamics addresses plurality and
hierarchy. Such work, combined with increased education
and communication, would help the field to be accepted more
broadly.
Further work could also be done to explore how these
criticisms apply to other approaches that are used instead of
system dynamics. Using criticism to compare different
paradigms can prove valuable when selecting the most
appropriate approach for modelling and problem solving.
9 References
Ackoff, Russell L. (1974), Redesigning the Future, John
Wiley & Sons Inc.: New York, New York, The United
States of America.
Ansoff, H. Igor & Slevin, Dennis P. (1968), An appreciation
of industrial dynamics, Management Science, Vol. 14,
No. 7, pp. 383-397.
Anderson, Philip Warren (1972), More Is Different: Broken
symmetry and the nature of the hierarchical structure of
science, Science, Vol. 177, No. 4047, pp. 393-396.
Barlas, Yaman (2007), Leverage points to march "upward
from the aimless plateau", System Dynamics Review,
Vol. 23, No. 4, pp. 469-473.
Beltrami, Edward (1987), Mathematics for Dynamic
Modeling, Academic Press, Inc.: Boston, Massachusetts,
The United States of America.
Bloomfield, Brian P. (1982), Cosmology, knowledge and
social structure: the case of Forrester and system
dynamics, Journal of Applied Systems Analysis, Vol. 9,
No. 1, pp. 3-15.
Bowen, Michael G. (1994), System dynamics, determinism
and choice: toward a reconsideration of the image of
"systems man", System Dynamics Review, Vol. 10, No.
1, pp. 87-90.
Boyer, Robert (2001), The Regulation Approach as a Theory
of Capitalism: A New Derivation. In: Labrousse, Agnes
& Weisz, Jean-Daniel (eds.) (2001), Institutional
Economics in France and Germany, Bertelsmann
Springer: Berlin, Germany
Checkland, Peter B. (1978), The origins and nature of ‘hard'
systems thinking, Journal of Applied Systems Analysis,
Vol. 5, No. 2, pp. 99-110.
11
Checkland, Peter B. (1981), Systems Thinking, Systems
Practice, John Wiley & Sons Inc.: Chichester, The
United Kingdom.
Checkland, Peter B. (1987), The application of systems
thinking in real-world problem-situations: the
emergence of soft systems methodology. In: Jackson,
Michael. C. & Keys, Paul (eds.) (1987), New Directions
in Management Science, Gower Publishing Company
Limited: Aldershot, Hampshire, The United Kingdom.
Churchman, Charles West (1967), Wicked Problems,
Management Science, Vol. 14, Dec, p. B141-42.
Cronin, Matthew A. & Gonzalez, Cleotilde (2007),
Understanding the building blocks of dynamic systems,
System Dynamics Review, Vol. 23, No. 1, pp. 1-17.
Cronin, Matthew A.; Gonzalez, Cleotilde and Sterman, John
D. (2009), Why don’t well-educated adults understand
accumulation? A challenge to researchers, educators
and citizens, Organizational Behavior and Human
Decision Processes, Vol. 108, No. 1, pp.116-130.
Coyle, Robert G. (1983), The technical elements of the
system dynamics approach, European Journal of
Operational Research, Vol. 14, No. 4, pp. 359-370.
Coyle, Robert R. (1986), Comment on ‘System dynamics and
operational research: An appraisal’, European Journal
of Operational Research, Vol. 23, No. 3, pp. 403-407.
Forrester, Jay W. (1961), Industrial Dynamics, The M.LT.
Press: Cambridge, Massachusetts, The United States of
America & John Wiley Sons Inc.: New York, New
Y ork, The United States of A merica (published jointly).
Forrester, Jay W. (1968), Industrial Dynamics: after the first
decade. In: Forrester, Jay W. (ed.) (1975 Collection),
Collected Papers of Jay W. Forrester, Wright-Allen
Press Inc.: Cambridge, Massachusetts, The United
States of America.
Forrester, Jay W. (1969), Urban Dynamics, The M.LT. Press:
Cambridge, Massachusetts, The United States of
America.
