1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
Defining System Dy ics Educati
Yaman Barlas
Bogazici University
Department of Industrial Engineering
80815 Bebek, Istanbul
Turkey
E-mail: Ybarlas@trboun.bitnet; Fax: 90 212 265 1800
Abstract
This is the second paper in a series that aims to start a debate on issues involved in university-level
system dynamics education. The first paper argues that the field has not experienced the growth that
one would expect from its potential and identifies several issues that need to be addressed by the
system dynamics community, before the field can proliferate in universities. This second paper tackles
some of those problems. More specifically, the paper discusses the academic definition of system
dynamics: what is the academic core of system dynamics? what other subjects are of immediate
relevance and importance with respect to this core? The paper offers answers to these questions. The
second issue that the paper deals with is the problem of terminology. I discuss different types of
terminology problems, the most significant being system dynamics, the very name of the field.
System d ics having an blished meaning in math ical and engineering sciences, does not
convey the specific meaning that we wish to attach it. I discuss various potential problems caused by
this situation. | then offer a short list of alternative, more specific names for the field. I conclude that,
once the academic issues are ri tackled, the university-level system d i i
should experience a growth, which would be a major step toward activating an exponential growth
process in the field in general.
Education, page 1
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
fining © . ducat;
D Dy E
This is the second paper in a series that aims to start a debate on issues involved in
level system dy The first paper argues that the field has not experienced
the growth that one would expect from its potential and identifies several issues that need to be
dd: d by the system d before the field can proliferate in universities. This
second paper tackles some of those problems. More ‘Specifically, the paper discusses the academic
of system d: ics: what is the academii re” of system dynamics? what other subjects
are of i di it and i with respect to this core? The paper offers answers to
these questions. The second issue that the paper deals with is the problem of terminology. I discuss
different types of terminology problems, the most significant being * “system dynamics,” the very name
of the field. “System d. ics” having an established meaning in math 1 and engineering
sciences, does not convey the specific meaning that we wish to attach it. I discuss various potential
problems caused by this situation. I then offer a short list of alternative, more specific names for the
field. I conclude that, once the academic issues are rigorously tackled, the university-level system
dynamics education should experience a growth, which would be a major step toward activating an
exponential growth process in the field in general.
The first paper of this series (Barlas 1993) is an overview and introduction paper that
identifies the various aspects of the problems involved in system dynamics education. The paper first
comments on different types.of system dynamics courses, based on personal experience of the author.
It then discusses four problem areas: i) lack of formal teaching material, ii) insuffi icient literature on
dagogy, iii) probl of: ‘inol and iv) inad on The
first paper concludes that once these problems are dealt with, the university-level system dynamics
education can experience a growth, which would be a significant step toward initiating an exponential
growth fmgess in the field in general, (Baties 1993
system di in an ever changing environment, with new or
revised ¢ hodologies and hes being added ly (eg. chaos, sii
gaming, soft systems methodology, systems thinking). Relationships between system dynamics and
these hes are being d d in the lit (Gould- ‘Kreutzer 1993, Richmond 1993, Lane
1993, Towill 1993 and Senge 1990). There are still various issues in these relationships that need to
be clarified. Plenary sessions of this very conference are devoted to a comparative debate on
“systems” meth ies. I believe that system dynamics and these sister fields can
establish extremely beneficial relationships. But I also believe that a pre-condition for this to occur, is
to first define the core of system dynamics. Unless this is lished, the field’s relationships with
other fields will increasingly be a source of confusion and vagueness, rather than beneficial
cooperation - a situation that already exists to some degree (especially vis-a-vis simulation games and
systems thinking; see Gould-Kreutzer 1993).
The purpose of this second paper is to focus on some of the issues raised by the first paper and make
some suggestions. Specifically, we deal with two issues: first, the academic definition, the “core” of
system dynamics. Second, we discuss the terminology problem, in lar the very name of the
field, “system dynamics.”
