Table of Contents
Go Back
A critical review of the use of System Dynamics for
organizational consultation projects
Alexander Zock, Ph.D.
Deutsche Lufthansa AG, Future European Operations,
Ground and Inflight Processes
Lufthansa Basis
D-60546 Frankfurt am Main
Phone.: +49 (69) 696 46983
Fax: +49 (69) 696 91558
E-Mail: alexander.zock@dlh.de
Abstract
System Dynamics (SD) based organizational consulting projects show a diversity of process
models, conceptual frameworks and terminology. As a consequence such practices do not
meet the criteria of fully consistent and solid consultancy approaches. Nevertheless, SD
based consulting contains a number of valuable elements. The author proposes to save and
integrate these into an existing consultancy framework, namely the systemic intervention
practice, that is derived from a systemic theory of social systems, which is based on
Maturana’s concept of autopoiesis. This strengthens the already existing subjective
worldview in parts of the SD community and thus integrates the exploratory strength of SD,
that allows a formalized reflection of the perceived logical structure of mental models, into a
coherent consultancy framework based on interpretive and constructivist perspectives.
Key words: Epistemology, Luhmann’s theory of social systems, autopoiesis, structural
coupling, organization theory, organizational intervention theory, systemic intervention,
intervention architectures
1. Introduction
The use of System Dynamics (SD) for organizational interventions is from a practical point
of view a well explored terrain as can be seen from the multitude of successful field studies
reported in the literature (Rouwette et al, 2002; Senge, 1990; Vennix, 1996; Sterman, 2000).
Unfortunately the same is not true with respect to the theoretical foundation of such activities
as can be seen from the fact that in the SD community
e no widely accepted and fully developed organizational intervention model for the use
of the SD methodology in an organizational context is available in the literature (e.g.
Rouwette et al. 2002),
e no clear reference to an organizational theory is established that would allow the
derivation of a consistent intervention model that could be used as a framework for
the use of SD methodology in an organizational context,
e the debate about the adequate epistemological interpretation (“Soft SD” versus “Hard
SD”) of the SD methodology is still highly controversial (e.g. Lane, 1994b).
This insight is not only theoretically unsatisfying but also highly problematic from a practical
point of view. Practical consequences are for example
e that a loop like learning approach with respect to the efficacy of organizational
interventions based on SD methodology is strongly hindered,
e that the SD community is split up into different practical schools for the use of SD
methodology in an organizational context,
e that the communities dialogue about its organizational intervention practice is
characterized by a missing standard terminology for the description of the set up of
such studies,
e that there is no clear consensus available in the community with respect to the
possible scope of the use of SD methodology in organizational intervention projects,
e that there are the big obstacles for the usage of SD based organizational interventions
by inexperienced but interested organizational practitioners, which is mainly due to
the fact that the community is not able to provide transparent and consensual answers
about topics like intervention process schemes, role models and skill requirements
etc.
Given the severity of these consequences it seems necessary to investigate in more detail the
fragmented theoretical foundations of the SD methodology. Based on such an analysis it
would then be of high interest to see if an improvement in the organizational practice of SD
can be achieved through a reformulation of the theoretical framework underlying SD. The
success of such a reformulation must be judged against its ability to overcome the above
stated problems and to enable the derivation of satisfying answers to questions like:
¢ What problem contexts are suitable to the applications of SD consultancy (e.g. Lane
1994b; Vennix, 1996; Rouwette, 2003)?
e What organizational contexts are suitable for the application of such approaches (e.g.
Lane 1994b; Flood and Jackson, 1991; Rouwette et al., 2002)?
¢ By what levers does an SD based intervention act in a given organization (e.g. Lane,
1994b; Rouwette, 2003)?
e What is the ultimate goal of the use of such interventions in an organization (e.g. Lane
1994b)?
e What is the optimum process for an organizational use of SD methodology (e.g.
Vennix, 1996; Andersen and Richardson 1997; Rouwette, 2003)?
¢ What role models and skill requirements need to be met in a SD based organizational
intervention (e.g. Vennix, 1996; Andersen and Richardson 1997; Rouwette, 2003)?
The discussion about such a reformulation might also trigger a clarification with respect to
the open question of the epistemological interpretation of SD and its practice. That is not to
say that there is only one interpretation, but that it might be possible to understand the
consequences of an adopted epistemological interpretation in a more consistent way, so that a
conscious and consistent choice of an epistemology and an accompanying practice might be
enabled in the community.
Based on the above considerations the given paper proposes a possible path for such a
reformulation that ranges in its scope from an epistemology to the formulation of a fully
developed intervention model that can be used as a practical framework for the use of SD
methodology in organizational problem contexts. A more concrete description of this path is
given via the following four core hypotheses of this paper:
Hypothesis 1: “Zhe existing SD based organizational intervention practice does not
appreciate the fact, that the application of SD methodology in an organizational
context represents an organizational consultation problem and consequently has to be
treated as such.”
Hypothesis 2: “The missing consensus about the epistemological interpretation of SD
in the SD community as well as the neglected discussion about the organizational
theory underlying SD is a major cause for the existing limitations in the
organizational intervention and consultation practice in the SD community.”
Hypothesis 3: A reinterpretation and reformulation of the existing organizational
theory underlying SD as well as its accompanying epistemological assumptions in the
frame of Luhmann’s theory of social systems (Luhmann, 1984) supplies a con.
subjectivist interpretation of SD and its practice due to the constructivist implications
of Luhmann’s theory and thus provides a more suitable theoretical framework for the
derivation of an organizational intervention practice.
Hypothesis 4: The existing “systemic intervention” paradigm in organizations and its
accompanying practice derived from Luhmann’s theory of social systems provides a
theoretically consistent and operationally elaborated consulting framework, in that
the existing SD based organizational intervention practices can be integrated and
reformulated.
Following these hypotheses this paper will elaborate the foundation and consequences of
these hypotheses in five steps. First a description of the existing organizational practice in the
SD community is given. This is then complemented by some thoughts on the theoretical and
epistemological foundations of SD and their relationship to the SD practice. After that an
outline of the autopoiesis theory of social systems is given. Based on this discussion we will
compare the organizational image of the SD community with the autopoiesis theory of social
systems of Luhmann (Luhmann, 1984). Here we will find striking similarities with respect to
observed organizational characteristics, but a much more coherent theoretical approach on the
side of Luhmann’s theory of social systems. This finding is especially interesting as
Luhmanns theory has since its introduction inspired a rich literature of organizational
intervention approaches that can be summarized under the labels ,,systemic intervention (SI)
or ,,systemic organizational consultation“ (SOC). This body of literature holds a great
potential to establish a practical framework for the use of SD methodologies in different
organizational contexts.
The paper closes with a summary of the main practical conclusions of the proposed new
framework for SD based organizational interventions, summarizes necessary further steps in
the ongoing analysis and points towards further possible discussion directions motivated by
the given analysis.
2. The existing organizational intervention practice in the System Dynamics
community
Jay Forrester defined the field of Industrial Dynamics (later called system dynamics) in his
book “Industrial Dynamics” (1961) in the following way:
“Industrial dynamics is the investigation of the information-feedback character of industrial
systems and the use of models for the design of improved organizational form and guiding
policy.”
This means that System Dynamics relies for the analysis of organizations on two essential
levers, namely the concept of causal feedback loops and the use of formal models to portray
the policy or decision structure of such organizations. Therefore, when we talk about the use
of SD methodology in the frame of organizational consulting efforts we mainly talk about the
use of formal modeling either it be qualitative or quantitative (including causal feedback
loops) of the causal structure of a relevant problem in an organization.
Forrester motivated this focus on formal modeling by pointing out the higher usefulness of
such models relative to verbal descriptions of organizational problems, which are (Forrester,
1961; 1968)
ill defined,
unclear or diffuse in their assumptions,
difficult to communicate through language,
not suitable for an inference of the dynamics contained in the respective problem.
In contrast to that formal representations of the causal structure of an organizational problem
are
a clear expression of the structure and assumptions underlying the model,
easy to communicate due to their formal clearness,
suitable for the inference of the problem dynamics due to the possibility to
incorporate a mathematical framework in this formal modeling process.
Formal models can never be a “realistic” picture of a real world system, but should be
understood, according to Forrester (1961), as subjective mental representation of a considered
system-problem complex. In such an understanding models are subjective communicative
and exploratory devices for the analysis of the causal and dynamical structure of a considered
system-problem complex. Putting it a bit differently one can state, that formal models enable
the exploration of individual or group believe systems (Forrester, 1961).
In practice the employment of such models as exploratory devices in an organizational
context has been found to be helpful from a practical point of view for a variety of
organizational problems, such as (see for example Vennix, 1996; Rouwette et al. 2002 etc.):
e knowledge elicitation and management,
e problem analysis, structuring and visualization (PSM-methodology),
e¢ improvement of internal communication (comparison and alignment of mental
models, establishment of common language),
qualitative and quantitative policy design and analysis,
fostering agreement and commitment,
conflict solution (explication of mental models),
training and learning.