Forrester, Jay W. (1971a), World Dynamics, Wright-Allen
Press Inc.: Cambridge, Massachusetts, The United
States of America.
Forrester, Jay W. (1971b), Counterintuitive behavior of social
systems, Technology Review, Vol. 73, No. 3, pp. 52-68.
Forrester, Jay W. (1985), "The" model versus a modeling
"process", System Dynamics Review, Vol. 1, No. 1, pp.
133-134.
Forrester, Jay W. (2001), Future Development of the System
Dynamics Paradigm, Keynote address for The 1983
International System Dynamics Conference, Chestnut
Hill, Massachusetts, The United States of America.
Forrester, Jay W. (2003), Economic Theory for the New
Millennium, proceedings of the Twenty-First Annual
Conference of the System Dynamics Society: New Y ork,
New Y ork, The United States of America.
Forrester, Jay W. (2007), System dynamics-the next fifty
years, System Dynamics Review, Vol. 23, No. 2/3, pp.
359-370 (based on a speech to the International System
Dynamics Conference, Boston, Massachusetts, The
United States of America in August, 2007).
Charles R. Featherston and Matthew Doolan: A Critical Review of the Criticisms of System Dynamics
Forrester, Jay W. & Senge, Peter M. (1980), Tests for
building confidence in system dynamics models, TIMS
Studies in the Management Sciences, Vol. 14, pp. 209-
228.
Goodwin, Richard (1951a), Economics in Business-Cycle
Analysis. In: Hansen, Alvin H. (ed.) (1951), Business
Cycles and National Income, W. W. Norton &
Company, Inc.: New York, New York, The United
States of America, pp. 417-468.
Grofler, Andreas (2007), System dynamics projects that
failed to make an impact, System Dynamics Review,
Vol. 23, No. 4, pp. 437-452.
Hayden, F. Gregory (2006), The Inadequacy of Forrester
System Dynamics Computer Programs for Institutional
Principles of Hierarchy, Feedback, and Openness,
Journal of Economic Issues, Vol. 40, No. 2, pp. 527-
535.
Homer, Jack (1997), Structure, Data, and Compelling
Conclusions: Notes from the Field, System Dynamics
Review, Vol. 13, No. 3, pp. 293-309.
Jackson, Michael C. (1982), The nature of soft systems
thinking: The work of Churchman, Ackoff and
Checkland, Journal of Applied Systems Analysis, Vol. 9,
No. 1, pp. 17-29.
Jackson, Michael C. (1991), Systems Methodology for the
Management Sciences, Plenum Publishers: New Y ork,
New Y ork, The United States of America.
Jackson, Michael C. & Keys, Paul (1984), Towards a System
of System methodologies, Journal of the Operational
Research Society, Vol. 35, No. 6, pp. 473-486.
Keys, Paul (1990), System Dynamics as a Systems-based
Problem-Solving Method, Systems Practice, Vol. 3, No.
5, p. 479-493.
Lane, David C. (2000), Should System Dynamics be
Described as a ‘Hard’ or ‘Deterministic’ Systems
Approach?, Systems Research and Behavioral Science,
Vol. 17, pp. 3-22.
Meadows, Donella H. (2008), Thinking in Systems: A Primer,
Chelsea Green Publishing: White River Junction,
Vermont, The United States of America.
Meadows, Donella H.; Meadows, Dennis L.; Randers, Jorgen
& Behrens III, William W. (1972), Limits to Growth,
Universe Books: New York, New York, The United
States of America.
Meadows, Donella H.; Meadows, Dennis L. & Randers,
Jorgen (1992), Beyond the Limits, Chelsea Green
Publishing Company: Post Mills, Vermont, The United
States of America.
Meadows, Donella H.; Randers, Jorgen & Meadows, Dennis
L. (2004), Limits to Growth: The 30 Year Update,
Chelsea Green Publishing Company: White River
Junction, Vermont, The United States of America.