The Need to Define the Core of System Dynamics
In the development and growth of any discip formal university educati plays an
role. Only the ed (the “reproductive” group), through formal (graduate) university
education, can produce educators. Thus, there is a positive feedback loop that must be set in motion
for the field to experience substantial growth, and a mini (threshold) number of edi
needed to activate the loop. This situation is especialy: true for system dynamics, which is a dificult,
Experts in such a discipline can not be produced by i t-
time training; formal university education is necessary. Currently, very few universities offer regular,
formal system dynamics courses.
A necessary condition to define formal system dynamics Siuestion is to first define the “core”
subjects of the field. Phil a mature ized by a well-defined core. The
core represents those aspects, ions of the di; m that are never ioned during what
Thomas Kuhn calls “normal science” (Kuhn 1962). The ‘core of the paradigm, unlike its peripheral
Education, page 2
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
aspects, is very rigid. Peripheral ions of a digm may be th fioned and
occasionally thrown away; but when the core assumptions are questioned, this signals a scientific
revolution, which in turn may mean the end of the paradigm. It is the constancy of the core
assumptions that makes a paradigm a very efficient “puzzle-solver” during periods of normal science
(Kuhn 1962). That is why established engineering sciences monitor their core subjects, by
periodically publishing standard curricula and list of subjects for their core courses. System
dynamics should also start a structured debate on what constitutes the core of system dynamics. Most
work on education focuses either on high schools (eg. - Davidsen etal. 1993, Forrester 1993 Brown
1992, Draper and Swanson 1990), or on short | training (Ri
1993,Ford and Gardiner 1987, the entire special issue of EJOR 1992, especially Graham et al. 1992).
With a few ptions (eg. And and Richardson 1979 and 1980, Clauset 1985), there has been
little work devoted to university-level system ics educati
The Academic Core of System Dy
One way to approach this issue is to ask, “what is system dynamics?”But most likely, a single
definition will not provide a complete answer to this question. There are various books that provide
different definitions,each emphasizing different aspects of the field(for eg. see Wolstenholme 1990,
Richardson and Pugh 1981, Forrester 1961).An al h which, I believe will simplify
this task, is to ask: “In what ways is a system dynamicist unique?What are those qualities of a system
dynamicist that clearly distinghish him/her from professionals trained in other similar
disciplines?”Having formal education in industrial engineering, operations Tesearch and
ing, and p with from these di I find out
that a system d icist is distinguished in two fi d 1 ways: First, system dynamicists are
causal, feedback model is imp I believe that we are Uniquely
qualified in building excellent causal models of probl that involve signi
Thus, system d: ion must first teach the knowledge that students must acquire
to be causal modelers of feedback p second. system ici are. model-based
analysts of dynamic policy issues.Again, the is is that, compared to other Is, we are
uniquely qualified in making use of models to tackle dynamic problems involving long-term policy
questions.The core of system d: must th include the p phy, tools and methods
necessary for this second set of skills.
Using the above core i of system d ics as a it is now possible to
discuss what specific subjects must be included in a given course.but we also need to define first what
kind of course we have in mind.As we discussed in the previous paper, there are many different types
of courses, depending on the nature of the student audience, the level of the course, whether the
course is stand-alone or part of a larger subject, and whether the course is intended as an elective or
part of a required sequence.(Barlas 1993).In this paper, I will focus on the type of course which I
believe is most crucial for the growth of the field: A sequence of two courses exclusively dedicated to
system dynamics. At least the first one (and p bly both) must be und duate courses.
Our purpose is to come up with a list of suggested topics for an undergraduate course
dedicated to system dynamics.In a similar attempt, And and Richardson (1979) ask the foll 2,
question:”what do we want everyone to know after having had an introduction to system d ics?”
We try to answer the same question, while keeping in mind the definition of the “core” given in the
previous paragraphs. For the first course, we come up with the suggested list of topics shown in
Figure 1. All topics are quite self-explanatory, hence we will not discuss them further.
Education, page 3
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
TOPIC SEQUENCE:
- Systems and Models
- Structure and Dynamic Behavior; Illustrations
- Problems of Dynamic Nature; Illustrations
- Systems Thinking
- Causality and Causal-loop Diagrams
- Model Conceptualization; Illustrations
- Tools for Dynamic Systems Modeling: Stocks and Flows.
