For the second time in two days, I came across the terms emic and etic, so it's time to have a bit more of a look into what they mean. This was prompted by a paper on JASSS: Agents in Living Color: Towards Emic Agent-Based Models, which I found following up on the links between wp:Prospect Theory and ABM. (Not while looking for ways to model crack dealers.) wp:Emic and etic refer respectively to insider and outsider views on a situation, springing from the linguistics phonemic/phonetic distinction. Emic accounts are in terms which can be understood by the people being modelled - they are from the perspective of insiders. Etic accounts are outsider views, and strive towards some form of objectivity or at least generality (given the problems with any kind of objective account).
Why is this relevant to ABM?
Firstly, as argued in the linked paper, one has to really understand how the modellees work and behave in their world to create an accurate model of behaviour, with the correct processes captured.
Secondly, there is the issue of communication, in particular with stakeholders. Often model results are presented as figures and graphs, a highly scientific, etic viewpoint, which may be entirely opaque to those involved. This relates mildly to what James Millington was finding with his Spanish farmers - it was very hard to ask them general questions about the landscape, as they always wanted to frame things in terms of their farm.
However, agent based models have a real potential for relating with stakeholders, as it is their decisions which are being modelled, so the process can be created in their own terms to a large extent. Conrad Rider has being doing some work in this area, using BDI agent models to look at land use change, but getting them to output narratives along the way. If the model can output sufficiently *emic* narratives, there is then a real chance to show the output to stakeholders in a way that lets them understand the processes which have been modelled, and use it as a form of verification that the important issues have been addressed.