This variety of potential applications has led to the development of a rich literature of field
studies based on the use of SD methodology for organizational interventions (see for example
Sterman (2000) and Rouwette (2002) for extensive lists of SD based field studies).
If one takes a closer look at the way modeling has been used in these field studies it is
possible to outline two principal intervention strategies, namely the use of precoded models
that have been developed outside of a considered system-problem complex and the use of
models that are developed in the frame of the respective system-problem complex. Typical
examples of the first intervention strategy are the use of so called management flight
simulators (Dérner, 1997; Sterman, 2000) for training purposes or the use of system
archetypes for the discussion and reflection of typical organizational dynamics (Senge, 1990;
Sastry, 1998).
The second intervention strategy relies on the construction of formal models tailored to the
specific system-problem complex of interest. This strategy is the most common approach to
using SD models in the frame of organizational interventions and can be applied in two
working modes, namely in an expert and a participative mode. Historically the expert mode
precedes the participative mode of model construction. The main characteristics of the expert
mode are
e the application of a modeling process (see for example figure 1),
e the application of an expert modeling process scheme that relies heavily on back
office modeling work without client involvement,
e a role model for the involved SD practitioner that focuses both on the modeling
process and the content of the modeling process.
Based on these findings, the expert mode of performing SD based organizational
interventions has received, a lot of criticism of which most has been directed towards its
model construction practice and the implicit helping model underlying the expert mode.
The main critique of the treatment of the model construction process in the expert mode is
related to
e the unavailability of a standard model construction process despite the long tradition
in model creation (e.g. Forrester, 1961; Richardson and Pugh, 1981; Wolstenholme,
1994; Sterman, 2000; Keating, 1998 etc.), which hinders the proliferation of the SD
modeling skills into a wider audience of organizational practitioners,
e the claim that guidelines with respect to questions like: “What is good modeling
practice and what makes a good model (e.g. Forrester 1961; Richardson and Pugh,
1981; Morecroft, 1982; Sterman, 2000)? What should be the preferred modeling
mode: qualitative or quantitative (e.g. Vennix, 1996)? What is the optimum size of a
model (e.g. Forrester, 1961; Forrester, 1968; Sterman, 2000)?” are not answered in a
consistent way in the SD community.
Policy
iweme se
Understandings
ef ofa system ay
Policy Problem
analysis definition
‘Simulation System
a
Model
formulation
Figure 1: Seven step modeling process (Start: Problem
definition) proposed by Richardson and Pugh (taken from
Richardson and Pugh, 1981).
The critique related to the helping model (role models) implicitly underlying the expert mode
has interestingly not been formulated as a claim to introduce a different formal helping model
including for example role definitions for the practitioner and the client, but is formulated out
of a less abstract perspective as deficiencies observed in practice. These deficiencies are
e that the results of the modeling process tend to lack client ownership and are therefore
often rejected by the client,
e that the expert role claimed by the SD practitioner not only for the modeling process,
but often also for the content of the modeling process is often rejected by the client.
It is interesting to note, that this critique is typical for the so called “doctor-patient” helping
approach in consultation as it is described and discussed by Schein (1969). Under this
consulting paradigm the consultant is called in to “find out what is wrong, and recommend
how to fix it” (Schein, 1987), an approach that is only promising if certain requirements are
fulfilled e.g. a clear mandate for the consultant from all involved clients, willingness to
accept the diagnosis of a “doctor”, adequate diagnosis by the consultant with respect to the
organizational symptoms, cooperation of client organizations members even if they are part
of the “sickness”, willingness of the client organization to implement the prescription of the
consultant and last but not least the ability of the client organization to stay “healthy” once
the consultant leaves (sustainability of remedy). Usually these requirements are not met in
real doctor-patient relationships too, as can be seen from the frequent critique of the expert
mode of modeling.
To overcome part of the above critique, over the years a more participative modeling
approach has been introduced into the SD community’s organizational intervention practice,
guided by the question: “How can we increase the commitment of a client into the use of a
model?”. This participative mode originally introduced at the end of the 80s into the SD
community differs from the expert mode mainly through
e its employment of a much more interactive model construction process,
e a focus in the SD practitioners role model on the task of group facilitation and client
involvement.
Examples for the successful application of the participative mode are next to others the
reference group approach by Randers (Randers, 1977), the strategic forum approach by
Richmond (Richmond, 1987 and 1997), the modeling as learning approach by Lane (1994a)
and the group modeling approach by Anderson and Richardson (1997) and Vennix (1996).
With the introduction of the participative modeling mode, the usability of SD methodology
for organizational interventions has surely been improved with respect to the problem of
client commitment and ownership. Nevertheless, even in the case of the participative mode
the integration of the practices of organizational interventions in the SD community into an
existing consultation process model e.g. process consultation (Schein, 1987) or systemic
intervention in organizations (Wimmer, 1992) has not been performed. This is pitiful, as an
explicit reference to an available consultation model would offer an optimum framework for
the discussion of deficiencies in the current system dynamics intervention practice as they
have been discussed above. Such a framework would also be very helpful for broadening the
scope of the discussion of the given intervention practice with respect to questions out of the
organizational consultation practice like:
e How should one establish and keep a helping relationship between the client and the
practitioner based on trust?
How is it possible to identify the relevant problems in a considered organization?
How can we define the scope of SD for organizational problem solution?
How can we ensure the suitability of SD for organizational problem solution?
How can we strengthen the operational character of the SD practice?
How can we measure and increase the efficacy of SD based intervention practices?
Despite the fact that such a reference has not been established, it is possible to find
substantial contributions to the discussion of the above questions in the SD literature. A first
example for such a discussion is provided through the study of Rouwette et al. (2002), who
compared the modeling processes and assessments of results of 107 case studies that
employed SD based intervention concepts. Their main findings are
e that the practices employed in the frame of SD based organizational interventions are
still fragmented with respect to process models, role models and terminology,
e there is no research program available that would allow to generate and test
hypotheses about the effectiveness of system dynamics based interventions.
The first finding clearly supports the claim that a coordination frame for the intervention
practice in the SD community in the form of a consultancy model is needed to overcome the
fragmentation in the communities practice. The second finding relates to the fact that in
addition a clear reference between the intervention practice and a theory of social systems is
required, that enables the formulation of hypotheses about the cause and effect logic behind
that practice.
A second example for a discussion of part of the above questions is the work of Flood and
Jackson (1991) who propose to integrate the SD intervention practice into the wider
framework of a “System of Systems Methodologies” that they call Total Systems
Intervention (TSI). TSI comprises besides SD several other systems methodologies out of the
wider System Thinking (ST)' community (including Soft Operations Research (SOR)
methodologies like Soft System Methodology (SSM), Viable System Diagnosis (VSD),
Critical Systems Heuristics (CSH) etc.). The main idea behind this meta-approach is Flood
and Jackson’s assumption that all of these systems methods are only applicable in certain
organizational problem contexts (they claim for example that SD is not capable of dealing
with situations that are characterized by nonconsensual or coercive structures). So, through
the combination of different systems methodologies into one toolbox Flood and Jackson hope
to provide a framework that is applicable in diverse problem contexts (for a more detailed
discussion of this issue see also Flood and Jackson, 1991 and Lane, 1994b). A similar but less
structured approach has been proposed by Lane (1994a and 1994b) who hopes at increasing
the practical effectiveness of SD interventions through the use of cross over techniques e.g.
the combined use of SOR methods or other methods like the hexagon technique in the frame
of an SD based group modeling approach etc.. Flood and Jackson’s approach is a relevant
contribution to the above discussion as it proposes a quasi-consulting model comprising a
clear consulting process. Unfortunately, Flood and Jackson do not elaborate on the topic of
the consultant-client role model issue and through this characteristic constrain their approach
to a quasi-consulting model. It appears further on to be a strong limitation in their model that
they do not formulate a coherent theoretical framework for their system of systems
methodologies, which gives their approach a patchwork character with strong
phenomenological orientation.
In addition to thess more practically oriented comments, Flood and Jackson also formulate a
more fundamental critique of the current SD intervention practice, which criticizes that
e SD diverges in its modeling practice from the normal scientific method by employing
causal relationships in the models that can not be supported by sound empirical data
but are more or less only “believes” of an observer about for example soft factors in
an organization like employee motivation and satisfaction, that are surely of high
relevance for the understanding of organizational dynamics,
e SD produces distorted and one-sided reflections of organizations that are not adequate
representations of the multi-purpose and multi-perspective social complexity of a real
organization,
e SD’s external perspective (outside observer) on organizations is fundamentally
flawed, as it does not offer any means to include the subjective intentions of human
beings that are believed to be a crucial factor in the organizational dynamics,
e SD is blind for the limitations of the perspective out of which it constructs models,
e SD does not offer means to question the taken-for-granted purpose underlying every
SD model.