- Introduction to Stock-Flow Diagrams and SOFTWARE.
" - Feedback loops: Positive and Negative Feedback
~ Behavior of Positive Feedback; Growth Processes
- Behavior of Negative Feedback; Goal-seeking Processes
- Linear and Non-linear Equation Formulation
- Coupling of Positive and Negative Feedback Loops
- S-shaped behavior and “boom-then-bust” patterns
- Importance of Time Delays
- CASE STUDY in Model Evaluation and Analysis.
~ Structure of Oscillatory Systems; Illustration:
~ Generic and Generic Sub-systems.
- CASE STUDY in Policy Design
~ Term Project Discussions.
Figure 1. Topics d For an Introduction to System Dynamics Course.
Some pedagogical aspects of this course outline need to be mentioned. First, there must be a stream of
short assingments through the course to keep the student alert. Second, there must be a term project,
but care must be exercised to keep it simple enough.My experiance has shown that students tend to be
weak on problem identification and attempt to build models that are too big and too complicated to be
fully und d. i d and analyzed in one half of a term. Third, the course must be rich in
examples illustrations and case studies. There are many aspects of system dynamics methodology
pecially in problem ideti ion, model ptulization and policy analysis phases) that are
almost impossible to convey in the abstract. (See Andersen and Richardson 1979 and 1980 for a rich
discussion). Fourth, the level of mathematical analysis of system dynamics covvered in the course
would depend on the particular student body. But I would nevertheless suggest that mathematical
analysis be kept at a minimum in this first course; in one term, there is simply not enough time to
teach the mathematical aspects, in addition to all the philisophical, methodological and conceptual
subjects that constitute the core of the field. Finall, Here is a tentative list of textbooks and/or
references for the course; High Performance Systems (1992) ;Wolstenholme (1990) ; Roberts et.al.
(1983) ; Richardson and Pugh (1981) ; Goodman (1974) ; Forrester (1968) and (1961).
A second course in system dynamics (either graduate or undergraduate) is really a must, if our
goal is to produce a growth in the field, via a growth in system dynamics education. Such a course
would have several objectives. First, it would discuss in more depth, some of the topics already
covered in the first course, such as model validity, sensitiviy and policy design. Second, it would
introduce new concepts and tools, such as stochasiticity in systems, discrete models, chaotic behavior,
and in general a host of th ical analysis techni Third, it would introduce a number of
related fiels that have important ties with system dynamics. And finally, being a second course, it
would give the a students an oppertunity to work on a major term project. Suggested topics for such a
course in system dynamics is shown in Figure 2.
Education, page 4
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
TOPIC SEQUENCE:
- Systems and Models and Differential Equations
- Systems Simulation as a Numerical Method
- Mathematics of Linear System Dynamics
- Equilibrium and Stability Analysis
- Phase-plane Techniques
- Non-linear Systems and Linearisation
- Model Sensitivity
~ Model Validation
- Policy Design
- CASE STUDY
- Optimisation
- Stochasticity in Systems
- Discrete versus Continious Models
- Chaotic Behaviour
- Role of Interactive Simulation Games
- Survey of Related Fields
- CASE STUDY
- Term Project Discussions
Figure2. Topics Suggested for a Second Course in System Dynamics
Note that this being a second course in system dynamics, it should provide a much greater
degree of freedom to the instructor. Therefore, the suggested topics are rather tentative in nature,
compared to the ones suggested for the first course. The ii ding on his/her exp and
the student audience, may wish to emphasis certain topics, and spend little time on others. For
instance, an instructor may prefer spending more time on the simulation-related analysis and design
issues, and less time on mathematical techniques. In addition to the list of textbooks given for the first
course, the second course would naturally revolve around a variety of reading material, as suggested
by the instructor.
