Book Reviews

Transcribed from the printed issue; not yet reviewed. Read it in the PDF of the issue, page 33

Diagnosing the System for organisations¶

by Stafford Beer, John Wiley & Sons, ISBN 0 471 90675 1

Diagnosing the System is the latest in a series of books in which Professor Beer sets out to explain the theories of the managerial cybernetics of organization, i.e. what are the fundamental requirements of an organization’s structure in order for it to be viable. The companion books in the series – Brain of the Firm and The Heart of Enterprise – are both large volumes which develop the subject in principally theoretical terms, explaining the rationale behind the model for a viable system and the basis for the the various cybernetic laws underlying it. However the drawback with this type of theoretical approach is that it does not provide the reader with the right sort of understanding to go away and start using the approach. This book attempts to redress the balance by showing how to design a suitable organizational structure and how to diagnose a faulty one.

The first impression of the book – at least to a reader of its precursors – is its size, a mere 150 pages compared to the several hundred in Professor Beer’s other works. The second impression is of a highly visual presentation of the subject material. The previous books present the viable system model in diagrammatic form, but Diagnosing the System takes the graphics a stage further by liberal use of red ink and expressive hand-drawn diagrams. The book looks as though the author has personally taken your copy and added chosen comments as an afterthought.

The style of the content is equally distinctive with continual emphasis on the practical. By page 6, under the bold red heading

“NOW DO THIS”

the reader is instructed to look at his personal position as a viable system, and on page 16 the tone of the rest of the book is set with a similar instruction to choose and define a specific system that the reader is to model throughout the rest of the book. As each stage of the theory unfolds revealed, so the exercises require the reader to relate it to his chosen system.

This strong emphasis on the practical is reinforced by a punchy, almost abrupt presentation. The style is more like a lecturer’s notes than an academic treatise and shows evidence of its origins in the material for a post-graduate course in the cybernetics of organizations. Few paragraphs are longer than 5 or 6 lines; sections and subsections are not numbered, but are distinguished by colour, indentation, etc.; diagrams abound; and the jargon is kept to a minimum and is summarised, along with the major conclusions, at the end of each chapter.

This practical style enables the book to be much shorter than a corresponding purely theoretical work, but puts obligations on the reader, since the principal message comes from the application of the concepts to the reader’s own chosen system. In this sense the reader can out take out of the book as much as he is willing to contribute.

The target audience for this book are those readers who face a real organizational problem, either as managers or their consultants, and who use the book as a handbook to diagnose and redesign their own systems. This will include people who have already read up the subject of the cybernetics of organizations (perhaps in the previous books in this series) as well as those who are new to it. However the book should also provide a valuable introduction for newcomers to the subject whose present interest is more conceptual than practical, but who find the practical style of this book easier than the more conventional approach of its predecessors.

References.
Brain of the Firm, S. Beer (1972), Allen Lane, Penguin, Harmondsworth.
The Heart of Enterprise, S. Beer (1979), Wiley, Chichester.

The Visual Display of Quantitative Information¶

by Edward R Tufte
Graphics Press, Box 430, Cheshire, Connecticut.

This book is not a recent issue (it first appeared in 1983), but is well worth a belated review because its valuable contribution to an increasingly important area deserves a wider readership than it seems to have achieved. It is one of the few books that provide the development of a theoretical approach to a subject in a book that is visually attractive and a pleasure to read.

The subject area is the effective presentation of data in a visual form. Whilst this is not specifically concerned with computer graphics, Tufte emphasises its importance is increased by the potential of computers to generate mediocre graphics.

“At least a few computer graphics only evoke the response ‘Isn’t it remarkable that the computer can be programmed to draw like that?’ instead of ‘My, what interesting data.’”

The book comprises two main sections, entitled Graphical Practice and Theory of Data Graphics. The first section starts with a chapter on Graphical Excellence, which contains some of the best examples of graphical displays of data from the last 200 years. (1986 is in fact the bicentenary of the publication of the first known time-series). My personal favourites illustrate the Paris-Lyon train schedules of 1880 and Napoleon’s Russian campaign of 1812-13. It is a well-worn cliche that “A picture is worth a thousand words”; Tufte shows how a graph is worth 700 words by comparing one with a newspaper report exposing how US representatives exploit their privilege of free postage in the run-up to elections.

The next chapter (on Graphical Integrity) introduces the Lie-factor, i.e. the ratio of the size of an effect as shown by a graph to its size in the data, again illustrated by published examples mostly from an assortment of the world’s best-known newspapers. Tufte makes the point that one lie in a newspaper is a lie repeated hundreds of thousands of times – a message that should be taken to heart by those designing chart layouts for use on computer systems. He also points out that the occurrence of even the well-known abuses of data is continuing to increase.

