Tensions Between Scientific Programming and the Scientific Method

Abstract

This thesis evaluates eleven scientific programmers against six best practices. The six best practices extracted from literature are: 1) the presence of a Distinct Design Phase, 2) Documentation, 3) the Use of Existing, Trustworthy Code, 4) the Use of Formal Version Control, 5) a Testing procedure, and 6) the Public Release of Code. A survey and interview were conducted on eleven participants to produce a case study on each participant’s purpose of their code, their self identified process difficulties, their tool use, their self identified areas for growth, and their perceptions of the importance of programming to their discipline. An evaluation of the extent to which the six best practices were followed is then presented. To conclude, three recommendations on how to increase adherence to best practices are presented: education about existing processes and tools, the adaption of existing processes and tools, and incentivizing adherence to best practices.