External Experiments – A Workable Paradigm for Collaboration Between Industry and Academia

Abstract

Results of empirical investigations are key input for industrial software process improvement activities. The relevance of an investigation for a given environment, however, depends on the similarities or dissimilarities of the investigation environment compared to the industrial environment which is interested in using a new technology. The optimal (or most relevant) results may be gained when performing the experiment in the industrial environment itself. Unfortunately, this takes place only rarely as experiments in industrial environments are either expensive (when new and old technology are used in parallel) or risky (when only the new technology is used).To overcome this dilemma, we introduced the concept of external experiments, i.e. experiments which are conducted in an environment different from the industrial target environment and which tries to simulate the characteristics of the industrial environment as best as possible. This paper describes this concept in detail, presents a process for identification, conduction and exploitation of such experiments and shows some results from past investigations which used this concept.