This critique is of great interest, as it clearly highlights the importance of a profound
understanding of the conceptual and epistemological theory underlying an intervention
' The term ST is meant here in the way as it is used by Lane (1994b), who summarized under
the term ST both SD and Soft Operational Research (SOR) methodologies. This is in contrast
to Senge (1990), who attributed the term ST only to the qualitative modeling practice in the
SD community.
approach. Without such an understanding the relevance, adequacy and reliability of such a
practice appears to be very questionable, which might easily translate into a low credibility of
such an approach. Therefore, it appears necessary to broaden the above requested
establishment of a reference between SD based intervention practices on the one hand and a
full consultation theory on the other hand, by including also references to
e atheory of organizations underlying SD,
e anepistemological theory underlying SD.
Such an integrated approach might also hold additional benefits as it might shed some light
on why for example the observed helping model in the SD intervention practice has emerged
and how it is related to the organizational theory underlying SD. The same is true for the
question if the addressed theoretical constructs can be integrated in a consistent way in their
current formulation and state of reflection. Therefore, to answer these questions and to
diagnose if a reformulation of parts of the theoretical constructs underlying the SD
framework is necessary, we will have in the next section a more detailed look at the
organizations theory underlying SD as well as its epistemological basis.
3. A critique of the organizations theory underlying System Dynamics
Tf one surveys the existing literature in the SD community with respect to its theory of social
systems, it is possible to outline a hierarchical framework consisting of the following
elements:
* asystem metaphor,
e aset of system principles,
e aphenomenology of system characteristics.
A system metaphor is a natural starting point for the formulation of an organizational theory
as it ,,....implies a way of thinking and a way of seeing that pervade how we understand our
world generally.“ (Morgan, 1986). It is a very powerful tool for the study of organizations as
it defines the perspective through which we view organizations. In the case of SD, Forrester
(1961) used servomechanical devices as a metaphor for the complex industrial organizations
that he was interested in. Based on this metaphor he defined industrial organizations as
information feedback systems“ (IFS). An IFS he defined as follows: ,,An information
feedback system exists whenever the environment leads to a decision that results in action
which affects the environment and thereby influences future decisions.‘ (Forrester, 1961).
This system metaphor leads to a picture of organizations as self-steering entities capable of
interacting with the environment in an autonomous way, whereas the self-steering capability
is the result of the internal causal feedback structure of the system. Given this picture of an
organization, it is not surprising, that the resulting SD theory of such an organization is one
that very much focuses on the endogenous causal feedback loops that relate the system state
variables with internal managerial decision making and acting (i.e. information processing).
To put it a bit different, one can state, that the organizational theory underlying SD can be
understood as a theory that tries to endogenously explain system behavior through the
causally closed loop structure inherent in the behavioral network created through the
managerial actions in an organization (Forrester, 1961; Richardson, 1991).
Principle 1
Closed Boundary: In concept a feedback system is a causally closed system. Its dynamic
behavior arises within its internal feedback loop structure. Any causal interaction which is
essential to the behavior mode of the system being investigated must be included inside the
system boundary.
Principle 2
Feedback loop — the structural element of systems: The feedback loop is the basic
structural element in systems. It is a causally closed path coupling system state (level),
observation of this systems state (information), decision based on that information
(“policy”) with an action that changes the considered system state. Dynamic behavior is
generated by feedback. The more complex systems are assemblies of interacting feedback
loops.
Principle 3
Decisions always within feedback loops: Every decision is made within a feedback loop.
The decision controls actions which alters the system levels which influence the decision. A
decision process can be part of more than one feedback loop.
Principle 4
Levels and rates as loop substructures: A feedback loop consists of two distinctly
different types of variables — levels (states) and rates (actions). Except for constants, these
two are sufficient to represent a feedback loop. Both are necessary.
Principle 5
Levels are integrations : The levels integrate (or accumulate) the results of actions in a
system. The level variables can not change instantaneously. The levels create system
continuity between points in time.
Principle 6
Levels are changed only by the rates: A level variable is computed by the change due to
rate variables, that alters the previous value of the level. The earlier value of the level is
carried forward from the previous period. It is altered by rates that flow over the
intervening time interval. The present value of a level variable can be computed without the
present or previous values of any other level variables.
Principle 7
Levels completely describe the system condition: Only the values of the level variables
are needed to fully describe the condition of a system. Rate variables are not needed
because they are computed from the levels.
Principle 8
Rates not instantaneously measurable: No rate of flow can be measured except as an
average over a period of time. No rate can, in principle, control another rate without an
intervening level variable.
Principle 9
Rates depend only on levels and constants: The value of a rate variable depends only on
constants and on present values of level variables. No rate variable depends directly on any
other rate variable. No rate equations (“policy statements”) of a system are of simple
algebraic form; they do not involve time or the solution interval; they are not dependent on.
their own past values.
Principle 10
Rate substructure — system sub-substructure — goal, observation, discrepancy, and
action: A policy or rate equation recognizes a local goal toward which that decision point
strives, compares the goal with the apparent system condition to detect a discrepancy, and
uses the discrepancy to guide action.
Principle 11
Level variables and rate variables must alternate : Any path through the structure of a
system encounters alternating level and rate variables.
Principle 12
Levels and rates not distinguished by units of measure: The units of measure of a
variable do not distinguish between a level and a rate. The identification must recognize the
difference between a variable created by integration (“level”) and one that is a policy
statement in the system (“rate”).
Table 1: List of hierarchical system principles that establish the underlying decision and
action oriented theory of organizations in the SD community (adapted from Forrester,
1968).
Based on this understanding of organizations, the natural perspective in SD on a considered
system is one of an outside observer (“external view”) trying to outline the systems internal
causal decision and action structure. Out of this perspective Forrester defined a set of
hierarchical system principles for an IFS (see Forrester 1968), which are listed in Table 1 and
are meant as guidelines for performing such an analysis of an organization. The basic logical
structure behind these principles is depicted in Figure 2, which shows the endogenous causal
decision and action structure responsible for the behavior of a system (“Structure drives
behavior”) that is delineated from its environment by the depicted causal boundary. This
10
structure is characterized by a self-referential decision operation that translates information
received from an observation into system specific meaning via a differential process that
compares the perceived information to a system internal expectation. This meaning is then
input to a mental decision structure which drives endogenous system action. The result of this
action is a change in a system state, which finally feeds back into the next observation,
starting the next cycle in this circular decision operation.
Causally Closed System Boundary
Goal TN OBeervaton!
i Discrepancy Information :
: Feedback j
H Decision/ Loop '
H Action t
Level '
H Rate '
Figure 2: Causal diagram of the hierarchical system
structure used for the representation of decision structures
in organizations in the SD methodology (for details see
Table 1).
Applying these principles to the study of organizations has led to the identification of a few
general behavioral characteristics of complex social systems, namely (Richardson, 1991):
Complex systems are remarkably insensitive to changes in many system parameters.
Complex systems counteract and compensate for externally applied corrective efforts.
Complex systems resist most policy changes.
Complex systems contain influential pressure points, often in unexpected places, from
which forces will radiate to alter system balance.
e¢ Complex systems often react to a policy change in the long run in a way opposite to
how they react in the short run.
e¢ Complex social systems tend toward a condition of poor performance.
Given this list of system characteristics one can easily recognize the footprint of the system
metaphor described at the beginning of this section, which attributed to social systems a self-
steering autonomous character analog to a servomechanical device constructed by an
engineer to be able to operate in an autonomous way (e.g. auto pilot etc.). This high degree of
autonomy observed in the behavior of organizations in conjunction with the servomechanical
metaphor has led to a working approach in the SD community with respect to organizational
interventions that is guided by the idea of system analysis, design and reengineering. The
ultimate goal of such an intervention approach is according to Forrester (1961):
11
y+. enterprise design to create more successful management policies and organizational
structures.”
The way to achieve that goal is to change the causal system architecture (causal loop
structure) of the system, which is coded into the decision and action practice in the
organization. The natural assumption in SD is that such a change concept can be derived in
an adequate way from the insights generated in the system analysis that is performed while a
SD model representation of the causal system structure is formulated.
This means that the naive role of a system dynamics practitioner is that of a system analyst,
designer and ultimately engineer, which inherently brings with it a helping approach to a
client organization that resembles the before mentioned doctor-patient model (Schein, 1969),
with its described problems and limitations. Based on this result it is possible to state that
the theory of organizations underlying SD is clearly responsible for the restricted
helping model found in the organizational intervention practice in the SD community.
This is an important finding as it points towards the necessity of questioning the adequacy of
this theory if one wants to improve the given SD based organizational intervention practice. It
might even be more fruitful to consider the replacement of the given theoretical framework
with an alternative theoretical construct from which one would require (next to the theoretical
and epistemological questions already raised in the last section)
¢ compatibility with the SD methodology,
¢ compatibility with the main results in the SD community about organizations,
e the provision of a fully developed organizational consulting approach consistently
derived from this theory that is capable of providing an alternative helping approach
for the SD based organizational intervention practice.
