Information Technology Reference
In-Depth Information
agents interact, i.e., meet randomly, to learn about the best performing strategy by
comparing the utility of its own behaviour with that of another agent. The interaction
with the physical environment results in receiving utility as a consequence of the
chosen behaviour.
The Challenge of Matching Theory and Case.
The question how well suited the
ostracism model is for the Bali case guides the iterative process of model design and
model adaptation. To identify whether the ostracism model matches the irrigation
context in Bali, the concepts of the ostracism model are placed in relation to what we
know from irrigation context of Bali [6], see table 2. Without going into detail about
each concept, we can identify 'easy' mappings' such as, resource is water and payoff
is the rice harvest, however most concepts do not exist in an one-to-one correspon-
dence with the real case. These notions are either more rich in the Bali context (light
grey cells) or the data is not available (dark grey cells).
We see that the resource management context from the ostracism model matches
the irrigation dilemma in Bali well in terms of: resource (dynamics), utility inter-
linked with the resource and the presents of social factors that could reflect the varia-
tion in adaptive capacity. More specifically, these social factors, relate to variables,
e.g., sanctioning, norms, that are considered important to establish self organisation
[17]. However, the Bali context also indicates aspects that are not addressed by the
ostracism model. For instance, in simplistic representation of the physical environ-
ment: pests dynamics are not included, however play a crucial role in the Bali irriga-
tion dilemma. Coupling the model to the existing models of Bali ecology is therefor
the intended solution. Concerning the social environment (topology) is minimally
existing on the micro-level, there could be good reasons to introduce a spatially de-
pendent structure that affects how the agents meet. Another example would be (hete-
rogeneous) agent attributes in which the role of caste could play a role in the way
agents from different caste interact with each other.
When comparing these missing elements with other models of cooperation [14-16],
we can identify some mechanisms that target some of the missing components ad-
dressed above to explain cooperation
5
. For instance, spatial explanations, such as
network reciprocity, graph selection, or set selection, describe the influence of the
network topology of an agent on the cooperation. Other explanations focus more on
explaining cooperation based on the group that one belongs to, e.g., green beard,
group selection or kin selection, which could be an option for representing the role of
being heterogeneously part of a caste. Overall, most models of cooperation focus on
one or a few mechanisms/drivers of cooperation. Probably the ABM model will result
in a merger of theories/mechanisms. For now we consider the ostracism model good
enough to continue, it is up to the next model specification phase to define what
seems to be a good fit of model & context.
5
Theories that describe the evolution of cooperation focus on identifying mechanisms.