Problems of Terminol
My previous paper indicates that system dynamics field has some terminology problems. Some
concepts central to the field do not have a unique technical name. For instance, the same technical
concept is called "stock" by some authors, "level" by others and "state" by yet others. There is "flow"
and then there is “rate” that define: the same concept. Model diagrams ¢ are sometimes called "flow
di i "and Rich bulary is
good for a natural vergupge but not necessarily for a technical field, especially if it is in the
development phase. It creates unnecessary communication difficulties and gives the impression that
the field is somewhat ill-defined and immature. Students in introductory system dynamics classes are
d by this multiplicity of terms, especially since they have to read extensively from a variety of
sources. As part of our attempt to then system d: ics as a formal academic field, we must
make sure that there is only one term for each major technical concept. For instance, I propose that we
consistently use the term “stock” (instead of “level” or “state”), “flow” (instead of “rate”) and “stock-
flow diagrams” (instead of “structure” or “flow” diagrams).
Another terminology problem is our usage of certain terms in ways that differ from their
standard usage in other established disciplines. For example, Seeger (1992) explians that the terms
“open* and “closed™ have hed fields of social sciences that differ
substantially from our usage of these terms. ( {See Richardson (1991), for a thorough historical and
philosophical discussion of this and other similar terminology issues). Similarly, “influence diagram“
in decision theroy is quite different than the one used in system dynamics, I suggest that, we use the
term “closed-loop system” (instead of “closed system) and “causal-loop “ diagram (instead of
“influence diagram ) .
Education, page 5
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
wi 6 eld. « jeg”
The most dramatic example of terminology problem lies in the very name of our field,
“system dynamics." In my previous article, I state that: “this term has an established meaning in
mathematical and engineering sciences. In short, in applied math ics it means ‘math ical
analysis of dynamics of systems’ and in electrical, mechanical and systems engineering it means
‘analysis and design of dynamical engineering systems.’ (Barlas 1993. Also see Towill 1993). There
are many articles and books in applied mathematics and engineering that have the keyword "system
dynamics" (or a minor variation of it , such as “dynamics of systems” or “dynamic systems”) in the
title. (See tables 1 and 2). Again, in my previous article, I observe that: “System dynamics is an old
mathematical and engineering term with a rather general coverage. The choice of such a general and
established term to name an emerging new field with a very specific philosophy and methodology
was, in my view, a mistake. It was like calling a newly i ib-branch of statisti istical
analysis". Such a general term undermines the rigor of the field and renders its boundaries fuzzy. The
term ‘system dynamics’ in the title of a presentation, course, book etc. does not convey the specific
meaning we attach to it.” (Barlas 1993).
Having made the above observation, I later carried out a library search to test this personal
“impression. The results of this limited and “non-scientific“ title search are showm in tables 1 and 2 .
Results of table 1 were obtained by carrying out a computerized search in Science Citation Index and
Social Scienes Citation Index, for years 1988-1993 . The figures represent the number of times the
exact term “system dynamics“ appeares in titles of the articles. Thus, in Science Citation Index , we
found 8 articles that had the term “system dynamics“ in their titles , that we would classify as “core
system dynamics work“ . We found 42 articles with the term “system dynamics“ in their titles , that
belong to some other established field . Thus in our limited data set,only 16% of all articles found in
Science Citation Index with the keyword “system dynamics‘ in their titles actually belong to our field
. The next row of table 1 displays the results obtained from Social Sciences Citation Index . Here, we
observe that the numbers are reversed ;41 out of 51(80% ) of all such articles were identified as
“systems dynamics work“ as we understand it . Finally , when we pool the two data sets , a total of 49
out of 101 articles (49%) were identified as core system dynamics work . My personal impression is
that this percentage is rather low , that there are too many articles with the keyword “system
dynamics“ in the title, that do not belong to our field . A much more objective (and very time
consuming!) way to judge this would be to conduct similar searches using keywords that characterize
other comparable fields , and see if there are significant differences between our results and the ones
obtained from other fields.