This alternative is additionally supported by the fact, that the adapted helping model used in
the participative mode of SD based interventions can not be derived from the above outlined
basic SD metaphor of organizations. It is more or less imported into the practice of SD based
interventions from the theory of group processes and the process consultation framework
developed by Schein (1969 and 1987). Based on these findings it will be of core interest
for the rest of this paper to investigate the ability of the autopoiesis theory of social
systems and its accompanying consultancy framework called systemic intervention
practice to meet the above identified theoretical requirements.
Another point of critique formulated by Flood and Jackson (1991) in the last section dealt
with the principal question, if SD is an adequate methodology for the analysis of
organizations based on the claim that inner states and motivations of human beings strongly
influence the dynamics of such organizations. Here a positive judgment seems possible, as
the policy concept in SD (see Table 1 and Figure 2) outlined above, allows for the portrayal
of subjective decision making practices in organizations including soft factors like motivation
or satisfaction and concepts like bounded rationality (e.g. Sterman, 2000). Nevertheless, such
a decision structure does usually not represent the decision practice of a single individual, but
that of a functional role in that organization that can be occupied in an organization by a
single individual or a group of individuals. This is again an interesting result as it means
that the interaction of a psychological system with its respective social system is
represented in the SD formalism via the decision making concept.
12
To finalize the analysis of the outlined theoretical framework of SD we will turn our attention
in the last part of this section to the critique formulated in the last section about the
epistemological basis of SD. Here two main topics need to be evaluated. First the charge that
SD diverges in its modeling practice from the normal scientific method and second the claim
that the concept of observation can not be reflected in the frame of the existing theoretical
framework of SD and thus leads to a fundamental blind spot in SD. Translated into other
words the latter means simply that the observer of a system and his relationship to the system
including for example goals, interests, sympathy or antipathy are not suitable objects of the
epistemological theory underlying SD. As the second topic requires a clarification of the first
charge before it can be decided, we will start in our analysis with the question of what kind of
explanations (knowledge) SD provides about social systems.
If one talks about knowledge generation, knowledge communication or the formulation of
explanations in the field of SD the primary device for this endeavor is, as mentioned earlier
already, the construction of models. In the existing SD methodology, these models are
available in two basic forms or mixtures of these forms (e.g. policy structure diagrams):
e Causal Loop diagrams (CLDs),
e Stock and Flow diagrams (SFDs).
CLDs portray the causal structure (i.e. feedback loops etc.) of a system and enable mainly so
called qualitative modeling, which means that such models can not be used for quantitative
predictions of the temporal behavior of a system. SFDs focus on the level, rate structure of a
system and do not display the causal structure of the system as clearly. On the other hand
SFDs enable a quantitative modeling of a considered system (as they can be translated into
differential equations) and can therefore be used to generate dynamical behavior patterns of
the system of interest via the solution of the derived differential equations.
Despite their conceptual difference both model types allow the transformation of existing
knowledge about a considered organization (e.g. verbally available expert knowledge, written
knowledge etc.) into a formalized logical or mathematical format. The interesting question
that arises with this formalization step is what kind of knowledge we get. Do we get realistic
knowledge? Or do we simply get useful knowledge and if so, how should we define the term
useful? These two questions and the distinctions behind them are representative for the two
epistemological schools that can be found in sociology. First the functional school with its
objective realism (knowledge is true and an objective representation of the real world). This
view is sometimes also called ,,naive realism“ or ,,ontological view“. It is the perspective of
the natural sciences as it could be found for example in physics before the discovery and the
debate about quantum mechanics”. Secondly, the interpretivism school with its subjective
constructivist paradigm (see also Lane (1994b) and Forrester (1968)). This view is sometimes
2 Interestingly this debate is not confined to the social sciences but can be analogously found
in the natural sciences (Goldstein, 1998; Whitaker, 1996; von Weizsacker, 1985), e.g. the
debate about the interpretation of quantum mechanics in physics that can be summarized by
the question: Describes Q.M. a reality that exists independent of an observer or is it a theory
of possible knowledge of an observer in interaction with a reality that can not meaningfully
be described without an interaction with an observer. In the latter interpretation QM is
sometimes called a theory of the perceptive structure of the human observer. Following this
interpretation one arrives at an interpretation of the scientific endeavor that is similar to the
famous inscription at the oracle of Delphi which read ,,Recognize yourself.“.
13
also called epistemological view. The differentiation between both paradigms can be nicely
laid out on the base of Maturana’s definition of an explanation (Maturana, 2002).
Explanations according to Maturana entail two conditions that must be satisfied together. The
first one is an informal condition that requires the proposition of a generative mechanism.
The second is a formal condition demanding the satisfaction of a criterion of validation. The
key to the distinction between both paradigms lies according to Maturana now in the
interpretation of the respective criteria of validation. Here he distinguishes two criteria
namely, “the experimental scientific method” and “the criterion of validation of scientific
explanations”. Of these the first one leads to an ontological understanding relating an
explanation to the existence of a reality independent of what the observer does, whereas the
second one leads to an understanding of an explanation as a generative mechanism that can
be used to explain the experiential coherences experienced by an observer without any
reference to some independent reality but only to the experiential coherences themselves
perceived by an observer (Maturana, 2002).
This distinction has a high relevance for the above formulated questions about SD as it states
that dependent on the belief one adopts about the adequacy of the above described criteria
one comes to completely different interpretations of the explanations that are provided by SD.
According to the belief system of the author it is most appropriate to adopt the second criteria
and thus define the knowledge provided by SD as useful knowledge in that it provides an
explanation for the experiential coherences perceived by the observer with respect to his past
perception of experiential coherences. If one additionally assumes that belief is based on the
perception of experiential coherences one gets back to the understanding, described at the
beginning of the last section, that SD models need to be understood as exploratory devices for
individual and group belief systems. Based on this reasoning it is possible to rephrase the first
charge made about the divergence of the SD modeling practice from the normal scientific
method, as a simple statement of the fact, that SD, in using this practice, does not adopt as its
basic epistemological criterion of validation the “the experimental scientific method”, but the
“the criterion of validation of scientific explanations”. Due to this argument we get to the
result, that the outlined charge is actually not a charge but simply a statement about a
different epistemological believe.
This leaves us with the problem of the treatment of the observer in the theoretical framework
of SD. Here it appears unavoidable to agree to the objections put up by Flood and Jackson
(1991) with respect to the current practice in the SD community, which does not display the
relationship of the observer to the system of interest. Nevertheless, from a principal point of
view it appears possible to adopt the perspective of a meta-observer that observes the
observer while observing the system of interest. Such a situation could also be portrayed in
the form of a model, in which the displayed policies would be a representation of the internal
goals, interests, sympathy or antipathy of the observer with respect to the system of interest.
A change in the internal configuration of the observer would then translate into a change of
the system representation, i.e. the SD model of the considered organization. Thus we come
to the result that the integration of the observer into the theoretical framework of SD is
theoretically possible if one describes the observer-model relationship out of the
perspective of a meta-observer that observes the observer while observing.
In the remaining sections of this paper we will now concentrate on the question, if the theory
of autopoietic social systems and the consultancy framework derived from this theory are
suitable candidates to fill the above identified theoretical gaps in the theory of organizations
underlying SD.
14
4. Social systems as autopoietic systems
In 1984 the German sociologist Niklas Luhmann proposed in his book “Social Systems”
(Luhmann, 1984) a theory for social systems that tries to understand these systems on the
base of modern system theory. His motivation for doing that was twofold. On the one hand he
wanted to foster the theoretical foundation of sociology which he perceived to be in a
theoretical crisis, on the other hand he was interested to see what the result of a reformulation
of sociology in the framework of system theory would lead to. That he choose system theory
as a framework for that endeavor was motivated by his believe that modern system theory
shows the potential to develop into some kind of a “super theory” (Luhmann, 1984) with
universalistic scope. Due to this potential his theoretical analysis is in part not only restricted
to sociological topics but covers also much more general aspects of system theory. This broad
approach to a new theory of social systems is also reflected by a three level hierarchy of
system terms that Luhmann defined at the start of his analysis. Here we find at the top of this
hierarchy the most general term “system”, which is applicable to all possible instances of
systems and is meant in its most abstract sense. On the second level he placed system terms
like machines, organisms, social systems and psychological systems. These terms are
examples for specific realizations of system types. On the third level Luhmann then focused
on the field of sociology and introduced under the term social system three types of systems,
namely interaction systems, organizations and societies. These system types are according to
Luhmann representative for the possible structuring mechanisms that can lead in society as a
whole to the establishment of a system (Luhmann, 1984). For Luhmann this scheme is not
only a classification of system terms, but also an expression of the different degrees of
abstraction underlying these terms e.g. the definition of a social system is more general and
abstract than that of an organization. Due to this structure in the system term hierarchy
Luhmann points out, that only certain kind of questions are meaningful with respect to the
proposed terms. A comparison in between different terms is for example only meaningful for
terms that are on the same level in the hierarchy. This means that it makes sense to compare
organisms and social systems, but that it is meaningless to compare organisms and
organizations or to try to establish an interaction theory of social systems, due to the different
levels of abstraction underlying these terms.
