It may also be that the theorists view their research domain as the whole unit rather than one component of the discipline.
What is sometimes missing is the big picture, i.e.,
what is the collection of components and how do they fit together?
What are the various program abstraction methods and when is each appropriate?
For what applications are they not effective?
Under what conditions are they most effective?
What is the relationsidp between processes and product?
What is the effect of a particular technique on product reliability, given an environment of expert programmers in a new domain, with tight schedule constraints, etc.