Building computational social science models from crowd insight

Accurate models of social processes are invaluable tools for understanding, exploring, explaining, and predicting phenomena that impact decisions about policies and courses of action. However, computational social science (CSS) models today take months to create and can only be generated by a fraction of modeling experts. Building models in this way is an “Ivory Tower” approach where highly specialized modelers spend an impractical amount of time building a model, creating “heavy” models that are slow to develop, hard to maintain, and may not reflect rapidly changing factors “on the ground.” CSS models must be developed more easily and in a more timely fashion to support the dynamicity of the social space. This work describes a vision and proof of concept in which observations of average people are harnessed by a system capable of generating adaptive models that leverage individual and collective insights and experiences. Rather than relying on few experts with similar experiences, our approach relies on a larger crowd of individuals with diverse knowledge. This approach augments the Ivory Tower with the wisdom of crowds to enable speedier development and seamless update and maintenance of models.