Following this hierarchical approach Luhmann proposes in his analysis different definitions
for the term system, dependent on which level of his hierarchical scheme he placed the
discussion. In this context we will restrict ourself to the second and the third level of his
hierarchy at which he introduced a very general operator based definition of a system.
According to that definition, a system is a closed sequence of operations of a specific type
(Luhmann, 1984), whereas operations in this context can be e.g. chemical operations
(molecular interactions), communications, thoughts etc.. Based on this definition it is a
straight forward approach to differentiate particular systems according to their base operation
type. An important constraint for the system phenomenology that follows out of this
definition is the fact, that there are no systems with mixed operation types. This statement
also means that a human being can not be considered to be a system as he comprises more
than one operation type e.g. chemical operations and thoughts. Nevertheless, it is possible to
say that human beings comprise psychological (based on thoughts) and biological (based on
chemical operations) systems and that if the biological system dies, also the psychological
system breaks down.
If one adds to the above definition of a system a self-referential reproductive capability one
gets to the definition of a so called autopoietic system (Luhmann, 1984). The term
15
autopoiesis was originally introduced by Maturana (Maturana, 1980), who used that term to
describe the auto-reproductive capability of living cells, which are capable to reproduce
themselves on the base of the molecular interactions that take place in a cell. Following this
concept Luhmann translated the principle of autopoiesis from the biological context into a
general principle of self referential systems. In such an abstract formulation autopoietic
systems are characterized by the fact that they are capable of reproducing their system
operations by the network of their operations that by themselves are produced by the system
operations (Luhmann, 1984). Therefore, autopoietic systems do not only have the capability
to reproduce their system elements in an autonomous way, but are further on also self
organizing with respect to their structure, i.e. there is no structure import into the system from
the environment. Given this more abstract definition Luhmann adopted the term autopoiesis
and claimed that social systems can be understood as autopoietic systems and proposed that
social systems are based on communications. This means that communications are the
elements of social systems and that the communications of a social system produce patterns
of communications that then again produce the communications of the social system. He
further on proposed that psychological systems are also autopoietic systems that are based on
thoughts. Given these assumptions it is possible to state that biological, social and
psychological systems are similar with respect to their autopoietic capability but can also be
clearly differentiated on the base of their specific system operation type (molecular
interaction, communication, thoughts).
This operator based view on systems has important consequences for the characteristics of a
system. First, as autopoietic systems are operationally closed, all processing in such a system
is performed exclusively by system internal operations. There is no operational contribution
of the environment to the system, so that a clear difference between the system and the
environment can be established via the reach of the system and the environmental operations.
The operational closure does not imply that autopoietic systems are closed systems with
respect to for example material flows from the environment. Second, autopoietic systems are
structure determined systems. This means that the continuation of system operations by other
system operations requires the existence of system structures, i.e. patterns of operations.
These patterns of operations are not stable in a static but in a dynamic sense, which means
that they have to be continuously reproduced by the system operations, while they are
operating.
A radical consequence of the above outlined characteristics is that social systems and
psychological systems are operationally separated and can only establish a relevant
environment for each other. This also means that human beings and their accompanying
psychological systems do not belong to social systems but are only a relevant environment
for such systems and vice versa. This brings up the question of how the relationship between
social and psychological systems can be described, especially as it is obvious that social
systems can not exist without human consciousness. Here Luhmann proposes the term
structural coupling, which also goes back to Maturana (Maturana, 1980). Structural coupling
describes the phenomena of structural congruence between a system and its environment.
Typical examples for structural coupling can be found in biology e.g. the eye and the
sunlight, the ear and sound etc.. The structural congruence between the system and its
environment is the result of a mutual history of coevolutionary events in between system and
environment. An important constraint for this adaptive dynamics in the system structuring is
the requirement, that changes in the system structure must be compatible with the systems
autopoiesis.
16
In the case of social and psychological systems, Luhmann proposes that the structural
coupling between the communication in the social system and the thought structure in the
individual consciousness takes place through a process of bi-directional meaning transfer via
language (Luhmann, 1984). This means that communication and psychological systems have
undergone a history of coevolution through the use of language. Therefore, excluding human
beings and their accompanying psychological systems from social systems does not mean,
that they are unimportant or not needed by social systems. As stated already before, the
opposite is true as communicative systems (social systems) can not exist without individual
consciousness with thoughts (psychological systems) although consciousness never directly
participates in communication and communication only influences consciousness through
meaning transfer via language. The term structural coupling establishes in that respect an
antipode to the term autopoiesis as it represents the dependence of a social system on its
environment, whereas the term autopoiesis implies the independence of a social system from
the environment when it comes to its operational capability of self-creation (Luhmann, 1984).
In this context it is important to understand that due to the operational closure of social or
psychological systems, external disturbances to such systems from the environment trigger a
system response that is completely endogenously driven (based on the system internal
operations). This also means that external disturbances can not provide information to a
system but instead can only trigger the production of information in the system that then in a
subsequent step can further on be translated into meaning via a process of system internal
interpretation. Thus, dependent on the operational characteristics of the system, the same
input that one system will consider to be meaningful information will be considered in
another system as noise. Therefore, social systems as well as psychological systems need to
be considered as distinct “meaning worlds” that are based on individual reconstructions of the
environment in the form of internal models used for the interpretation of environmental
“noise” (Teubner, 1997). Based on this insight, it appears meaningful to say that social
systems and psychological systems are capable of observing their environment and
themselves (through the self-referential application of this capability on themselves). The
latter implies that social systems are also capable of self-reflection, through the self-
referential nature of their operational structure. This capability is constrained by the nature of
their operational structure, which constitutes a “blind spot” in their self-reflection that can
only be seen by an external observer who could observe the system in its act of observation.
This mode of observation is sometimes also called “observation of the 2™ order” (Luhmann,
2002). It is important to understand that qualitatively an “observation of the 2™ order” is only
for the observer that is observed an “observation of the 2" order”. For the observer that
observes the other observer it is a normal observation with its own “blind spot” determined
by the operational structure of that observer (Luhmann, 2002).
Due to its capability of self-reflection a social system is also capable of building an internal
model of itself or parts of itself, which means that it is capable to replicate the basic
distinction between system and environment in itself. Luhmann (Luhmann, 1984) calls this
process, following Spencer-Brown (Spencer-Brown, 1979), “re-entry” and states that through
this process, social systems are subject to an internal differentiation process that establishes in
the system functionally differentiated autonomous autopoietic subsystems. An example for
such an internal differentiation would be in the case of the whole society the functional
differentiation into subsystems like science, economy, law or politics. Each of the resulting
subsystems is characterized by Luhmann based on three references, namely the reference to
the overall system which is called “function” of the subsystem, the reference to another
subsystem which is called “achievement” and last but not least the reference to itself which is
17
simply called “self-reference”. This three dimensional referencing of a subsystem is of great
interest as it leads to a picture of a social system as a functionally differentiated system with
no hierarchical structure, but a structure based on functionality and achievement for its
autonomous and autopoietic subsystems. In such a picture every subsystem is of equal
importance due to its exclusive functions and achievements (Luhmann, 2001). They are
further on due to their autopoietic nature also “meaning worlds” of their own, that
consequently develop also their own communicative codes (Luhmann, 2001).
A further important step in Luhmann’s theory is the application of the above concepts to
organizations as they can be found in highly developed social societies in the form of e.g.
industrial companies, governmental agencies, non-profit organizations etc.. According to
Luhmann’s hierarchical scheme, organizations are social systems whose operational basis are
given by the communication of decisions’ (Luhmann, 1998). They produce decisions from
decisions and attain in this way operational closure. This view on organizations implies also
that any kind of organizational structuring is not only the result of the internal organizations
decisions, but is actually established by these decision structures independent on if they are
formal or informal decisions (Luhmann, 1998). It is further on important to understand in the
context of this organizational understanding that the applied decision model is not one that is
based on an intra-psychological model of an individual, but one that is defined on the base of
the relationship between the meaning of an action and an expectation on that action. It is
through this understanding of a decision operation in an organization, that it is possible to say
that individual consciousness and action are a prerequisite for the organizational autopoiesis
but not part of it. Individual action in that respect is only the trigger for an organizational
decision and nothing more.
If we come back now to the questions raised at the end of the last section about what
relevance the above theory for the SD community can have, the author aims at proposing a
far-reaching answer. The above characteristics imply that organizations need to be
understood as autonomous, functionally differentiated systems with the ability to reconstruct
meaning and evolve through their interaction with themselves or the environment (including
the psychological consciousness of the human being that couples to these systems). This view
on organizations has profound implications for the analysis and work with organizations. It
especially has the implication that it is self-referential in that it is applicable to itself, i.e.
that the autopoietic theory of social systems can be understood as a meta-theory for
formulating models and reconstructing meaning for functionally differentiated and
closed worlds of meaning. The theory in that respect is also a suitable theory to reflect
its own way of observing social systems. This means that through this theory we become
aware of our way of observing the social systems that we are dealing with (Willke, 1997).
