Pitfalls in the search for a global iconography for the web

What do the ancient Maya have in common with contemporary computer users?

According to Paul Honeywill, both belong to cultures that communicate with structured visual icons that have developed in a way analogous to natural spoken languages. But whereas the system of the Maya apparently had a productive syntax that allowed complete and novel sentences to be expressed, the icons we find in computer interfaces are much more limited in scope.

This book wavers between ambitiously arguing that computer icons actually do constitute a general visual language, and somewhat more convincingly suggesting that they could if only people were motivated to use them more coherently.

The latter notion is pursued through discussion of a number of earlier proposed visual languages. Isotype, Blissymbolics, The Elephant's Memory, various systems of airport signage - all are fascinating examples of attempts to encode concepts of objects and actions into simple visual terms that can be linked and compounded. But none can be used successfully without significant learning investment, and Honeywill argues plausibly that none has become widely used because there are insufficient incentives to learn them.

Icons as we know them in common computer interfaces are widely used simply because the vast majority of computer users use the same operating systems, and some investment in learning is unavoidable. There is widespread "standardisation" of the 32-by-32 pixel format, and considerable commonality in the iconography of the images used to suggest various functions. Through the worldwide web, these standards are now known and used on an unprecedented scale. Here, perhaps, is a situation ripe for the successful introduction of a global visual language.

Yet there are problems. Identical or closely similar icons are often used inconsistently in different contexts, and are often not understood. In the interfaces of, say, word processors, it can often be shown that experienced users are simply working from their knowledge of the location of buttons for particular functions, while novice users are baffled by the icons until they see a "tool tip" or similar textual clue. On the web, icons might have at least two kinds of uses: on the one hand to provide access to interactive functions, as in application programs; on the other to assist more generally in navigation and communication. Addressing the latter, Honeywill extensively covers the use of visual features in magazine layouts and "corporate identities".

However, the book presents an uneasy compromise between

theoretical discussion and design manual. In considering the syntactic and semantic nature of icons - terms rarely used with any precision here - the Maya analogy is suggestive, but no more. Several empirical studies are described, but too loosely and briefly to illustrate learning incentive clearly. While offering detailed advice on the graphic design of icons, and a spectacular anthology of icons from websites around the world, Honeywill remains vague on the relationships between these and the graphics used in magazines and elsewhere.

The rate of change in the world of the web makes it even more difficult to identify the important issues. Icon standards are rapidly becoming anachronistic. The 32-by-32 pixel icon, already seen less often on web pages, seems doomed as a standard even for operating systems, denounced for example by the designers of the new Apple Aqua interface (see www.apple.com/macosx/aqua.html). Still, visual aspects of our culture continue to gain emphasis over the verbal, on the web and elsewhere.

Honeywill rightly highlights the importance of good design for effective communication. It is odd, however, that amid carefully composed graphics, curiously neglected flaws in editing and typesetting appear everywhere.

John Lee is deputy director, human communication research centre, University of Edinburgh.