Marxsen, Craig S. (2003), Prophecy de Novo: The Nearly
Self-Fulfilling Doomsday Forecast, The Independent
Review, Vol. 7, No. 3, pp. 325-342.
Morecroft, John (2007), Strategic Modeling and Business
Dynamics: A Feedback Systems Approach, John Wiley
& Sons, Ltd: Chichester, The United Kingdom.
Newell, Barry (2012), Simple Models, Powerful Ideas,
Submitted.
Newell, Barry & Proust, Katrina (2009), I See How You
Think: Using Influence Diagrams to Support Dialogue,
Working Paper, The Fenner School of Environment and
Society, College of Medicine, Biology and
Environment, The Australian National University.
Newell, Barry; Proust, Katrina; Wiltshire, Gabrielle &
Newell, David (2008), Taking a systems approach to
estuary management, 17th New South Wales Coastal
Conference: Wollongong, New South Wales, Australia.
Newell, Barry; Marsh, Debborah M. & Sharma, Deepak
(2011), Enhancing the resilience of the Australian
National Electricity Market: Taking a Systems
Approach in Policy Development, Ecology and Society,
Vol. 16, No. 2, Art. 15.
Osgood, Nathaniel (2009), Lightening the performance
burden of individual-based models through dimensional
analysis and scale modeling, System Dynamics Review,
Vol. 25, No. 2, pp. 101-134.
Peterson, David W. (1975), Hypothesis, Estimation, and
Validation of Dynamics Social Models -- Energy
Demand Modeling, Unpublished PhD Thesis,
Massachusetts Institute of Technology (MIT):
Massachusetts, The United States of America.
Phaal, Robert; Farrukh, Clare & Probert, David (2010)
Roadmapping for strategy and innovation: Aligning
technology and markets in a dynamic world,
Unspecified, ISBN 978-1-902546-82-7.
Piaget, Jean (1928), The Child's Conception of The World,
Routledge & Kegan Paul Ltd: London, The United
Kingdom.
Piaget, Jean (1930), The Child's Conception of Physical
Causality, Kegan Paul, Trench, Trubner & Co. Ltd:
London, The United Kingdom.
Popper, Karl R. (1945a), The Open Society and Its Enemies,
Vol. 1: The Spell of Plato (5th ed, revised 1966),
Princeton University Press: Princeton, New Jersey, The
United States of America.
Popper, Karl R. (1945b), The Open Society and Its Enemies,
Vol. 2: Hegel, Marx, and the Aftermath (5th ed, revised
1966), Princeton University Press: Princeton, New
Jersey, The United States of America
Popper, Karl R. (1957), The Poverty of Historicism,
Routledge & Kegan Paul Ltd: London, The United
Kingdom.
Radzicki, Michael J. (2004), Expectation Formation and
Parameter Estimation in Uncertain Dynamical Systems:
The Systems Dynamics Approach to Post Keynesian-
Institutional Economics, The 22nd International
Conference of the System Dynamics Society, July 25-29,
2004: Kelbe College, University of Oxford, Oxford, The
United Kingdom.
Radzicki, Michael J. & Tauheed, Linwood (2007), In
Defense of System Dynamics: A Response to Professor
Hayden, Conference Proceedings of the 2007
International Conference of the System Dynamics
Society and 50th Anniversary Celebration, July 29 -
12
The 30th International Conference of the System Dynamics Society
August 2, 2007: Boston, Massachusetts, The United
States of America.
Richardson, George P. (2011), Reflections on the
foundations of system dynamics, System Dynamics
Review, Vol. 27, No. 3 (July-September 2011), pp. 219-
243.
Rouwette, Etiénne A. J. A.; Korzilius,Hubert; Vennix &
Jacobs, Eric (2011), Modeling as persuasion: the impact
of group model building on attitudes and behavior,
System Dynamics Review, Vol. 27, No. 1, pp. 1-21.
Saeed, Khalid (1992), Slicing a complex problem for system
dynamics modeling, System Dynamics Review, Vol. 8,
No. 3, pp. 251-261.
Schmandt, Jugen (2010), Mitchell, George P. and the idea of
sustainability, Unspecified. ISBN: 978-1-60344-217-6.
Senge, Peter M. (1990), The Fifth Discipline: The Art &
Practice of The Learning Organization, Century
Business: London, The United Kingdom.
Sibbet, David (2010), Visual Meetings: How Graphics, Sticky
Notes and Idea Mapping Can Transform Group
Productivity, John Wiley & Sons Inc.: Hoboken, New
Jersey, The United States of America.
Sibbet, David (2011), Visual Teams: Graphic Tools for
Commitment, Innovation, and High Performance, John
Wiley & Sons Inc.: Hoboken, New Jersey, The United
States of America.
Simon, Herbert Alexander (1953), Administrative Behavior:
A Study of Decision-Making Processes in
Administrative Organisation (2nd ed.), Macmillan
Publishers Ltd: New York, New York, The United
States of America.
Simon, Julian L. (1981), The Ultimate Resource, Princeton
University Press: Princeton, New Jersey, The United
States of America.
Solow, Robert M. (1972), Unknown, Newsweek, March 13,
1972, p. 103.
Sosna, Marc; Trevinyo-Rodriguez, Rosa Nelly & Velamuri,
S. Ramakrishna (2010), Business Model Innovation
through Trial-and-Error Leaming: The Naturhouse
Case, Long Range Planning, Vol. 383, pp. 407.
Sterman, John D. (2000), Business Dynamics: Systems
Thinking and Modeling for a Complex World, McGraw-
Hill/Irwin: New Y ork, New Y ork, The United States of
America.
Sterman, John D. (2010), Does formal system dynamics
training improve people's understanding of
accumulation?, System Dynamics Review, Vol. 26, No.
4, pp. 316-334,
Ulli-Beer, Silvia; Gassmann, Fritz; Bosshardta, Mathias &
Wokauna, Alexander (2010), Generic structure to
simulate acceptance dynamics, System Dynamics
Review, Vol. 26, No. 2, pp. 89-116.
Von Bertalanffy, Ludwig (1968), General Systems Theory:
Foundation, Developments, Applications, George
Braziller, Inc.: New Y ork, New Y ork, The United States
of America.
13
Wolstenholme, Eric F. (1982), System dynamics in
perspective, Journal of the Operational Research
Society, Vol. 33, No. 6, pp. 547-556.
Wolstenholme, Eric F. (1983), Modelling National
Development Programmes -- An Exercise in System
Description and Qualitative Analysis Using System
Dynamics, J ournal of the Operational Research Society,
Vol. 34, No. 12, pp. 1133-1148.
Zagonel, Aldo A. Zagonel; Mojtahedzadeh, Mohammad;
Richardson, George P.; Brodsky, Nancy S.; Brown,
Theresa J.; Conrad, Stephen H. & Glass, Robert J.
(2011), Developing a theory of the societal lifecycle of
cigarette smoking: Explaining and anticipating trends
using information feedback, International Conference of
The System Dynamics Society: Washington, D.C., The
United States of America.
Bibliographical Note
DATE 4th ofJ une, 2012
Charles Featherston
Charles is a Doctoral student and the Australian National University (ANU). His topic is exploring if
and how system dynamics can be used to inform scenario planning. Charles received his Bachelor of
Engineering with honours and bachelor of Commerce double degree from the ANU. Charles has also
collaborates heavily with the Institute of Manufacturing at the University of Cambridge. His research
interests are in scenario planning, modelling, technology development and innovation.
Dr Matthew Doolan
Matthew Doolan has 13 years experience in the manufacturing industry. He undertook a PhD in
manufacturing process control at Ford Australia, based at the Stamping Operations Facility. Matthew
is in an academic position with the School of Engineering at the Australian National University. He
has been involved in numerous manufacturing projects including dimensional control projects with
Ford Australia, Supply Chain companies and the AutoCRC. Matthew lectures in Manufacturing
Technology, Engineering Management and Operations Management. He has research interests in
Strategic Technology Development and Manufacturing.
1
