The medium is the message

Most scientists live in a world densely populated with graphic representations of data. The 15 June issue of Nature contained more than 70 graphs or graph-like presentations of results, to say nothing of a clutch of advertisements that incorporate such devices. Every graph or chart embodies analytical assumptions, and at a deeper level the implicit cognitive structures that shape every phase of a scientific study.

A graph is not only a crucial tool in plotting data within parameters that are taken as significant, but is also an integral part of the style of a presentation. It serves the rhetoric of irrefutable precision that operates whenever we aspire to proceed in a 'scientific' manner.

These remarks are occasioned by a coruscating attack upon PowerPoint presentations in Edward Tufte's latest book, Beautiful Evidence (Graphics Press, 2006). Tufte is an expert in public affairs at Yale University; he is also an expert in the presentation of statistical evidence and a sculptor, and has established himself as the world's leading analyst of graphic information.

Centring his attack on the revelations in the 2005 report on the 2003 Columbia space-shuttle disaster, Tufte lays bare the special kinds of hierarchies, abbreviations, neologisms, linguistic tropes and implicit cognitive structures that were embedded in the PowerPoint presentations used by NASA officials. While the damaged craft wheeled in orbit, their modes of presentation worked against a proper analysis of the complex uncertainties and risks occasioned by the foam debris that had struck the wing during take-off.

Credit: E. TUFTE

Echoing the criticisms in the report itself, and comparable strictures expressed by Richard Feynman at the inquiry into the 1986 Challenger disaster, Tufte concludes that “the pitch-style typography of PowerPoint is hopeless for science and engineering”. The rat-a-tat style of bullet points, arranged in aggressive hierarchies and coupled with compressed jargon, insider acronyms and summary graphics, took over from the need to present supporting data, documentation and modes of analysis.

Rather than being narrative reports, capable of debating the complexities and laying them open for decision-makers, the NASA presentations relied on a series of PowerPoint slides, each typically containing 40 or so words (equivalent to a mere 8 seconds of normal reading). Typically, each slide is succeeded by the next in such a way that vital logical continuities fall between the gaps.

A typical template in PowerPoint is the standard line graph, such as the one of cancer survival rates pictured left. It is linked directly to a data sheet and can be adapted for any purpose in which such a display is desired. The three-dimensionality of the plot increases its graphic impact and enhances its air of reality. But we do not necessarily know anything about the nature of the data sets that lie behind it, nor of the methods of data gathering and modes of analysis. A 'result' is all too readily presented as 'significant' without meeting the statistical criteria for significance.

Such charts, delivered with pleasing facility by software packages, permeate business and other non-scientific subjects that generate quantitative information. Government and other official reports, and articles in the press on economics and the social sciences are permeated by facile graphics that look all too convincing. Of course, real science never resorts to such sloppy shorthand — does it?

Let us leave the last word to Tufte: “Making an evidence presentation is a moral act as well as an intellectual activity. To maintain standards of quality, relevance, and integrity of evidence, consumers of presentations should insist that presenters be held intellectually and ethically responsible for what they show and tell. Thus consuming is also an intellectual and moral activity.”