In the same issue of Transactions on Software Engineering that has the pair programming study I just mentioned, Hannay et al. [1] –again from the very prolific Simula Research Lab–, give us a glimpse of the immaturity of the software engineering research field by reporting on a survey about the amounts and kinds of theories used in software engineering experiments. While other disciplines take theory for granted to guide their studies (why would a physicist look for ether anymore when it isn’t a part of any current theory?), most software engineering researchers haven’t read the memo yet:
“Of the 103 articles, 24 use a total of 40 theories in various ways to explain the cause-effect relationship(s) under investigation. (…) Several articles comment explicitly on the lack of relevant theory. (…) Most of the articles (that) use theory (use it) to justify or motivate experimental research questions. However, we found no evidence of theory-driven research, in the sense of empirically based theories that encompass and define the research questions of empirical software engineering.”
To use Kuhn‘s terminology, we’re still in the prescientific stage. We will hopefully get out of it soon (that is, in a decade or three).
—
[1] Jo E. Hannay, Dag I.K. Sjoberg, and Tore Dyba. “A Systematic Review of Theory Use in Software Engineering Experiments”. IEEE Transactions on Software Engineering, 33:2, Feb. 2007.
Catenary 
A decade or 3 lol !
I have the same feelings, it’s still a young industry and a baby science.
I bet on code reviews instead of pair programming, but in my theory, this part is less important than the functional spec and the design description.
…and the benefits of all of these techniques haven’t been properly assessed. Even more troubling, there are teams that do not use any of them and still develop software successfully!
Pingback: Should we care about evidence-based software engineering? « Semantic Werks