Browse Books

Reviewer: Bonnie A. Nardi

Winograd presents an engaging ramble through software design. Mitch Kapor, Paul Saffo, John Seely Brown, Donald Scho¨n, and others put in appearances. The book evolved from a workshop held at Stanford in 1992. Each contributor to the book is thoughtful, well-spoken, and worth listening to. The thesis of the book is that a new discipline of software design, as distinct from software engineering, is needed. There is a welcome emphasis on human-computer interaction as a major part of software design. The book gets off to a weak start, with a reprint of Mitch Kapor's “Software Design Manifesto” [1]. While this short piece is of historical interest, it has dated quickly. “No one is speaking for the poor user,” states Kapor in what now sounds like a somewhat condescending voice. Many of the other chapters in the book belie this statement. Kapor asserts that designers should learn languages like Pascal and Forth. Pascal and Forth__?__ This contribution might better have been moved to another part of the book, at least if you agree with the old vaudeville adage that you should always open and close with your best acts. The book ends with a wonderful chapter on human-centered design and participatory design by Sarah Kuhn. Kuhn argues that bad design is expensive and that, by not listening closely to users' needs and taking into account the context of their work, expensive mistakes will continue to be made. She offers a number of case studies and points to the success Intuit has had by “observing the customer's context of use, and anticipating new features even before customers themselves have identified new needs.” Laura DeYoung's fine chapter goes into detail on just how Intuit gathers customer input and how seriously it is taken, all the way up to the highest levels of management. I particularly liked the chapter by Denning and Dargan, in which they analyzed several successful systems (Quicken, MeetingMaker, Topic, and the Macintosh user interface) to arrive at the following recommendations for good design: choose a domain in which many people are working and where there are constant breakdowns (for example, in the area of handling personal finance); study the actions, especially the repetitive actions, that people take; define software routines that imitate familiar patterns; and deploy prototypes early and often. Of course, the immense creativity that goes into a design such as the early Macintosh interface is not captured in such a procedure, but the general idea is sound. If more companies and researchers followed this advice, there would be a huge improvement in the products consumers are offered. This book has a carefully selected bibliography and is a valuable resource for that reason alone. At the end of each chapter is a short list of “Suggested Readings.” I found the selections to be right on the mark. At the end of the book, a bibliography helpfully compiles all the references. The book is packaged in a magazine-like format with essays, interviews, and profiles of products. While this format <__?__Pub Caret>works well overall, it was surprising to find such inadequate design (in a book on design) for the profiles. Black type is set against a gloomy gray background, making the profiles unattractive and hard to read. The profiles are useful, however, and include Kid Pix, Mosaic, the Star, HyperCard, Director, Visual Basic, spreadsheets, Microsoft Bob, and Quicken (Quicken is the product star of the book). I would have liked a little more analysis and sharper-edged evaluation here. The descriptions are informative but bland. The Microsoft Bob interface, for example, has not been a market success, but it is presented as an example of a simple interface. This book will be a good resource for students trying to get a grip on the whole business of software design and especially the field of human-computer interaction. While the chapters by Norman, Scho¨n, Saffo, and a few of the others are not their best work, they will introduce new scholars and practitioners to these important thinkers. The suggested readings provide the needed follow-up. Overall, this work will admirably serve its intended goal of bringing a higher level of professional recognition to software design.