The social role of simulation models

Steffen Bayer
Imperial College Business School
South Kensington campus
London
SW7 2AZ
[email protected]

Timothy Bolt
School of Management
University of Southampton
Southampton
SO17 1BJ
[email protected]

Maria Kapsali
Imperial College Business School
South Kensington campus
London
SW7 2AZ
[email protected]

Sally Brailsford
School of Management
University of Southampton
Southampton
SO17 1BJ
[email protected]

1
 The social role of simulation models

Abstract
This paper suggests a classification of the social roles simulation models can play. Two dimensions are
distinguished according to the context and use of models: models can be boundary objects or
representative objects and they can be epistemic or technical objects. These two dimensions allow a
classification of four types of model roles. Models can be ascribed different roles over time and different
roles by different stakeholders involved in their development and use potentially leading to
misunderstanding and conflicts. The suggested classification framework can be applied to a variety of
problems around the use models including the discussion of the differences between System Dynamics
models and Discrete Event Simulation models and the comparative analysis of model use.

2
Introduction
Simulation models and simulation modelling are used in many different ways. Context and objectives of
modelling projects vary as much as approaches and tools used. The system modelled is only part of what
determines the modelling process and the modeller often is only one of the stakeholders influencing or
being influenced by the model. Other model users, such as decision makers or students in a teaching
context or participants in a group model building project, may also interact directly with the model. Many
other stakeholders might be influenced indirectly by decisions made based on the model. This paper aims
to understand the social roles of simulation models and puts forwards a framework to classify these roles.
In doing this, the paper continues the work of Zagonel (2002) on group model building and builds on the
literature on boundary objects. In addition, this paper suggests that the distinct literature on objects as
epistemic and technical objects allows for the development of a classification which captures more of the
differences in the social role of models.

The paper assumes that not only would reflecting on the use of simulation models and the role they are
given by stakeholders support the modeller in making a modelling project more effective, more
implementable or more insightful for the different stakeholders, but could also help to avoid conflicts and
misunderstandings if different stakeholders understand the role of simulation models and the modelling
process differently or if the role of a model evolves over time within a project.

The paper first discusses the application of the concept of boundary objects to models and then reviews
insights from the literature on epistemic and technical objects in order to develop a framework for
analysing the social role of models. This framework is then applied to two issues: first the reoccurring
discussion on the differences between system dynamics and discrete event simulation and then an
empirical study comparing to consultancy projects developing simulation models for the same system
with different stakeholders.

Models as boundary objects or as representation of reality

Models can be used to make predictions about the real world and allow decision-makers to experiment in
a safe, quick and low cost way with different courses of action. However, as has been shown in the case
of engineering, (Dodgson et al., 2007b) simulation modelling can also help to shape the conversation
between stakeholders in problem solving and foster collaboration. Zagonel (2002) identified a tension
between conceptualising modelling as representing reality and as negotiating a social order. He contrasted
simulation models as boundary objects or as micro-worlds (see table 1, summarized from Zagonel 2002).

3
Table 1: Zagonel’s (2002) conceptualization of SD group model building

Models as “micro-worlds” Models as “boundary-objects”

• problems are pre-existent in the system • problems emerge from debate and
discussion
• create a realistic representation
• come upon a shared understanding
• accurately address the content of the
issue • understand our complementary and
competing views,
• strive to find the “correct” solution
• build a joined picture reconciling our
• focused upon the results and outcomes different views
Therefore, our group process needs to be effective The process we use to “negotiate” this model is as
at getting at the answers we need. important, if not more important, than the accuracy
of the model as a representation of our reality.

Boundary objects are artefacts shared between communities of practice, which have their own specific
informational codes (Star and Griesemer, 1989; Carlile, 2002; Sapsed and Salter, 2004). Boundary objects
can address some of the difficulties of communicating and creating knowledge across (disciplinary and
organisational) boundaries. These difficulties include not only the syntactic and semantic challenges of
having to overcome differences in language and interpretation, but also the challenges inherent in creating
new shared knowledge and dealing with the negative consequences for the participants arising out of this
shared knowledge creation process. (Carlile, 2002) Boundary objects such as repositories of knowledge,
standardized forms and methods, objects or models or boundary maps have been shown to support
interdisciplinary working (Star 1989). However, while boundary objects can be the basis of negotiation
and knowledge exchange, they can also be ineffectual, precisely because their role is at the margins of
communities, and their use depends on the frequency of interaction and level of understanding within
groups (Sapsed and Salter, 2004).

