Information Technology Reference
In-Depth Information
always referred to more than one category in figure 2, agreement in one category is
quite likely and thus not a reasonable basis for cooperation. The lowest possible re-
quirement thus would be an agreement in at least two of the goal categories.
After the selection of possible cooperation partners, the agents send cooperation
queries to their possible partners in distinct time intervals that are replied positively or
negatively depending on their degree of agreement concerning the development goals.
Agents with higher shares of positive replies are selected for cooperation. The final
step is to have the agents design a project together. It is assumed that agents intersect
their goals and subsequently select a project category that matches that intersection
best. At the end of this process, the agents may evaluate their cooperation e.g. by the
number of projects that they designed with a particular partner or by the share of their
own development goals that the designed projects achieved.
5
Discussion
The objective of this paper was to investigate and model a political process that in-
volves a number of stakeholders with their respective viewpoints, and which relies
fundamentally on the communication and collaboration among them (a detailed de-
scription of the results may be found in [10]). The achieved simulation results should
be validated, i.e. be compared to the empirical evidence from the social phenomena
investigated [11].
It became evident that the text based representations of actions of the stakeholders
can in principle be transferred into rules of an agent based simulation model, in part
requiring substantial empirical effort. This may be illustrated by means of some of the
model elements: Empirical evidence suggested that political stakeholders had closer
relations with each other than with stakeholders from the other groups, explaining this
fact by their more frequent meetings. This interrelation could be exactly represented
in the simulation. Also, the simulated discussions in the regional assembly showed a
high consistency with the empirical data of their real world counterparts.
Discrepancies were detected concerning the voting behavior: While according to
the observation of the assembly meetings the stakeholders decided the projects pre-
dominantly concordantly positive, the interviews revealed that stakeholders were
themselves skeptical about this procedure, because it resulted solely from a relatively
low usage of the available budget in the first half of the period. They hence expected a
need to prioritize and thus reject funding for at least some of the projects in the near
future. This served as a justification to direct the simulation model towards addressing
this prospective challenge for the stakeholders by formulating a dynamic decision
process depending on the course of the discussions. This setup also served as a base
model to experiment with the variability of the decisions in a number of different
discussion settings.
However, some methodological challenges remain. While role playing games can
monitor the overt behavior of stakeholders in specific situations and thus serve as a
data source for a simulation, these situations will always be artificial in their setting.
When using interviews, one has to be aware that the resulting narratives are nothing