The first half of the book provides a very clear and enjoyable coverage of material that is, on the whole, available from other well-known books. It is when Tufte starts to develop his theory of graphic design that new ideas are expressed. There are many publications that offer rules for good graphical design, including some from computer graphics equipment suppliers and most of the advice that they give is good. However their recommendations are simply rules, and don’t allow much scope for imagination; properly applied they will prevent the worst excesses of bad design, but they tend to suppress the innovation needed to produce excellent graphics. This book differs in that it establishes a clear theory explaining why some charts are misleading and some are particularly effective.

The central themes of this theory are the maximisation of the “data-ink ratio”, i.e. the proportion of ink devoted to the non-redundant display of data information, and the elimination of chartjunk. The reader is rapidly led to question the need for boxes, grids, axes and tick-marks and to keep a watchful eye for the three main forms of chartjunk. These are described as unintentional optical art, grids and “ducks”. Any user of the standard computer graphics packages and devices will have encountered the range of “highland-tartan” patterns which can automatically shade an otherwise conventional bar-chart and serve principally to distract from the message of the data. Fewer will have detected the less immediate impact of grids which, unless used with caution, can dominate the information content of a chart. The “duck” is the example where the construction of the decoration of a chart has taken precedence over the presentation of the information:

“The symptoms of the We-Used-A-Computer-To-Build-A-Duck Syndrome appear in a display from a professional journal: the thin substance; the clotted, crinkly lettering all in upper-case sans serif; the pointlessly ordered cross-hatching; the labels written in computer abbreviation; the optical vibration.”

The final three chapters concentrate on how to use the theory to create excellent graphics – the benefits and risks of multifunctioning graphical elements; the advantages of high information graphics using data density and multiple small charts; and the incorporation and aesthetics and good technique.

This book should be compulsory reading for anybody actively involved in the design of quantitative graphics, particularly for computer systems which will be released to untrained users. My major complaint concerns availability as the book used to require a direct order to the publishers and was hindered by the presence of at least two Graphics Presses in the U.S.A. However I understand that it can be obtained within the U.K. from Graphics Press UK, PO Box 8, Godalming, Surrey.

Graphics & Image in Office Systems¶

by S.J.Newton
N.C.C. Publications. ISBN 0 85012 425 5

This book is one of the N.C.C. series surveying various computer-related topics. Its stated aim is to “highlight developments in the hardware and software of business graphics, explore the use of graphics systems and consider the relevance of standards.” However the book goes further by including sections on image processing as well as more conventional graphics.

A list of the chapter headings should indicate the extent of the coverage of the book. They are:

This is a very ambitious list of contents for a fairly short book, and my initial fear was that the task of compressing the material into so short a space would reduce the book to an unacceptable degree of generality.

The hardware section starts by identifying three basic issues – the method of generation (vector or raster), resolution, and the choice of colour or monochrome. Newton then categorises current graphics devices under the headings of displays, input devices and output devices. I detected some confusion between devices for graphics input and for control of graphics systems. For instance a light pen can be regarded as a graphics device when used for drawing images, but for selecting from a series of options it is irrelevant whether the system under control is a business graphics system or not. Consequently I found much of the discussion about the relative merits of keyboards and voice input irrelevant to the subject of business graphics.

Having just finished re-reading Tufte’s Visual Display of Quantitative Information (reviewed above), I was inevitably going to be disappointed by Newton’s section on choosing and using charts. Newton’s is a much more straightforward approach describing the different types of chart in common usage and giving some of the basic rules for good chart design. I remain unconvinced that a ‘word chart’ is really a business graphics application, and would strongly disagree with almost any advocate of the pie-chart. The rules for good graphics were perfectly reasonable – Determine your message, Choose the appropriate chart form, Keep it simple, . . . – but their impact would have been much better if the first charts in the book (from another NCC publication) had actually obeyed them.

The subsequent chapters continue in like vein. They provide a fairly simple coverage of the subject in a reasonably conventional approach without providing any startling revelations. Most readers will find some new material in the book, but it won’t change their thinking overnight.

The book fails for me because it tries to cover far too wide a subject area and has no clear audience. It attempts to combine a state-of-the-art review of current hardware with recommendations on chart design, a justication of the use of graphics and a review of standards – each of which could merit a book of its own. The explanation of how to create a pie-chart:

“the data is computed as a percentage of a total and each data value (percent) is converted to its appropriate segment of a full 360 degree circle”

does seem somewhat unnecessary when the facing page shows the result. On the other hand the organisation structure of the International Standards Organisation and the levels of adoption of six different standards are likely to interest only the specialists in the area.