In a variety of domains, modelling has been shown to be able to support situations where disparate
stakeholders need to create new knowledge. In large, complex transdisciplinary arenas, models can
become the facilitators of interdisciplinarity, integrating the different knowledge bases (Mattila, 2005).
Simulation modelling has been shown to act as a boundary object in engineering (Dodgson et al. 2007a)
helping to bridge disparate communities involved in innovation and in particular allowing disparate
groups to engage with innovation projects and contribute potential solutions to engineering problems
(Dodgson et al. 2007b).

4
Modifying Zagonel‟s (2002) use of terminology slightly we can distinguish between models as boundary
objects and as representative objects.

This classification of models can be further refined if we draw on a distinction in the science studies
literature on objects in experimental systems which makes it possible to speak more precisely about the
purposes for which different types of models are used: the distinction between epistemic and technical
objects. Epistemic objects help to create knowledge and are fluid, while technical objects are static and
seen as unproblematic tools to make knowledge available (Ewenstein and Whyte, 2009; see table 2).

Table 2: Objects in experimental systems (Ewenstein & Whyte, 2009)

Epistemic objects Technical objects

Abstract and evolutionary in-flux artefacts used in Unproblematic, static, technocratic instruments
expert work to negotiate meaning – usually used in expert work between the boundaries
political

Boundary objects and representative objects can both also be epistemic and technical objects. These two
dimensions therefore allow a classification of four types of model roles. Models which as boundary
objects facilitate communication between stakeholders with different knowledge bases can be used to
create new knowledge (as epistemic objects) by the stakeholder group or can be used to make knowledge
available across the group (as technical object). In the first type of use of a model as a boundary object the
emphasis would be on learning as a group while in the second it would be on expression of the knowledge
in a form accessible to others and on experimenting with that knowledge in the group, i.e. showing what
would happen under different scenarios. Models which are primarily used to represent a reality which is
seen as principally unproblematic can again be used in two different ways: to be explored as a micro-
world or management flight simulator in order to allow the user to learn or as a predictive tool thereby
allowing the user to draw on the knowledge embodied in the model without necessarily requiring an
understanding of the relationships within the system.

These four types are ideal types – in practice the social roles of models might not always be clearly fit
into any one of these types, but instead be a mixture of them. Different stakeholders might have different
views of the role of the model: a client might for example have at the outset a predictive tool in mind,
while the modelling process might show that what is required (or maybe in some cases achievable) would
be to learn as a group. Over time the role of a model might change: learning as a group might be
followed, by expression of knowledge and experimentation, followed by the development of a predictive
tool for other users or a micro-world as a learning environment for students to explore.

This paper introduces this framework; the remainder of the paper gives some initial directions and
indications on how this framework might be used in research about model use.

5
Table 3: A framework to classify the social roles of simulation models
Epistemic Object Technical Object

(create knowledge) (make knowledge available)

Boundary Object Learn as group Experiment and express

(facilitate communication
across boundaries)

Representative Object Explore Predict

(represent reality)

Social roles and modeling approach

Different roles of models entail different demands for the chosen modelling approach. When considering
the ideal types of models in this framework some potential implications can be sketched.

Models can support learning as a group particularly well if they are easily changeable so that suggestions
of a group model building group and experiments can be rapidly implemented and interactively explored.
Such models would typically be simple and visually accessible to the stakeholder group which might have
only limited experiences with simulation modelling or understanding of the mathematical underpinning of
models. Frequently in such modelling projects, insights into relationships between variables or parts of
the system might be more of a focus than precision of the modelling output. Models used primarily as a
tool for prediction might in contrast be more fixed, detailed and precision focused. While the visual
interfaces might still be important depending on the context of their use the emphasis might now be more
on the visual attractiveness of the output rather than on the degree to which it makes an understanding of
the relationships between elements of the system accessible (see figure 1).

The requirements for the other two types of roles in our framework will fall between these extremes.
Models used to experiment with and express knowledge off a model used to explore a system should
make insights into relationships easily accessible but need not to be easily changeable in their model
structure.

