Page 257 -

256 O. Barreteau et al.

Quality of the simulation model is understood here rather classically with the
following indicators:

– Realism: is the simulation model able to tackle key features of the social
complexity it aims to represent?
– Efficiency: is the simulation model representing its target system with a minimum
of assumptions and minimal simulation run times?
Quality of the representation according to its use is another classical indicator of
a simulation model’s quality. It is specifically tackled in the following subsection.

12.2.1.1 Taking Social Diversity and Capacity to Evolve into Account

One of the key features to be taken into account when representing a social
system is to deal with its diversity. This diversity is related not only to individual
characteristics but also to viewpoints, expectations towards the system and positions
in the decision-making processes. Dealing with diversity in simulation of social
complexity involves embracing it as well as profiting by its existence.
Classically, dealing with diversity is a process of aggregation or selection. Aggre-
gation consists of the identification of classes of individuals and representatives for
them. Selection consists of choosing a few cases with all of their characteristics. This
may lead to very simple simulation models with a generic diversity. Aggregation is
rather greedy on data and modelling time and is still dependent on the viewpoint of
the observers who provide the information leading to the categorisation. Selection
is weak to cope with relations among various sources of diversity.
Involvement of stakeholders in the modelling process allows them to bring
their own diversity. Concerns over representation are then transferred onto the
constitution of the sample of participants. Fischer and colleagues have shown
through development of situations to support creativity in various fields, such as art,
open source development and urban planning, that diversity, as well as complexity,
is important to enhance creativity (Fischer et al. 2005). This creativity is expected
to pave the way for surprises in the simulation model.
Involvement of stakeholders in the modelling process is a way to externalise part
of this diversity outside the model towards a group of stakeholders. The issue is then
to work on the relation between the model and a number of stakeholders to allow a
transfer of knowledge and ideas.
Social systems are open and evolving. Their definition depends on the viewpoint
of the analyst. As far as simulation is concerned, this means depending on
the viewpoint of the model designer(s). This choice means framing: cutting a
number of links around the boundaries of the system studied, as well as around
the interpretation which might occur based on the simulation outcomes (Dewulf
et al. 2006). Firstly, participation provides the opportunity to consider problem
boundaries which would be plurally defined, increasing the potential coherence of
the model. However, it is still an operation of cutting links out of the real-world
situation, even though these chosen cuttings are more grounded and discussed.