On the same token, one is able to consider SD to be an autopoietic entity of its own, which is
part of the wider functional (autopoietic) subsystem science in our modern western society.
3 Luhmann proposes in his theory to understand the term decision as a relational term that is
based on the relationship between the meaning of an action and the expectations on that
action. In that respect a decision is a selection between the two alternatives of conformity and
deviation in such a situation. An important aspect of this definition is that the attribution of
meaning to an action can be done by the acting individual or an observer and is thus no
absolute characteristic of an action. The same is true for the definition of the expectation on
the action. This means that an individual is able to perform an organizational decision
through his action without knowing that he made such a decision (Luhmann, 1984).
18
Based on these finding we can reformulate the question posed in the last section in the
following way: “What can we learn from reflecting and reformulating the theory and practice
of SD in the frame of Luhmann’s system theory of social systems?”. This question implies a
far-reaching and voluminous scientific program which can surely not be treated in a
comprehensive way in the scope of this paper. It also reaches further then the question placed
at the end of the last section about the suitability of this theory to meet the theoretical
requirements for an SD meta-theory deduced from the given theoretical framework of SD.
Therefore, keeping this overarching programmatic question in mind, we will focus in the
remainder of this paper on the treatment of three questions believed to provide a suitable start
for the discussion of the above raised question. These questions are:
e How does the epistemological foundation of SD fit into the epistemological theory
behind Luhmann’s theory of social systems?
e In what respect differs the servomechanical metaphor of SD and its related
theoretical understanding of organizations from Luhmann’s theory of social
systems?
e Is the systemic intervention practice derived from Luhmann’s theory of social
systems a suitable consultancy framework for the SD practice of organizational
interventions?
Starting with the epistemological question it appears clear from the above outline of
Luhmann’s theory, that the subjective interpretation of the SD practice as it has been outlined
in the last section is perfectly compatible with the claims made in Luhmann’s theory. Both
theories do not comprise ontological claims, but make only claims about the subjective
structure of either belief systems about a system (in the case of SD) or internal models of a
system (in the case of Luhmann), which is actually the same. Therefore, it appears that from
an epistemological perspective both theories are not in contradiction to each other.
Nevertheless, a difference in the kind of knowledge that is created in both theories can be
stated. The reason for that is that in SD the produced knowledge is mainly based on normal
observations, whereas in Luhmann’s theory the preferred observational mode is that of an
observer of the 2™ order. This means that the knowledge created in the latter theory
comprises a higher order of reflection than that typically derived in SD. This limitation in the
SD practice has already been mentioned in the last section, when we discussed the
implications of Flood and Jacksons (1991) critique of the SD practice. Here we also stated
that this deficiency in the SD practice is not a conceptual but mainly a practical one. This
means that from a principal point of view it could be overcome by an additional level of
reflection in the SD methodology, which would have to include a more thorough observation
of the model creation process, including an appreciation of its constraints inherently
contained in the perspective of the observer of the system of interest.
If one further on compares the major characteristics between the servomechanical metaphor
underlying SD and the organizational metaphor resulting from Luhmann’s work, one can
outline the following similarities:
e in both theories the system boundary between the system and the environment is
defined on the base of the operational closure of the system operations, that means
through the reach of these operations (endogenous versus exogenous system
variables),
e the dynamics of organizations is considered in both theories to be structure
determined and to be endogenously generated, in that respect, that external
19
influences can only establish a disturbance but not an instructive authority for the
system,
e based on this characteristic both theories attribute to social systems the capability of
autonomy,
both theories consider organizations to be functionally differentiated,
both theories use an impersonal perspective to display the system structure,
both theories use model building to analyze the structure of a system.
Despite these similarities there are also significant differences:
in the SD practice the concept of the observation of the 2" order is not known,
in contrast to the practice in SD in Luhmann’s theory the relationship between an
external observer (e.g. consultant) and the system of interest is part of the theory,
e the modeling process in SD is mathematically oriented and highly formalized,
whereas the modeling process in Luhmann’s theory is an abstract process of a
system, that is not formalized but an act of cognition,
e the term autopoiesis with its reproduction function has no equivalent in SD, although
the causal circularity of the systems communicative operations resembles the
feedback loop concept from SD,
e the policy and decision structure of SD diagrams relies heavily on intra-
psychological decision models in contrast to the more contextual decision model in
Luhmann’s theory.
Summarizing the above one can state that both theories show on a phenomenological level
some striking similarities with respect to their claims about the structure and dynamics of
organizations. On the conceptual level there are some similarities but also some significant
differences. Therefore, if one aims at reformulating the SD paradigm in the frame of
Luhmann’s autopoiesis theory of social systems it appears necessary to reflect in more detail
on the identified conceptual differences to see if the SD practice can be consistently derived
from the theory of autopoietic social systems. An important question in this context is for
example the question of how SD methodologies like CLDs and SFDs diagrams have to be
interpreted in the frame of Luhmann’s theory. An answer to this question is not in the scope
of this paper and will be subject to further investigations. The same is true for a detailed
answer to the question about what role the practice in SD can play in the frame of the
autopoietic theory of social systems. All in all it appears extremely promising to compare the
structure and claims of both theories to enable a detailed reformulation of the current SD
practice in the frame of Luhmann’s theory, which would for example enable also the
integration of the existing SD organizational intervention practice into the systemic
interventions practice in organizations. This last topic, which also points to the third question
,given above, will be the subject of the next section.
5. ,,Systemic intervention* practice in organizations
Since the introduction of Luhmann’s theory of social systems in 1984 the interpretation of
Luhmann’s theory has been in the focus of organizational consultants. These efforts have led
to the definition of an independent consulting style derived from the main implications of the
theory. This consulting practice is called “Systemic Intervention” (SI) or “Systemic
Organizational Consultation” (SOC) as already mentioned before (see section 1). The term
systemic relates in this context to the system based view on the considered organizations in
20
that a particular consultation project takes place. The term intervention is defined’ as (Wilke,
1992):
“\..a goal driven, directed communication between psychological or social systems, that
respects the autonomy of the system that is the target of the intervention.”
Based on the above cited work and the definitions provided in the last section, the main
characteristics of the SI approach can be specified as follows:
a) Creating a consultation relationship and defining clear role models is part of ST
Due to the operational closure of organizations a consultant is from a principle point of view
not able to impose any kind of instructive change model on an organization. Therefore, it is
of great importance that the consultant defines at the beginning of every organizational
intervention project an adequate relationship to the client organization that reflects this fact.
Usually this is done via the establishment of a consultation system with selected
representatives of the client system (Simon, 2002; Wimmer, 1992). The natural role model
for the consultant in such a setup is that of a structural coupling to that consultation system,
which means that the consultant enters into a process of coevolution with the consultation
system. In this situation it is the main task of the consultant to manage this process in an
active fashion (process consultant). With respect to this management task it is important for
the consultant to understand, that every disturbance that he provides to the consultation
system either it be intentional or unintentional might be interpreted by the consultation
system as a relevant information and might trigger a change in the system structure and
behavior. This awareness is especially important at the beginning of a consultation project,
when the consultant has not developed his intuition about the information processing
characteristics and capabilities of the consulting system and the client organization.
b) Consciousness about predominant schemes of observation within a client system is key
and is produced through observation of observation patterns
Through the establishment of a consultation system that comprises members of the client
organization the consultant is able to create a situation in which usually typical observation
patterns of the client organization become apparent in the consultation system so that they
can be observed by the consultant 2nd order observation”). This offers the possibility to
bring these patterns and their consequences (e.g. “blind spots”) to the attention of the client
organization members that participate in consultation system.
c) SI focuses on changing observation schemes
Based on the identification of predominant observation schemes in the consultation system it
is possible to discuss with the members of the client organization alternative observation
schemes that might offer some potential benefits for the client organization.
d) Sl aims at perturbing old and introducing new communication patterns
Analog to the identification of predominant observation schemes it is also possible to address
the issue of persistent communication patterns in the consultation system. Here two principle
4 Translation by the authors from the original German text.
21
intervention path are available to the client. First, the possibility to irritate these patterns in
the consultation system via system interventions by the consultant that might lead to
permanent changes in the communication patterns of the consultation system and potentially
the client organization. Second, the discussion of the observed communication patterns in the
frame of the consultation system based on the consultants observations. These discussions
might also lead to changes in the prevalent communication patterns in the consulting system
and potentially also in the client organization. Such changes are usually very difficult to
achieve as they are intimately related to the system structure of the client organization that
due to its self organizing character does not allow the simple import of structure.
e) SI consultants focus on becoming experts of process architectures
As it was said before, one of the main levers for organizational change is the intentional use
of communication events that are capable of triggering an internal dynamic in the
consultation system of interest. Out of this perspective the consulting process is simply a
coordinated sequence of interventions i.e. organizational communication events. Such a
sequence is under the SI paradigm usually hierarchically structured according to the
following categories (KGnigswieser and Exner, 1998):
1. The process or event architecture (also called intervention architecture or social
architecture).