While specific models would not be expected to correspond completely to these ideal types and while the
exact model requirements will be context specific, nevertheless this classification in ideal types is
informative and should be recognised.

6
Figure 1: Model characteristics and social role

The literature on simulation more widely contains a increasingly lively debate on the characteristics of
different modelling approaches, in particular System Dynamics (SD) and Discrete Event Simulation
(DES) have been frequently compared. Systems Dynamics and Discrete-Events Simulation can been seen
to represent the two ends of a spectrum in their emphasis and explanatory power, though both may be
applied to the same situations. There has been discussion and comparison of the methods in the literature
since around the mid 90s, most notable early discussions being Sweetser (1999), Lane (2000), and
Brailsford & Hilton (2001). These themes become more fully explored by Morecroft &Robinson (2005,
2006) and Tako & Robinson (2009a, 2009b), as well as by modellers subsequently looking at strategies
for combining Systems Dynamics and Discrete Event Simulation in hybrid models (for a summary of this
literature see table 4).

Several papers offer summaries of the differences beyond the purely technical distinctions between the
two approaches, which we class into four areas:
 characteristics of the problem / decision under consideration,
 data requirements and the development process
 type of understanding derived and
 model output and their usability by clients (often based on visual representation).

7
Table 4: Some differences between SD and DES modelling based on the model comparison literature.

Areas Characteristic typical SD use typical DES use

problem / decision level strategic decisions in systemic operational decisions
decision type and population levels

perspective systemic overviews to operations level where events

population level where impact one another and variations
individual variation is statistical of individuals cumulate or interact
subsumed
data base data qualitative to identify system model build from individual
requirements sources behaviour, find feedback loops components, putting together
and then supported by data to entities
development complete stock levels and flow
process rates

uncertainty deterministic runs based on explicit randomness in parameters

and provided parameters, feedback for each modelled activity and
randomness loops and delays event
type of key technical systemic interactions and impact of randomness/variation
understanding learning feedback effects and potential bottlenecks under
derived runs

scope of overall population level changes variation expected for service

learning for long-term planning efforts delivery decisions and
contingency planning
model output primary usage not optimisation, understanding playing with the models, what-ifs
and usability mode influences
by clients
representation system represented as stocks & system represented as events &
flows with explicit feedback queues with implicit feedback
effects

common user lack of individuality among probability distributions for each

concerns human entities event and entity
about entities

common user continuous, smooth curves and rearranging components

concerns stock accumulation does not completely changes interactions
about match perceptions of users
structure

While it has to be recognized that the actual domains of use of SD and DES might overlap to a wide
extent and modelers can successfully apply SD and DES tools to problems across the spectrum from

8
strategic to tactical (Tako & Robinson, 2006), the characterizations of SD and DES literature nevertheless
allows formulating a hypothesis about the “natural domains” of both modeling approaches in Figure 2:
the top right corner could be suggested as the natural domain of System Dynamics while the bottom right
seems more the home of Discrete Event Simulation.

Figure 2: Model characteristics and social role

Empirical work is required to analyse whether this suggested understanding of natural domains of
different modelling approaches corresponds to the actual use of these two approaches and to successful
outcomes. Clearly such work will also have to consider the differences between stakeholders - in terms of
knowledge domains, language used, incentives and social ties - as well as the problem characteristics and
the system (e.g. importance of randomness and feedback, relevant level of aggregation; operational vs.
strategic focus) together with the goal of planning process or of the modelling engagement.

Illustrative case study of two modelling projects of the same system

We are currently conducting case studies of consultancy engagements using simulation modelling in
healthcare planning. In this work we have observed modelling workshops and interviewed of clients,
modelling consultants, expert group members and other stakeholders.
9
As part of this work, we compare two case studies of SD projects tackling the same problem issue, one
from a local perspective and the other from a national perspective. In both projects a similar group model
building approach was used. However, the composition of the expert groups involved in the model
building process was different. The local project aimed to inform commissioning of healthcare services
for that particular local jurisdiction, while the national project aimed to develop tool for local
commissioners nationwide .Table 5 compares the two case studies and includes some illustrative quotes
highlighting the role of the differences in goals and group compostion which influenced the roles of the
models in both cases.

