Is there a future for empirical software engineering?

This talk provides a personal perspective on the evolution of empirical software engineering. The evolution has included the focus of the studies, e.g., from attempting to demonstrate the effectiveness of various processes to the building of knowledge around a domain or a set of processes. The study designs have changed from strictly quantitative studies to all forms of controlled experiments, quasi-experiments, preexperimental designs, case studies, field studies, and focus group activities. There has been recognition of the importance and influence of context variables and the domain in interpreting the results of studies. The issue of meta-analysis has become more important and is still an elusive goal. A community of empiricists has evolved that allows for the replication of studies leading to the identification of context variables. But there still does not exist a community consensus on the right ways to build bodies of knowledge or a "blessed" set of approaches. We argue over what is appropriate evidence when we review papers. We struggle with the concept of experimental replication. We do not know how to break an empirical study into significant small enough chunks for publication in the various standard forums. We have not convinced the software engineering community that empirical study is a necessary or even valuable research paradigm for the field. What other problems that remain? How can they be solved? What will the future hold for empirical study? This talk will address these issues and propose a possible path towards a mature empirical software engineering discipline.