The architecture of the consultation process defines the event structure of the
considered consultation process. That means it describes when an event will cause a
disturbance in the considered system and via what means such an event is organized.
Typically it is possible to discern different events along four dimensions (TSSC),
namely:
e Time (T): When does an event takes place and what is the duration of the
event.
e Space (S): The spatial condition under which an event takes place e.g. indoor,
outdoor, office, meting room ete..
e Social (S): The number of people that are involved in the event and how they
are involved. Here certain social entities have to be distinguished e.g. single
person, pairs, groups etc..
e Content (C): What is the content that is discussed in the frame of an event and
from what perspective/s is it presented.
The process or event architecture defines the amount and type of social spaces in
which the structural coupling of the involved systems (consultant and client
representatives) can take place via planned (and unplanned) interventions. Typical
architectural elements are: Group-Workshops, Interviews, Steering Board sessions
etc.. K6nigswieser and Exner (1998) differentiate in total 12 typical architectural
elements. A process architecture is usually displayed by an event diagram. Figure 3
displays a schematic example for such a diagram.
2. The intervention design.
The design of an intervention describes the “interior” of a certain intervention that
takes place in a social space provided by the process architecture. The structure of a
design can range from simple to complex designs, whereas the latter can comprise
Sub-Designs. Like architectures, designs also need to be specified along the four
22
dimensions TSSC. An example for a simple design would be an interview comprising
two people of a duration of 2 hours taking place in an office with the aim of data
collection. An example for a complex design could be a daily WS with six
participants, taking place in a hotel, that comprises four interventions of 2 hours
length each having a specific design. K6nigswieser and Exner (1998) provide in their
book a collection of about 70 designs for a multitude of intervention goals.
Intervention Architecture
Architecture
Elements
Element | H#_@-@ ¢ ¢ @ oo o_
Element 2
Element 3 @-@-@
Element 4 hs
Element 5 OOOO
Blement 6 | @-@-__@ @ @ @ —@ _|
Element 7 f fi
Time
Figure 3: Schematic event diagram (intervention architecture) of a
consultation process. The different shapes and colors display
different architectural elements (after K6nigswieser and
Exner, 1998)°.
3. Intervention Techniques
Interventions techniques are used by a consultant while working in an intervention
design and are used to facilitate the process of the ongoing intervention. Typical
examples of intervention techniques are story telling, use of metaphors, circular
questioning, reframing, splitting, positive attributing etc. (for further details see
K6nigswieser and Exner, 1998). Intervention techniques are used in the frame of an
ongoing intervention to punctuate the process, regulate distance between the
consultant and the client, to add perspectives or emotions to the process.
This process framework is highly operational and characterized by a clear standard
terminology that is very well accepted throughout the SI community (e.g. K6nigswieser and
Exner, 1998). The typical working mode of a consultant working in this framework is, as it
was outlined before already, that of a process consultant or architect that is responsible for
directing the involved client representatives through for example a WS design etc., a working
approach or role model very much in agreement to the one used in the participative modeling
mode in SD.
5 In the original publication of Kénigswieser and Exner (1998) the symbols are standardized
and have a distinct meaning e.g. a triangle represents a group of three people, whereas a
rectangle represents four people. Such a pictographic use of displayed symbols is not adopted
here.
23
6. A new operational framework for the use of SD methodologies in organizational
interventions
Given the above analysis we can now pose the question of what the benefits of using the SI
consulting approach as a framework for SD based organizational interventions could be. Here
we find four main benefits, namely:
1) The SD consultant can be understood as a process consultant
When we discussed in section 2 and 3 of this paper the different role models used in the
expert and the participative modeling mode in the existing SD practice, we emphasized the
fact that the process oriented role model used in the latter mode appeared to be unrelated to
the organization theory underlying SD. Based on this observation it appears to be of high
interest that a process oriented consulting approach seems to be the natural working mode in
the frame of the SI, which is not merely a coincidence but a clear consequence of the
autopoietic theory of social systems underlying the SI framework and its claims for the
relationship between an observer and a social system of interest.
2) The SD modeling process becomes part of an intervention architecture
In section 2 we identified the missing reference to a full consulting process model as one of
the main limitations of the organizational SD intervention practice. This is mainly due to the
fact that the process model underlying SD based interventions is very much restricted on the
issue of how to build a good SD model and therefore neglects other important process
questions that arise in the frame of an organizational consultation project (e.g. see the
questions raised in section 2). This situation can now be easily overcome by applying the
process model used in the SI framework and acknowledging the fact that the existing process
model in the SD community is nothing more than a process description in the frame of an
intervention design according to the SI terminology. This fact enables a consultant to fit the
given SD practice nicely into the intervention architecture of an organizational intervention
project without loosing sight of the other important elements contained in a full intervention
architecture.
An illustration of the practical consequences of the above claim is shown in figure 4, which
depicts the group modeling approach of Anderson and Richardson (1997) in the form of an
event diagram of the implicit intervention architecture used by them. The diagram shows all
in all 7 architecture elements (to be filled with intervention designs) that are more or less
clearly mentioned in the group modeling process description of Anderson and Richardson
(1997). It is interesting to see that the architecture diagram covers much more aspects of the
typical group modeling approach than the modeling process itself. That means that Anderson
and Richardson (1997) implicitly use a more complex process model for their overall
organizational intervention without explicitly mentioning it. The implicit nature of the other
elements is also shown by the fact that Anderson and Richardson (1997) describe the design
of the modeling WS very detailed through a compilation of more than 20 so called scripts,
whereas the other architecture elements are only shortly mentioned in the text without any
details. The application of the SI framework does provide in this context not only a nice way
of displaying this disproportion between the different architecture elements, but it also
enables a structured analysis of the given approach with respect to missing architecture
elements. The description of the process model of Anderson and Richardson (1997) clearly
misses for example an important architecture element usually called in the SI frame “problem
24
diagnosis”. The problem diagnosis is used to clarify issues like: who is the client, what
problem needs attention of the consultant, what is the organizational context of that problem,
how do different stakeholders in the organization perceive the problem of interest etc..
Therefore, based on such an event diagram it is very easy to discuss the proposed consulting
process model for a given consulting project that might include SD methodology. Such an
integrative perspective might also be very useful to explain to potential clients the importance
and relevance of the use of SD in the frame of a consultancy project.
Intervention Architecture
Architecture (Anderson and Richardson (1997))
Elements
Selection of gatekeeper ly
Interviews with managers
Diagnosis (selection of CY
people for WSs Sy
Pre-WS meetings with Gate-
keeper and modeling team O-O-O-2
Contract for WSs A.
Mewings ofS IWO-O-O-O-0-0-0-0-O—
‘modeling team
Group modeling WSs { H }
Time
Figure 4: Schematic process architecture of the group modeling approach
by Anderson and Richardson (1997) (architecture elements
adapted by the authors).
Another important aspect of using the proposed three hierarchical categories of the SI
framework is the fact, that it supplies a clear standard terminology that can help to
standardize the fragmented terminology in the SD community. A nice example for this
problem is depicted in table 2, which shows the corresponding terminology in the group
modeling approaches of Vennix (1996) and Anderson and Richardson (1997). Here two
findings can be emphasized. First, not even in the group modeling practice, which is only one
of the practical schools in the SD community, a consistent terminology is in use. Actually
Vennix (1996) and Anderson and Richardson (1997) use for all three SI categories
completely different terms. Second, both approaches do not use a clear hierarchical
distinction between the terms they apply. Both approaches are in that respect much more
oriented towards a phenomenological perspective than the SI approach. Given these
observations it appears very promising to call for the use of the SI terminology in future SD
projects as this would establish a clearer hierarchical positioning of a discussed topic in a
project description. Such a clarification of the logical positioning would decrease the danger
of misunderstandings. It would also support the standardization of the elements in these
categories, as is has been proposed for example by Anderson and Richardson (1997) for the
documentation and use of scripts. The use of the SI framework would enable an extension of
such a standardization or cataloguing effort to the two other categories of the SI framework.
Here especially a standardization or cataloguing of process architectures for SD based
organizational interventions could be of high interest and could trigger a loop like learning in
the SD community with respect to such projects.
25
ST Vennix (1996) Anderson and Richardson (1997)
Architecture | Project design Process phase model
Design Group process techniques/Formats Scripts
Technique Facilitation technique Group tasks
Table 2: Comparison of terminology used under the SI and the group modeling paradigm
(K6nigswieser and Exner, 1998; Vennix, 1996; Anderson and Richardson, 1997).
3) The focus of attention lies on observations of observations and thus on belief systems
As stated already in the last section, it is one of the core beliefs in the SI paradigm that a
consultant can best provide added value to an organization by supporting an organization in
identifying its blind spots in its observation practices. Through that belief, which is also
consistently supported by the underlying organizational theory of autopoietic social systems,
the SI paradigm strengthens the subjectivist interpretation in the SD community. This
interpretation (described already in section 2) states, that SD modeling is not targeted at a
formalized description of some “real system properties”, but at uncovering and aligning of
believe systems (based on unconsciousness observational patterns) about such systems in an
organization.
