Best known for his seminal work on software engineering and project management, Frederick Brooks expands here to ubiquitous design. Having led the development of the IBM 360 computer and its operating system, I suspect that many do not know about his pioneering work in virtual reality. Brooks draws upon both domains, and more pedestrian endeavors such as the renovation of his own home, to seek universal meaning in physical design and the design of software.
Like The mythical man-month, this book is organized into a series of independent essays--and the range is dizzying. The essays cover various aspects of design, open-source software, collaboration, telecollaboration, rationalism versus empiricism, aesthetics, art history, home construction, management theory, and biographies (of John Cocke, Ralph Gomory, Seymour Cray, and others). Each chapter is introduced by a visual representation of the theme: a painting of Jefferson, floor plans for the author’s beach house, a photo of Michelangelo’s David, a cartoon of the office of the future, and even sheet music. Annotated notes and references incorporate a unique, highly informative feature--extensive feedback from reviewers of the manuscript. The author states that “a science of design is an impossibility.” Therefore, he does not aim for a comprehensive set of design rules. Instead, he offers “a few opinionated essays” and observations. There are some real gems.
Brooks cautions us about ridiculous requirements. For example, the Comanche, a “self-ferrying helicopter,” was supposed to be able to fly itself over the Atlantic. This feature--even though it would be used very infrequently--dramatically affected the design constraints. In “Rationalism versus Empiricism,” we learn the difference between rationalism (a perfect design process can be used to reason a perfect design) and empiricism (a design can be perfected over time through an iterative process). Brooks states that he is clearly an empiricist, though, with respect to rationalism, he is a fan of cleanroom design and has guarded appreciation for formal methods. In the chapter on bad designs, he gives a fascinating history of a “bad” computer language--IBM’s Job Control Language, which was used as a scripting language (originally to be programmed on punched cards) for the OS/360 operating system.
Brooks likes spirals. There is the Boehm spiral model, which is mentioned in several of the essays, a picture of a spiraling bridge, and even a spiral staircase in the author’s beach house. He takes up issues of budgets and design constraints, contending that constraints may also be your friend because they shrink the design space. Experimental evidence given for this claim is compelling: give a student an assignment to write an essay about “anything” and she will procrastinate and complain; if you tell the student to write about “Julius Caesar,” she will get to work right way. Another important observation is that the designer is “better wrong than vague.” Wrong explicit assumptions will at least be questioned, whereas vague ones will not. Put more exquisitely, “An articulated guess beats an unspoken assumption.”
His discussion on collaboration argues that design activity has evolved from solo to group-centered over the last 100 years. In this regard, he argues that competition (both internal and external) can be as valuable a design-forcing function as collaboration. His essays on “mind to machine design” and “machine to mind design” reveal his long interest in virtual reality.
This book is many things, but mostly it is a look into the mind of one of the greatest designers of the automated age. If I were to make a movie based on this book, it would mimic Being John Malkovich, where the protagonist gets into the head of a famous American actor. Being Frederick Brooks would be a frenetic movie--thrilling at times, distracting at times, and puzzling at times--but there would be so much to be learned. I suggest that every computer scientist and software engineer buy this book.
