The Garbage Can Model: A Study in (Non)Reproducible Research

The classic paper 'A Garbage Can Model of Organizational Choice' by Cohen, March and Olsen (1972) both heralded a significant expansion of the study of organizational decision making to non-industrial settings, in particular universities, and served as a very early example of reproducible computational research, incorporating a Fortran 66 program in its appendix to permit others to reproduce their results and run further examples. In this work my extensive attempts to perfectly reproduce the original results show the inherent challenge of reproducing computational research in the presence of ever-changing computing platforms and technology. Indeed, exact values could not be reproduced in this study, nor any other published study, because of hypersensitivity. Despite this, additional statistics allowed by modern computer capabilities almost completely agree with the qualitative observations and conclusions in the original work. Finally, in light of the need for high precision, it will be worthwhile to retest and reevaluate later studies of the internal dynamics of the model that faulted the code for behavior at odds with the theory.