4) SD adds value to SI by formalizing the observation analysis process of organizational
believes and adding dynamical analysis capability to that process
As pointed out above the belief analysis process in organizations is in the focus of both the SI
and the SD approach. Therefore, it is of great interest that the more formalized analysis
approach offered by the SD methodology might also add some value to the SI approach. This
can take place through the formal rigor of the SD methodology that allows also (dependent
on if the modeling approach is qualitative or quantitative) for the inference of the behavioral
complexities implicitly contained in a given belief system about a system or organization of
interest. Thus, the use of such a more mathematically oriented analysis approach provides an
additional element for the analysis of the logical consistency of a given group believe system
in an organization of interest.
Summarizing the above, it appears very probable to produce some added value by
incorporating the given organizational intervention practice in the SD community into the
framework of the SI paradigm. The same appears to be true with respect to the uncertainty
observed in the SD community about the use of SD in the frame of organizational
consultation projects described in section | of this paper. Additionally, it appears also to be of
high interest to continue with the research program outlined at the end of section 4 in this
paper that called for a reinterpretation of the SD theory and methodology in the frame of the
autopoiesis theory of social systems of Luhmann.
7. Conclusions and further steps
It was the ultimate goal of this paper to propose the integration of the existing SD based
organizational consulting practice into the systemic consultancy framework derived from
Luhmann’s autopoiesis theory of social systems. To do that, the paper has analyzed the given
practice at
26
e the level of the consulting practice (standard terminology, role model, process model
etc.),
e the level of the organizational theory underlying SD (servomechanical theory versus
autopoiesis theory),
e the level of the epistemological basis of SD (constructivist interpretation of SD
versus the ontological “hard SD” and OR like interpretation style),
and has shown that such an integration does not lead to theoretical inconsistencies at either of
these levels. The paper further on proposes future work on all three levels with focus on:
e A translation of existing organizational SD practices into the proposed standard SI
scheme to build a standard catalog of intervention architectures, designs and
techniques.
e A more detailed reinterpretation of the SD methods in the frame of the autopoiesis
theory of Luhmann (e.g. what is the meaning of CLDs or SFDs in this context).
e A deeper discussion of the epistemological interpretation of SD in the frame of
Luhmann’s theory, which might foster the understanding of SD as a formal
methodology to causally analyze and dynamically explore the structure of believe
systems.
To close this paper the author would like to point to the fact that according to the presented
theoretical framework of Luhmann’s autopoiesis theory the SD community and the
community that has emerged around the work of Luhmann both constitute autonomous
autopoietic subsystems of the society subsystem science. Through that characteristic they
both represent distinct meaning worlds in between which “creative misunderstandings
...offer an escape from the impossibility of ever being able to translate accurately the
language of one world into another domain’s communication system. One discourse uses the
meaning materials of another as a provocative stimulus to reformulate it as something new in
its own internal context. Since a real translation is impossible, something is invented. This
inventiveness creates the surplus value ..., which is added to the autopoietic dynamics within
and between systems.” (Teubner, 1997).
8. Acknowledgment
The authors would like to thank Dr. Christoph Klingenberg of Deutsche Lufthansa AG for
supporting the given work.
9. Literature
Andersen, D. F. and G. P. Richardson 1997. Scripts for group model building, System
Dynamics Review, 13(2), 107-129.
Dérner, D., R. Kimber, and R. Kimber 1997. The Logic of Failure: Recognizing and
Avoiding Error in Complex Situations, Basic Books (Perseus Books Group), New York.
Flood R. L. and M. C. Jackson. 1991. Creative Problem Solving: Total Systems Intervention,
John Wiley & Sons Ltd., Chichester.
Forrester J. 1961. Industrial Dynamics, MIT Press, Cambridge, MA.
27
Forrester J. 1968. Principles of Systems, MIT Press, Cambridge, MA.
Goldstein S. 1998. "Quantum Theory Without Observers", Physics Today March 1998, pp.
42-46 (Part 1) and April 1998 pp. 38-42 (Part 2).
Keating, E. K. 1998. "Everthing You Ever Wanted to Know about Developing a System
Dynamics Model, but Were Afraid to Ask.". 16th International Conference of the System
Dynamics Society, Quebec '98, Quebec City, Canada, System Dynamics Society.
K6nigswieser, R. and A. Exner 1998. Systemische Intervention, Klett-Cotta, Stuttgart.
Lane, D. 1994a. Modeling as Learning: A consultancy methodology for enhancing learning
in management teams, in: Modeling for Learning Organizations, (Eds.) J. D. Morecroft and J.
D. Sterman, 85-117, Productivity Press, Portland, OR.
Lane, D. 1994b. With a little help of our friends: How system dynamics and “soft” OR can
learn from each other, System Dynamics Review, 10(2-3): 101-134.
Luhmann, N. 1984. Soziale Systeme, Suhrkamp Verlag, Frankfurt am Main.
Luhmann, N. 1998. Die Gesellschaft der Gesellschaft, Subrkamp Verlag, Frankfurt am Main.
Luhmann, N. 2001. Aufsdtze und Reden, Philipp Reclam jun., Stuttgart.
Luhmann, N.2002. Einfiihrung in die Systemtheorie, Carl-Auer-Verlag, Heidelberg.
Maturana H. R. and F. J. Varela 1980. Autopoiesis and Cognition: The Realization of the
Living, Dordrecht: D. Reidel, Vol. 42: Boston Studies in the Philosophy of Science.
Maturana, H. R. 2002. Autopoiesis, Structural Coupling and Cognition: A history of these
and other notions in the biology of cognition, lttp://www.isss.org/maturana.htm (Date:
15.06.04).
Morgan, G. 1986. Images of Organizations, Sage Publications
Morecraft J. D.W. 1982. A critical review of the diagramming tools for conceptualizing
feedback system model, Dynamica, 8: 20-29.
Randers, J. 1977. The potential in simulation of macro-social processes, or how to be a useful
builder of simulation models, Oslo, Norway: Gruppen for Ressurstudier.
Richardson, G. P. and A. L. Pugh III 1981. Introduction to System Dynamics modeling with
DYNAMO, MIT Press, Cambridge, MA.
Richardson, G. P. 1991. Feedback Thought in the Social Sciences and Systems Theory,
University of Pennsylvania Press, Philadelphia.
Richmond, B. 1987. The strategic forum: from vision to strategy to operating policies and
back again, High Performance Systems: Lyme, NH.
28
Richmond, B. 1997. The strategic forum: aligning objectives, strategy and process, System
Dynamics Review, 13(2): 131-148.
Rouwette, E., J. A. M. Vennix, and T. van Mullekom 2002. Group model building
effectiveness: a review of assessment studies. System Dynamics Review, 18 (1): 5-45.
Rouwette, E. 2003. Group model building as mutual persuasion. Ph.D. Thesis, Wolf Legal
Publishers, Nijmegen.
Sastry A. 1998. Archetypal self-reinforcing structures in organizations, In: Proceedings of
the 16'" International Conference of the System Dynamics Society, Quebec.
Senge P. 1990. The fifth discipline: The Art and Practice of the Learning Organization, New
York: Doubleday Currency.
Simon, Fritz B. 2002. The De-construction and Re-construction of Authority, Soziale
Systeme, 8 (H2), 283-293.
Schein, E. H. 1969. Process Consultation: its role in organization development, Addison-
Wesley, Reading (MA).
Schein, E. H. 1987. Process Consultation (vol.II), Addison-Wesley, Reading (MA).
Spencer-Brown G. 1979. Laws of Form, New York.
Sterman J. D. 2000. Business Dynamics: System Thinking and modeling for a complex world,
Boston, McGraw-Hill.
Teubner, G. 1997. Law as an autopoietic system, report on a presentation, ESRC Business
Processes Resource Centre: http://bpre.warwick.ac.uk/Isesg3.html
Vennix, J. A. M. 1996. Group model building: facilitating team learning using system
dynamics, John Wiley & Sons Ltd.., Chichester.
von Weizsicker, C. F. 1985. Aufbau der Physik, Carl Hanser Verlag, Miinchen-Wien.
Whitaker, A. 1996. Einstein, Bohr and the quantum dilemma, Cambridge University Press.
Willke, H. 1997. Autopoiesis and organised complexity, report on a presentation, ESRC
Business Processes Resource Centre: http://bpre.warwick.ac.uk/Isesg3 .html
Wimmer R. 1992. “Was kann Beratung leisten? Zum Interventionsrepertoire und
Interventionsverstdndnis der systemischen Organisationsberatung“, in: Ders. (Ed.),
Organisationsberatung. Neue Wege und Konzepte, Gabler: Managementperspektiven,
Wiesbaden, Germany, 59-111.
Wolstenholme, E. F. 1994. A systematic approach to model creation, in: Modeling for
Learning Organizations, (Eds.) J. D. Morecroft and J. D. Sterman, 175-194, Productivity
Press, Portland, OR.
