Summary

We are a group of computer scientists at the University of Karlsruhe believing that Informatics (or Computer Science), because of methodological problems or misconduct, falls short of what it could achieve. To prove this impression, we conducted a literature survey in 1994 and found, for instance, that as much as 40 percent of the journal contributions that would require experimental evaluation contained none whatsoever and that computer science journals were worse in this respect than two journals from other fields we used for comparison (Optical Engineering, Neural Computation).

Our work is driven by several goals:

1. We want to contribute to a solid methodological basis for empirical software engineering and other parts of empirical Informatics, in particular with respect to controlled experiments and benchmarking-type studies.
2. We aim to test unvalidated hypotheses that are used in may areas of software engineering (sometimes outspoken, sometimes implicit).
3. We try to build explanatory or predictive models that identify, quantify and relate the influence factors that underly software engineering phenomena. These models will sometimes indicate specific areas for improvement and hence give impulses towards the developments of better methods or tools. We will then also investigate these.
4. We are also interested in performing solid empirical evaluations of the performance and usefulness of tools or software systems (from all areas of Informatics) that originated from outside the core empirical research group.