β Scribed by H.T. Smith
No coin nor oath required. For personal study only.
The appearance of any work which has the appellation State of the Art excites a more-thanaverage degree of interest and comment. In this case the label is attached to a subject matter that is very under-represented in published from and therefore doubly welcome. The lnfotech Report on Man/Computer Communication consists of a two volume set. The first volume is entitled An Analysis of the State of the Art, and the second represents as set of specially commissioned papers, approximately half of which were given at a 1978 Infotech seminar in Amsterdam and the rest being obtained subsequently to round out the subject. (A list of the printed papers is given in an appendix to this review.)
As is usual with Infotech Reports, the two volumes do not make up two halves of one story. Instead the first volume consists of a review and analysis of the Man/Computer Communication area which draws heavily from the contents of 22 papers printed in the second volume. The review and analysis is intended to give the reader a global view of the subject and, at the same time, provide pointers towards the more detailed material in the second volume. It consists of an editorial framework into which are fitted quotations extracted from the printed papers and the general research literature. The quotations "are selected to give a balanced and comprehensive view of the problem and solutions". (Although not in distributed data processing as printed in the editorial notes!) Different typefaces are used to distinguish clearly between editorial comment, selections from contributory papers, and background sources. Personally, I dislike the overall style that results from the use of this system. Sometimes it resembles a fragmented series of epithets which hinders detailed consideration of a concept. Others may not agree. Certainly, when it works well, the reader is given a series of snapshot views of a particular topic by different authorities. However, the success of this strategy depends not only on the editorial framework but upon the contributory papers being representative of the area (and the presence of "quotable" quotes). For example, over 80% of the citations in the review come directly from the contributors papers or the conference proceedings. This causes difficulties when topics are not adequately represented by the contributions.
Any evaluation of the report must explicitly consider whether it provides a definitive coverage of the Man-Computer Communication (MCC) field. It should be obvious that the production of an acceptable definition of MCC is a difficult task. After all, it is relatively easy to produce a symposium on "Database Design" or "Structured Programming" but quite another thing to decide what characterizes the field of MCC. Given the large range of topics from which to choose, stringent decisions will have to be made about what to put in and what to leave out. The resulting balance will almost inevitably be idiosyncratic to the Editor and perhaps contentious to others.
At first sight the coverage provided by the editorial framework seems quite good. However, there is one important exception. The Editor cites two main areas of deliberate omission, specific applications (e.g. business and education) and programming language experiments (including studies of human programming). While the former is reasonable, I believe the latter omission is regrettable. At the present time, and for the forseeable future, one has to resort to some form of programming to make computers perform a desired function. The difficulties which are imposed by current programming languages and their associated systems have ramifications for everyone from analysts to end-users. Recognition of the importance of studying the users view of programming languages has led to a considerable amount of research in the last few years. This has included studies of the efficacy of particular language features, problem specification procedures, and the structure of programming projects. It seems difficult to justify the omission of this material. To discuss at length, as the report does, the content of the human/computer dialogue without reference to the principle means, and limitations, of its production, is surely a mistake.
469
π SIMILAR VOLUMES