Table 5: Comparison of two case studies

Case 1: producing a tool for others – national Case 2: learning as a group – local project
project

Goal: "influence policy" and "make a difference", Data clashed with perception of participants –
"reflect the work we had done in this", "Provide a learning about wider system, finding about the
tool for local authorities to make a robust business performance of the solutions, attention directed by
case." modeller towards solutions

Group composition: "it is a reasonably small field" Iterative process where the boundaries of the
and so "we all knew each other" model are negotiated with participants depending
on changing perceptions of the system
Some said the model building made them "look at
everything", while others don‟t see a difference "Three key points to help the participants use the
from other policy discussion events model constructively: a well defined issue, people
who have the power to make changes to take part
Welcome broad participation in group (different in the process and the simplest model to address
disciplines) the issue."
Model is ("looking at full spectrum of " ..... it is a group learning process – if you present
interventions") – model can communicate this to it cold through a model without the learning
others process it is very difficult to own the results ......"
Model can also distract because some find it "the model works best with those (participants)
difficult to understand who have a whole systems view and can articulate
what they see ...."

While case 1 (the national project) had the goal of producing a tool for others to use, case 2 (the local
project) gave considerable weight to the learning of the stakeholder group. The two groups also differed
in regards to the composition of the stakeholder groups. In case 1 the participant were nationally
recognised experts who knew each other well from meeting at conferences and other events. Even though
10
they came from different disciplines, they had an understanding of the other participants‟ knowledge and
had developed a shared language from their repeated previous encounters. Case 2 was a local group of
professionals involved in implementing the policy on a local level. Maybe surprisingly, the ties between
the stakeholders were looser and more work had to be done to create a shared understanding and a shared
language. In case 2 the modelling workshops had a function of creating the shared understanding of the
problem, while in case 1 the focus was more on codifying the shared understanding which was already
there.

The differences in understanding of the goal of the project and in the difference of the composition of the
stakeholder group determined the different social roles of the model in both cases. In case 1, the model
did not serve as a boundary object but was conceptualized as a decision making tool for others. In case 2
the model was observed to be serving more of a boundary object and an epistemic object role than in case
1: it had some role in bridging the difference of the knowledge bases of the stakeholders and supporting
the collaborative production of new knowledge.

Summary and conclusions

In order to classify the social roles of simulation models, two dimensions can be distinguished: models
can be boundary objects or representative objects and they can be epistemic or technical objects. The 2x2
matrix of these two dimensions allows a classification of four types of model roles. Models can be
ascribed different roles over time and different roles by different stakeholders involved in their
development and use potentially leading to misunderstanding and conflicts.

The suggested classification framework can be applied to a variety of problems around the use of models
including the discussion of the natural positioning of System Dynamics models and Discrete Event
Simulation models in a group model building context. The use of the framework for our illustrative case
studies of modelling project has highlighted how the nature of the group composition and the goals also
are key in determining the social role of models.

We are currently conducting a larger empirical study which aims to analyze the social role of simulation
models in simulation projects using both SD and DES. Through the use of our framework in this study we
hope to understand factors influencing the social roles and inform work identifying success factors for
modeling projects aiming to support diverse aims: to learn in a group, to express knowledge and
experiment with it, to explore systems or to develop tools for decision makers, fostering collaboration and
decision-making in diverse stakeholder groups.

11
References

Bechky BA., (2003). “Sharing Meaning across Occupational Communities: The Transformation of
Understanding on a Production”, Organization Science, 14(3): pp. 312-330.

Brailsford, S. and N. Hilton (2001). "A Comparison of Discrete Event Simulation and System Dynamics
for Modelling Healthcare Systems" Proceedings of the 26th meeting of the ORAHS Working
Group 2000, Glasgow Caledonian University: Glasgow, Scotland.

Carlile PR., (2002). “A Pragmatic View of Knowledge and Boundaries: Boundary Objects in New
Product Development”, Organization Science, 13(4), pp. 442-455.

Cetina K. K., (1997). “Sociality with objects: Social relations in postsocial knowledge societies“, Theory,
culture & society, US-based international Society for Social Studies of Science, 14(4): pp. 1-30.

Dodgson M., Gann DM., Salter A., (2007),"In Case of Fire, Please Use the Elevator": Simulation
Technology and Organization in Fire Engineering”, Organization Science, 18(5): pp. 849-864.

