���������������������������������������������������������������
er with information and perspectives from stakeholders or others with an interest in the specific problem . This requires tools that can help the policy analyst to extract , organize , and assemble knowledge , facilitate participation and communication , and develop a representation of a shared understanding .
The first task in knowledge synthesis is to elicit the knowledge that is to be synthesized . While a model does not directly extract knowledge in the same way as does administrative data analysis or conducting a survey , the modeler explicitly asks the policy analyst and other subject matter experts for key information about the target system . This occurs primarily in the Design phase , where competing understandings and different perspectives are drawn out and combined into a coherent set of features and relationships to be included in the model .
Additional refinement and understanding occurs in later phases as the rigor of Build and Test highlights remaining inconsistencies or incomplete information . Working with group discussion , rather than separate individual contributions , can be particularly valuable as the interaction identifies priorities , promotes clarity in the problem definition and system description and supports open discussion of conflict ( Cockerill et al ., 2009 ; Shackley , 1997 ). As the model is developed , it captures and organizes the knowledge about the features and relationships in a system . Such organization summarizes and makes accessible a shared understanding of the target system , and supports a range of policy relevant activities . For example , the organized knowledge provides a focus for ongoing design discussions , eliciting further knowledge as stakeholders respond particularly to those elements they believe to be incorrect in some way .
In addition , the model as a communication tool can provide additional knowledge synthesis functionality during the Use phase , transferring and consolidating understandings through the experience of using the model and discussion of analysis results . That is , participant users can understand and explore the implications of the relationships they have identified .
������������������������
Regardless of how much knowledge
can be amassed and understood in the available time , there will still be important unknowns that could impact on whether potential actions to address complex societal problems help or hinder . Models can expose unknowns and reduce or estimate their impact , but can also introduce new unknowns . Ignorance theorists have identified various types of unknowns ( Gross , ����� ���� ), of which five are important for modeling of complex societal problems : inconsistency or confusion , inaccuracy , inherent uncertainty , absence ( or gaps in knowledge ), and irrelevance .
As for knowledge synthesis , the discipline of the modeling process provides some of the functionality to manage unknowns , particularly during the Design and Build phases . The focus on a model purpose and the objective of simplifying the model to the extent possible sharpens discussions between participants about whether a proposed feature or relationship is required , thereby consciously ignoring irrelevant information as out of scope . Such decisions may , of course , introduce error by incorrectly excluding some factors with greater influence than those included . However , if done well , excluding lower priority knowledge is a pragmatic trade off that accepts incompleteness to gain simplicity and clarity .
The rigorous description of the features and relationships eventually included
70