Do One Thing And Do It Well

The phrase "Do one thing and do it well" is an admonition from C language expert Brian Kernighan about how to write good functions (and more recently from Arthur Riel's heuristic that each class should manage one key abstraction [BibRef-Riel1996], in addition to what are doubtless many other parallels). Analogous advice is starting to appear for classes in object-oriented systems. And the same might apply to the people who write those classes.

QPW is organized along lines of domain specialization. Domains important to QPW are dependency registration software, human interfaces, databases, and a few others. An individual was identified for each of those domains.
In their domain, each individual does what they are good at. They excel at bringing their domain expertise to the table in architecture meetings. They know what the right abstractions are. They know how to implement those abstractions. They bring C++ or DOS or Windows proficiency to the project, or quickly develop it (through analogy to related domain experience).

Equally important is what these individuals are not good at, and that they are not expected to take responsibility for domains not related to their specialization. Instead of working inthese domains, hey work with these domains. One good example is documentation. Developers are supported by a documentation organization that develops internal and external documentation. The time spent by developers in conveying this information to the documentation organization is far less than it would take for them to commit it to writing, put it into an acceptable format, and have the work edited for linguistic elegance.

By contrast, we knew that most of our AT&T developers wrote most of their own memos. It's not clear whether this owes to our history, our organizational boundaries, the nature of our business, or to reward mechanisms. Nevertheless, a deeply rooted cultural behavior at AT&T is that engineers draft their own memos. Developers spend much time (roughly 13%) of total development time creating and refining memos in AT&T. In Borland, that job is deferred to people who are expert at it. (AT&T information from process researchers Larry Votta and Nancy Staudenmayer (MIT), 1993.)