This book reports on error theory work in China. In the context of the book, an error is an outcome that is not in accord with a set of rules that specify the correctness of the system. Normally in a review of a book like this I would structure the review by summarizing the chapters in the order in which they appear. However, in the case of this book, this approach would not be in the reader’s best interest. The problem with the book is that a great deal of the material takes the form of abstract definitions aimed at creating the authors’ theory of error. This is a critical problem for readers because the definition of “error set” is not completely elucidated.
The authors spend a substantial portion of the book discussing formal ways of composing error sets. This material is quite dense and not a little difficult to follow. In fact, the typography of the book makes it even harder to follow by failing to make adequate use of white space in the layout of the definitions--the definition of an error set is a particular offender here, as the components of the definition are not laid out in a way that helps the reader understand them. To compound the problem, few concrete examples are given in the course of the abstract material. I had expected that this formal material would lead to a methodology for error analysis and detection, but this is not the case.
So, how should readers approach this book? It is probably best to start with Section 6.5, which describes an error system that could be used by referees in Nanquan--this is a form of Chinese martial arts in which the practitioner has to conform to a set of rules regarding the placement of hands, feet, arms, and legs. Chapter 7 gives an in-depth example using a water treatment facility in East Guangzhou. These examples will require some backtracking to find definitions, but the reader will gain an understanding of the authors’ approach.
An error set thus consists of a set of rules for determining correctness, domains that restrict the space and time of an application, for what may be acceptable at one time or in one location may fail in another. An important component of an error set is the notion of transformation that allows for the analysis of errors. Unfortunately, the provided definition of transformation is somewhat opaque. Only in the examples in the final chapters does the reader get a better sense of the idea. One specific example of a transformation is scaling.
Chapter 5--perhaps the most rewarding chapter in terms of the authors’ theory of errors--further describes the structure of error systems, where the system can be serial (errors in one component propagate through the system), parallel (errors in one component can be elided), and hierarchical. This is a useful way of looking at errors in systems and should appear earlier in the book.
The book is marred by a number of typographical errors. Usually the reader can get the correct sense even in those cases where a “not” is missing. As remarked above, the typography renders large sections of the book very hard to read--a better layout would make the book thicker, but easier to follow. Perhaps the best use of the book would be to give it to a graduate student and have her or him create a readers’ guide--a nice master’s thesis. Many important ideas related to structuring error sets and their analyses are implicit throughout the book, but require further elaboration.