Dodgson, M., Gann D, Salter A (2007b). The impact of modelling and simulation technology on
engineering problem solving. Technology Analysis & Strategic Management 19(4): 471-489.

Ewenstein B., Whyte J., (2009). “Knowledge practices in design: The role of visual representations as
'Epistemic objects'”, Organization Studies, 30 (1), pp. 7-30.

Ewenstein, B. and J. K. Whyte (2007) Picture this: visual representations as 'artifacts of knowing',
Building Research and Information, Vol. 35, No. 1, pp. 81-89.

Fassin Y., (2008). “Imperfections and Shortcomings of the Stakeholder Model‟s Graphical
Representation” Journal of Business Ethics, 80(4) pp:879, Springer.

Koskinen KU., Makinen S., (2009). “Role of boundary objects in negotiations of project contracts”,
International Journal of Project Management, 27, pp:31–38.

Kristiansen K., (2004), Discourse and dialogue. Some observations and further reflections,
Archaeological Dialogues, 11(2), pp:114-132.

Lane D.C. 2000. “You just don't understand me: Modes of failure and success in the discourse between
system dynamics and discrete event simulation”. LSE OR Dept Working Paper LSEOR 00-34,
London School of Economics.

Mattila E. 2005. Interdisciplinarity “in the making”: modelling infectious diseases. Working Papers on
theNature of Evidence: Howe well do facts travel., London Scholl of Economics.

Morecroft, J. D. W. and Robinson, S. (2005). "Explaining Puzzling Dynamics: Comparing the Use of
System Dynamics and Discrete-Event Simulation" In Proceedings of the 23rd International
Conference of the System Dynamics Society: Boston.

12
Morecroft, J. and Robinson, S. (2006). Comparing Discrete-Event Simulation and System Dynamics:
Modelling a Fishery. Proceedings of the Operational Research Society Simulation Workshop
2006 (SW06) (J. Garnett, S. Brailsford, S. Robinson and S. Taylor, eds.). Operational Research
Society, Birmingham, UK, pp.137-148.

Rheinberger HJ, (1997). Toward a History of Epistemic Things: Synthesizing Proteins in the Test Tube,
US: Stanford University Press.

Rixon AJ, Burn S., (2008). “Visualisation Techniques For Facilitating Stakeholder Decision Making In
Urban Planning”, The Journal of Community Informatics, 4(2) Special Issue: E-Governance and
Community Informatics.

Sapsed, J and Salter A (2004). Postcards from the edge: local communities, global programs and
boundary objects. Organization Studies. 25: 1515-1534.

Star. S and Griesemer J. Institutional ecology, „translations‟, and boundary objects: amateurs and
professionals in Berkeley‟s museum of Vertebrate Zoology Social Studies of Science 19:387-420.

Sweetser, A. (1999). "A Comparison of System Dynamics and Discrete Event Simulation." Proceedings
of 17th International Conference of the System Dynamics Society and 5th Australian & New
Zealand Systems Conference: Wellington, New Zealand.

Tako, A.A. and Robinson, S. (2009a). Comparing Discrete-Event Simulation and System Dynamics:
Users‟ Perceptions. Journal of the Operational Research Society, 60 (3), pp. 296-312. doi:
10.1057/palgrave.jors.2602566.

Tako, A.A. and Robinson, S. (2009b). Comparing Model Development in Discrete Event Simulation and
System Dynamics. 2009 Winter Simulation Conference.

Tako, A.A. and Robinson, S. (2006). Towards an empirical comparison of DES and SD in the supply
chain context. 26th International Conference of the System Dynamics Society.

Warmington P., Daniels H., Edwards A., Leadbetter J., Martin D., Brown S. and Middleton D..
“Conceptualising Professional Learning for Multi-Agency Working and User Engagement”,
Paper presented to the British Education Research Association Annual Conference, UMIST,
Manchester, 15-18th September 2004.

Whyte J., Ewenstein B., Hales M. and Tidd J., (2008). “Visualizing Knowledge in Project-Based Work”,
Long Range Planning, 41, pp:74-92.

Zagonel A. 2002. Model conceptualization in group model building: A review of the literature exploring
the tension between representing reality and negotiating a social order. Proceedings International
Conference of the System Dynamics Society, Palermo.

13