Table 1, Results of 1988-1993 Article Title Search
Appearances of the keyword “systems dynamics”
In articles that we In articles that belong Ratio
would classify as to other fields
“system dynamics”
Science Citation Index 8 42 8/50=16%
Social Sciences 41 10 41/51=80%
Citation Index
Total 49 52 49/101=49%
Education, page 6
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
Table 2, Results of Book Title Search
Appearances of the keyword “systems dynamics”
In books that we In books that belong —_—Ratio
would classify as to other fields
“system dynamics”
Keyword is “system 8 8 8/16=50%
dynamics”
Keyword is “system 8 31 8/39=20%
dynamics” or
“dynamics systems”
Finally , Table 2 displays results from book titles search . In the first row , we see a total of 16 books
with the exact keyword “ system dynamics “ in the title . Of these 16 books , 8 belong to our field ,
and the other 8 belong to “other“ (traditional hard engineering)fields . The ratio is 50% . During the
search , I noticed that there were too many book titles with terms “ similar “ to system dynamics (such
as “dynamics of systems or “dynamic systems”). Just to illustrate the point , the second row of Table
2 displays the results obtained when we allow the keyword “dynamic systems” in addition to “system
dynamics.” We see that in this case, the ratio of system dynamics books (in our sense) to the total of
books found drops to 20% (8 out of 39).
A more extensive and scientific library search is needed to reach a definitive conclusion as to
whether the field has a “name problem.” (Another interesting dimension of this is that, many system
dynamicists prefer not to use the term system dynamics in the titles of books, presentations, courses,
even p dings). Nevertheless, our limited data set at least tentatively indicates that
there is a problem. These empirical findings are in general agreement with my concern that there is a
widespread, established use of “system dynamics” and that the term does not convey the specific
meaning we wish to attach to it. Audiences already familiar with this general usage of the term fail to
appreciate that the author is referring to a very specific philosophy and methodology and ask
questions like "what type of system dynamics do you exactly mean?" “System dynamics” name has
contributed to the incorrect perception that the field is just an application of control engineering to
socio-economic systems. And those who were able to see the field’s distinguishing features, have
preferred to refer to our work as "DYNAMO models" or "Forrester models" or more recently as
STELLA. It is a shame that a field as profound and imp as system d: ics must be reduced to
software names. It seems to me that this has been happening because the field does not have a unique
and specific name. Here is how Richardson (1991, p.9) explains his usage of the term “feedback” in
his book: “ there is probably a normative component to my use of a single term to cover these myriad
ideas: I see a value to using a single label, so that what is in common to these ideas becomes the
" icable'te
pa
focus.” This is a very p PP and le to a collection of terms as well as a single
term, provided that the term we choose has certain unique and distinct features. The term "system
dynamics" really refers to the very general area of inquiry to which our field belongs, and it could be
kept and used as such. But in addition, we need a more specific and unique name which depicts. what
distinguishes us from other modeling fields. This need is going be felt more acutely in the future,
because, as mentioned earlier in the article, we are wi i proli ion of sy i
methodologies. Defining the core of the field and giving it a unique name would help clarify the
hierarchy and interrelationships among these fields.
In trying to select a specific name for the field, I suggest that we refer to the “core” defined earlier in
the article, and ask the question: “what terms most uniquely characterize the core of the field?” Here
is-a simple list of “keywords” that are all vital for the field: dback; d: ics; systens ic);
closed-loop; structural; causal maoeling;policy-orientation and simulation. The question is to select a
collection of terms that reflects the distingushing features of the field. Here are some suggestions:
Education, page 7
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
-Dynamic Feedback Systems(Methodology, Modeling, Approach, Simulation)
-Systemic Dynamic Feedback(...)
-Causal Systemic Feedback(...)
-Closed-loop Dynamic Systems(...)
-Causal Dynamic Systems(...)
The alternative terms in parentheses are not part of a fixed part of the suggested name for the
field; they would be appropriately chosen, depending on the context of use. (Such as Dynamic
Feedback Systems modeling; Dynamic feedback Systems approach, etc.). My list is not exhaustive. It
is possible to come up with many other representative names. Another approach would be to chose an
acronym, which would have the advantage of being a single term. What is important is that we start
this debate in the community as soon as possible.
