Once the results have been interpretted, steps 1 to 4 can be repeated again to explore another analysis question with a new set of initial labels. By making multiple evaluations, the modeler can investigate several “what-if” scenarios and make informed comparisons of the various routines or design alternatives contained within the model. During this process, if the evaluation results seem contrary to reality, this may indicate one of two things. First, it could provide a unforeseen insight into the domain. For example, if the PC Product Provider does not allow Peer-to-Peer technology, it may seem that this would reduce Piracy and improve Profit, but when the Desirability of Product is considered, the results are actually conflicting. Second, nonsensical results may indicate problems with the model including missing elements, actors or links which would have a significant effect, wrong links types, or other changes. Deciding amongst the two results gets back to the believability of the model and confidence in the model’s construction and underlying assumptions. Deciding if the results are believable is also a matter of human judgment.
As with any other analytical method, as the modeler becomes familiar with the evaluation procedure, the average time and effort required to conduct future evaluations decreases and the benefit and return on investment increases. Also, the level of the effort and time required to conduct an evaluation depends on the tool used in the analysis. For example, the propagation process is done completely manually using istar_stencil in Microsoft Visio. A version of this evaluation procedure is implemented in OME3, using a slightly different set of propagation rules. The rules described here have been implemented in a depreciated version of the OpenOME tool, but are currently being re-implemented in a the new version of this tool, to be available, hopefully, before the end of 2008.
For questions regarding i* syntax please click on the "discuss" tab. This feature is available for registered users only. For registration information please see the Registration and User Rights Policy for the i* Wiki.