Autumn 2003

Information visualisation (text in full)

Nico Macdonald
Overview

Modern-day map-making may be a way out of Web design’s stasis

In 1569 the Flemish cartographer Gerardus Mercator published a map of the world in which parallels and meridians were rendered as straight lines at right angles, spaced so as to produce at any point on a map an accurate ratio of latitude to longitude. Mercator’s work in developing visual presentation helped to change the perception of the new world, and for the budding mercantile classes, mapping the earth made it possible for new industries and modes of commerce to emerge.(1)

Over the past 50 years we have created another new world using micro-processors and networks. Digital technology – from cameras to keyboards, sensors to measurement devices – allows us to gather unprecedented amounts of data. Computers manipulate, store and present this data on a gigantic scale; the Internet gives its users access at high speed and low cost.

Society has become increasingly complex. Businesses co-ordinate more resources in more places ‘just-in-time’. Financiers monitor companies, industries, markets and commodity prices in real time. Researchers and scientists analyse everything from chemical compounds to geological data and the human genome in increasingly pressured environments. Governments at the centre of more dynamic populations need to plan everything from housing to trans-portation, healthcare to employment.

Yet, most of us still try to comprehend this glut of data using representations from the era of print – many based on text rather than image. Not only that, we have been viewing them on monitors that display less information than a letter-size sheet of paper. There is a clear danger of ‘information overload’ or maybe ‘understanding underload’ – a failure to focus, to prioritise, and to apply wisdom to information. Our ability to present information in a useful and intelligent manner is falling behind our ability to create and distribute the raw data.

For the designer, this is the challenge of ‘information visualisation’, the process of revealing useful, easily grasped insights by transforming abstract data into visual and manipulable forms. In other words: making new maps for the new, information-rich universe.

Evolution has left humans with brains that are as much visual as they are analytical, able to distinguish and group objects by size, colour, shape and spatial location. Our brains are adept at identifying patterns. ‘Millions of leaves on thousands of plants – and we can still pick out the ripe bananas’, notes artist and engineer Brad Paley.

Information visualisation taps into these powerful cognitive skills in ways that textual and numerical information cannot. A visualisation is more universally readable than any single language, and it can be ‘read’ more quickly. It can show quantitative information, but it is also infinitely variable, and aids comparison in ways that text can’t match. A good example might present multiple dimensions of information, trends and patterns, and show information in context. Or show the qualitative state of complex systems ‘at a glance’, allowing for quick identification of status and changes in the system.

The concept is not new. From Mercator through Minard (and the other lesser heroes celebrated by Edward Tufte) via Colin Ware(2) to contemporary protagonists such as Richard Saul Wurman(3) the concept of visualisation has been fervently evangelised – but only in print. In the digital world its develop-ment has been slow, though the idea of graphical displays, epitomised by Ivan Sutherland’s work at mit and the University of Utah, dates back to the early 1960s. Sutherland’s successors in the field include Xerox parc veterans Stu Card and Jock Mackinlay and University of Maryland hci Lab founding director Ben Shneiderman.(4)

The graphical user interface has been with us for some time, joined ten years ago by the graphical Web browser, but in these cases ‘graphical’ refers to the use of visual techniques for task execution and information navigation. Actual information is usually presented as text or – at best – as a static graphic. Though there is a need to develop visualisation techniques and models, several factors have limited development and there are many misconceptions – often based on the print heritage.

Complexity demands that we devise new ways of seeing data. ‘There are a large number of conventions about visualising information, especially in business’, notes Paul Kahn, founder of Paris-based Kahn+Associates. He gives financial data as an example: ‘The typical data visualisations – bar charts, pie charts – become so conventional that they more or less look the same and hence are less useful’. He believes that the ‘Map of the Markets’ tool created by SmartMoney magazine represents the kind of progress that is needed. This was a landmark, combining a cognitively powerful visualisation with elegance and aesthetics to create a high quality of experience.

‘People think of data visualisation as an insert into an application’, observes Mark Schindler of Visual I | O. ‘This is where Tufte stops – he focuses on static ways of representing data that is often dynamic’. Schindler, who describes traditional data visualisation as the translation of one set of facts to another, argues for an approach in which data can be seen in context from a number of different vantage points. User interaction with data presenting results in real time – ‘direct manipulation’ according to Shneiderman’s terminology – is a powerful element of information visualisation but one that is often missed out.

Three-dimensional visualisation is another area that escapes the limitations of print. One recent initiative is an interface for the knowledge mapping tool Antartica, which presents a three-dimensional interface to its geographical metaphor for information organisation. Kahn sees value in the use of an extra dimension for highlighting information, but regrets the way that 3d navigation ‘undermines your understanding of what you are looking at and where it is’.