Lusi
System dynamics, over thirty five years after its conception, has not experienced the growth
that one would expect from is potential. System dynamics i is not merely a tedious application of
feedback control engi Throneh the years, it has grown to become
an original, creative field of i raguiiye with solid phil phi dati Today,
in contrast with the r ductionist and logical iricist philosophies of the first half of this century,
are on the rise. Time is right for the field to turn its
potential into action and mote In this article, I discuss two problems that I believe have contributed
to the field’s stagnation. The first one has been the lack of emphasis on formal, regular, university -
level system d: i To hen system di in we need to sharply
define the core subject matters of the field. ‘l offer a tentative core definition and a list of core topics
that will Hopetully, ae a debate. The second problem i is one of ” There are
logy. More ifi , both my personal experience and my
(informal) library search indicate that the very name of the field is problematic. “System dynamics”
has an d meaning in math land engineering sciences. It does not convey the specific
meaning that we wish to attach to it. Worse, it is likely to give an erroneous perception of the field. I
offer a short list of alternative, more specific names for the field. I hope that this series of papers ‘nm
start a debate on academic aspects of system d y i . Once these academic issues are r
tackled, I believe that uni ty-level system di: ion will experience a growth, ‘which
would be a major step toward activating an exponential growth process in the field in general.
References
And D.F. and G.P. Richard: 1980. Toward a Pedagogy of System Dynamics.In System
Dynamics, Legasto, A.A., J.W. Forrester and J.M. Lyneis, eds. New York: North-Holland.
And D.F. and G.P. Ri 1979. A Core Curriculum in System Dynamics. State University
of NY Occasional Paper Series GSPA 1-79,
Albany, NY. Brown, G.S. 1992. Improving Education in Public Schools: Innovative Teachers to the
Rescue.System Dynamics Review. 8 (1): 83-89.
Ford, A. and P. Gardiner. 1987. Bringing the Results of the Bonneville Project to the Classroom.
University of Southern California. Los Angeles. Ca.
Forrester, J. W. 1993. System Dynamics as an Organizing Framework for Pre-college Education.
System Dynamics Review. 9(2):183-194.
Forrester, J. W. 1968. Principles of Systems. Cambridge MA:Productivity Press.
Forrester, J. W. 1961. Industrial Dynamics Cambridge MA:Productivity Press.
Graham, A.K. J.D.W. Morecroft, P.M. Senge and J.D. Sterman. 1992. Model-Supported Case Studies
for Management Education. Special Issue of EJOR on Learning 59 (1): 151-166.
Goodman, Michael R. 1974. Study Notes in System Dynamics. Cambridge, Ma: Productivity Press.
Education, page 8
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
Gould-Kreutzer. 1993. Fi d: System Dynamics in Education.System Dynamics Review. 9(2):
101-112
High Performance Systems. 1992. STELLA Il: An Introduction to Systems Thinking. Hanover, NH.
Karsky, M.1989. What is to be Done. Informal Written Communication. Paris, France.
Kuhn, Thomas. 1962. The Si of Scientific Revoluti Chicago: The University of Chicago
Press.
Lane, D.C. 1993. With a Little Help From Our Friends: How Third Generation System Dynamics and
the Problem Structuring Techniques of Soft OR Can Learn From Each Other. In Proceedings of the
1993 System Dynamics Conference Cancun, Mexico.
Richardson, George P. 1991. Feedback Thought in Social Sciences. Pennsylvania: University of
Pennsylvania Press.
Richardson, George P. and Alexander L. Pugh III. 1981 Introduction to System Dynamics M deli
with DYNAMO Cambridge, MA: Productivity Press.
Richmond, R. 1993. Systems Thinking: Critical Thinking Skills for the 1990s and Beyond. System
Dynamics Review. 9(2): 113-133.
Roberts, Nancy, D. F. Andersen, M. Garret, R. Deal. and W. Shaffer. 1983. Introduction to
Coimputer Simulation: The System Dynamics Approach.Reading, MA: Addison-Wesley.
Education, page 9
