Edward R. Tufte
Graphics Press, 1997
Edward R. Tufte
Graphics Press, 1997
ISBN 0961392126 (boards)
Buy it from Amazon.com
Buy it from Amazon.co.uk
Other Reviews
Tufte is a polymath. He would justifiably be famous for his beautiful and provocative books The Visual Display of Quantitative Information, and Envisioning Information, even if he were not also a Yale professor, a writer on politics and democracy, and the recipient of awards for statistics, science, and software documentation.
Visual Explanations is a fascinating book in its own right - though, as you may have guessed, it can also be seen as the third part of an extended and even impassioned monograph.
Tufte's thesis, roughly summarized, is that the human eye and brain are beautifully co-ordinated for appreciating complex information, presented visually in 2 or 3 dimensions. Presenters of information (such as requirements engineers) should therefore learn and exploit the principles of displaying, envisioning, and explaining visually as well as verbally.
It should be no surprise, then, that this is a beautiful book, carefully typeset on thick creamy paper, richly illustrated in monochrome and in colour with engravings, modern illustrations including many designed by Tufte himself, maps both good and bad, and not least with explanations both clear and confusing.
Users of DOORS (up to v4, or users of its 'classic' colour scheme after that) will be familiar with its characteristic 'sky blue' backgrounds and careful use of colours to distinguish objects with different properties, such as read-only from editable text. Richard Stevens, the inventor of that tool, showed his admiration in that practical way. The goal, of course, is to make properties immediately clear without the need for error messages and other textual devices. In Visual Explanations, Tufte ruthlessly compares several well-designed displays with 'ghastly' counter-examples.
The General Bathymetric Chart of the Oceans, for instance, uses a gently-graded series of blues to denote depth, and a similar gradation of browns to indicate heights above sea-level. The colours make it obvious without looking at the key where the oceans are deep. Tufte remarks: 'this classic map .. makes extraordinary use of small and effective differences .. with "the deeper or the higher, the darker the color" serving as the visual metaphor for the color scale. To indicate depth, the contour lines are labeled by numbers, a design that enhances accuracy of reading and nearly eliminates any need to refer back to the legend'. Tufte then constructs a rainbow-encoded map where red means shallowest, yellow through green and blue intermediate depths, and violets the deepest water. Despite the familiarity of the colour-sequence, the map is virtually unusable. Tufte comments: 'Although often found in scientific publications, such a visually naive color-scale would be laughed right out of the field (or ocean) of cartography. These aggressive colors, so unnatural and unquantitative, render the map incoherent...'
Tufte's strong opinions are supported by an impressive array of evidence, drawn from fields as diverse as landscape gardening - he shows how Repton exaggerated the impact of his improvements, for commercial effect; science, where Isaac Newton's own careful optical diagrams are seen to be successively corrupted by popularizers; and even modern dance, where Margaret Morris 'in graceful parallel' starts her book with a posed photograph of herself, and a diagram in her own notation of movement describing exactly her pose.
Tufte even ventures underwater to show how amateur divers can be helped to recognize tropical fishes with a waterproof guidebook - similar butterflyfishes are grouped on a sheet to highlight differences. He cannot resist pointing out that the foureye butterflyfish, 'recognized by large false eyes near the tail and by concealed real eyes masked by a dark line passing around the head. This is another example of disinformation design...' and he goes on to explain the effect of highlighting, as well as pointing out that since refraction magnifies objects seen underwater, the problem of turning 'all those wet, sticky pages underwater' can be reduced by printing the images of the fishes smaller than usual, saving pages.
A final suggestion, with reference to Herbert Simon's analysis of skewed distributions, is to show only the most commonly-seen fishes - to avoid distracting inexperienced divers with thousands of rare species. 'Good design should take into account how, when, and where the information is used.'
The book includes a detailed, beautiful, and horrifying account of the confused visual explanations leading up to the Challenger disaster. Tufte argues that if the temperature and O-ring failure (blow-by) data had simply been plotted or tabulated clearly, everyone would have seen what is obvious in hindsight - that a low-temperature launch would be disastrous. The fatal launch was at 29°F; 'the O-rings had persistent problems at cooler temperatures: indeed, every launch below 66° resulted in damaged O-rings ... there is a scandalous discrepancy between the intellectual tasks at hand and the images created to serve those tasks. [The images] failed to reveal a risk that was in fact present [and] failed to persuade government officials that a cold-weather launch might be dangerous.' Tufte devotes 16 pages to a careful analysis of the attempted and true explanations, including Feynman's famous, brave but admittedly unscientific public demonstration of the problem. Tufte naturally cannot resist producing a diagram in crisp Scientific American style of the controlled experiment that Feynman would have liked to have performed.
Tufte's combination of scientific precision and genuine aesthetic sensibility with a practical outlet is rare if not unique. 'Information-sensitive designs are exacting and laborious, requiring a deep appreciation of the particular content at hand.' Interface designers often (read, usually) take other 'more generic' approaches - in other words they do whatever is easiest and most familiar, instead of addressing the real requirements in the current situation. Tufte is a stern schoolmaster; he despises 'chartjunk', all those little coloured coffins depicting death-rates where simple scatterplots or graphs should be used. He closes with the words 'by virtue of the architecture of their arguments, [good synthetic visual designs] make reading and seeing and thinking identical.'
The job of requirements engineering includes communicating accurately the needs of one set of people (the stakeholders) to another (the developers). Tufte describes a set of visual and intellectual tools necessary for this kind of task.
© Ian Alexander 2000 You may also like:
