A new paper in the Journal of Sortition co-authored by prof. Paulina Pospieszna presents an application of the simulated annealing algorithm for selecting members of deliberative mini-publics such as citizens' assemblies. The method, developed by an international research team, prioritizes demographic accuracy in the selection process while maintaining satisfactory levels of inclusiveness and equality of chances for volunteers.
The research, based on 10 case studies from various countries, demonstrates that the simulated annealing method can achieve perfect demographic representation in most cases, as measured by specially designed Accuracy and Closeness Indexes. Notably, the method generates a high number of unique panel compositions compared to existing approaches, with 10,000 unique panels created in 8 out of 10 test cases.
The paper introduces new metrics for evaluating the quality of random selection, including the Accuracy Index and Closeness Index, which can be used to assess and communicate the precision of panel composition to the public. The method is implemented through an easy-to-use script that requires no prior programming knowledge, with parameters set through a regular Excel file.
The article was created in cooperation with: Marcin Gerwin, Piotr Szuca, Nikodem Mrożek, Grzegorz Kuriata and Alexander M. Geisler.
Full article available here.
More information about the algorithm.