So why should clients bother with information visualisation? Gong Szeto sees the value for organisations, noting that analysis – of marketplace dynamics, consumption patterns, and the like – is a core business activity and clients need to ‘figure out if the business is performing on plan or off plan and why, and to validate past decisions and support future ones’. An article in the Economist observes that ‘after rounds of lay-offs, companies have fewer people to take complex decisions’.

In the area of research and development, information visualisation can be the determining factor in success. Yet data analysis is not keeping pace with technological advances in other areas. The Human Genome Project is an example of the challenge of understanding large amounts of data.

Bob Quinn of Battelle, owner of visualisation tool developer OmniViz, believes that data visualisation is a way for business to gain competitive advantage: since many companies receive exactly the same data at the same time, ‘the winners will be those who can quickly transform raw data into sources of high value information’. Another area of business that can benefit is communication. Kahn sees a role for information visualisation in sharing knowledge by ‘helping to create a common mental model among groups of people who need to understand it’.

So if information visualisation is an appropriate solution to a design problem, the data is there, and the client and stakeholders are convinced of its value, how are designers to proceed? They need to begin by understanding the nature of the information, where it comes from, how and where it is used, what insights users will be looking for and their relative importance. Financial traders will be more interested in seeing changes in existing data relationships (see Thomson Financial example), while researchers may wish to find new data relationships. The needs of beginner and expert users may also need to be considered, particularly if training will be limited or some people will only use the tool infrequently.

Issues of perception, cognition and visual psychology need to be taken into account when considering how to present, contextualise and prioritise information, and support its manipulation. These relate to the number of dimensions of data presented, spatial positioning and relationships, colour and tone, and dynamic changes in these elements. At the level of interface and human-computer interaction, the designer needs to analyse the tasks to be supported, and address comprehensibility, navigability and the effectiveness of

any solution, including its usability. There may be a visual metaphor around which an interface can be structured. Visual I | O’s Angel Shen-Hsieh develops metaphors by ‘discerning the mental picture that expert decision-makers have.’

Developing interface concepts may begin with sketching but will likely require more sophisticated prototyping as, unlike many areas of design, sketches don’t provide a good basis for user (or even heuristic) testing. It is also important to work with the people who manage the data the tool will draw from, and those responsible for its implementation, to ensure the data can be delivered in the form it is needed and that the tool will perform quickly and reliably. The designer will also need to address the overall quality of experience of the product and the engagement of the user, considering visual appeal, information design and typography and interaction style.

Yet information visualisation is not well served by tools that support these challenges: designers find that they have to custom-create or re-combine tools for anything other than simple interfaces. Clients also need tools to input, structure and annotate information at the back end – a design deliverable that is often forgotten on even the most mundane projects.

Many of the developments that will enable information visualisation to become as commonplace in the future as the graphical user interface is today are either technological or infrastructural. Higher quality, cheaper and more portable displays will allow the combination of the qualities of paper and digital media. Pervasive networks will allow people to get to information where and when they need it. And structured languages will allow data from disparate sources to be combined for new and practical uses.

Others factors involve a greater commitment to the idea: on the part of clients to invest in the right solutions; and on the part of designers in developing skills in information visualisation, and in communicating its value to clients. Focusing on one industry sector’s needs, and doing a lot of testing and evaluation, presents one way forward.

The Web design industry appears to have arrived at a fixed set of solutions which, while adequate, preclude real innovation. Meanwhile, businesses and other organisations have moved from using computers as document processing and basic communication devices to deploying them for enterprise management, logistics and manufacturing control, business and financial analysis, realtime decision-making, and strategic planning. If designers and engineers could use information visualisation to do for these organisations what the spreadsheet did for them over twenty years ago, design would be taken more seriously as a business resource and the spin-off benefits to society would be tremendous. They may do for the twenty-first century what Mercator did for the sixteenth. But if we get to this point, don’t expect everyone to be talking about ‘information visualisation’: it will become the norm, barely recognised. We will merely marvel that people ever had to work in any other way.

Thanks to Alison Lee at IBM Research for help with this article.

References, including those in this article, can be found at writing.spy.co.uk/Articles/Eye/DataVisualisation

1. The Mercator map, centred around the Greenwich Meridian, shaped many people’s view of the world for centuries. In the 1960s, geographer Arno Peters created a map, the Peters Projection, in which each country’s surface area was represented more accurately.

2. US-based UK émigré and author of Information Visualization: Perception for Design.

3. Richard Saul Wurman (see Eye no. 28 vol. 7) created the pioneering Access map series, edited Information Architects and initiated Understanding USA.

4. Stu Card, Jock Mackinlay

and Ben Shneiderman co-edited Readings in Information Visualization: Using Vision to Think (Morgan Kaufmann, 1999). More recently, Shneiderman co-authored The Craft of Information Visualization: Readings and Reflections (Morgan Kaufmann, 2003).

Tracker Pixel for Entry