“Chart junk” is an infamous concept in data visualisation. Since the coining of the term in the early eighties, it has led to the creation of a school of minimalistic data visualisation design and the term has found its way into the language of both visualisation practitioners as well as researchers. In the past years, a counter movement, trying to abolish the use of the word and rejecting the minimalistic visualisation design style, has come to grow in popularity.
The term chart junk was first used by Edward Tufte in his book “The Visual Display of Quantitative Information”. Tufte, with a background in statistics and political science, criticised some practices of data visualisation designers in that time, like adding patterns to the data elements in a chart, and making gridlines too visually present.
To overcome these issues, in the same book Tufte introduced the concept of “data-ink”. The data ink ratio is the proportion of the ink in a visualisation that is used to display the data divided by the total amount of ink used. According to Tufte, non-data-ink is chart junk and should be removed, within reason.
But Tufte also mentions another category of chart junk, which he compares to the “Big Duck”, a famous duck shaped building in New York, built by a duck farmer and used as a shop to sell ducks and duck eggs.
Big Duck. Source: Mike Peel, CC-BY-SA-4.0
For Tufte, the whole structure of the building is decoration, and some visualisations can be classified as Big Duck graphics:
When a graphic is taken over by decorative forms or computer debris, when the data measures and structures become Design Elements, when the overall design purveys Graphical Style rather than quantitative information, then that graphic may be called a duck in honor of the duck-form store, “Big Duck”.
In part, Tufte is criticising the emerging field of computer generated visualisations. Making graphics with computers was new, and as a result there was a lot of experimentation, which not always produced effective charts.
But he was also criticising designers being creative in adding effects and decorations to visualisations, like 3D perspective and other decoration. According to Tufte, “graphics do not become attractive and interesting through the addition of ornamental hatching and false perspective to a few bars”.
With the concept of chart junk and the “maximize data ink” rule in mind, in his book Tufte redesigns a few visualisations, resulting in ultra minimalistic looking charts. This style of visualisation, and the philosophy behind it, became very influential in the decades after Tufte’s book was published.
In his next book, “Envisioning Information”, Tufte repeated his critique on illustration style data visualisation, in which the hand of the designer was very visible. As an example, he used a graphic by (then) Time magazine illustrator Nigel Holmes: the now famous “Diamonds were a girl’s best friend” chart:
Credibility vanishes in clouds of chartjunk; who would trust a chart that looks like a video game?
Source: Nigel Holmes for Time Magazine
This kicked off a Tufte vs Holmes debate, in which Holmes claimed that “Tufte, in his insistence on absolute mathematical fidelity, remains trapped in ‘the world of academia’ and is insensitive to ‘the world of commerce,’ with its need to grab an audience”.
So, do illustrations and “chart junk” hinder the interpretation of a chart? In 2010 researchers tried to find out and settle the Tufte vs Holmes discussion by exposing people to visualisations designed by Holmes and versions of these charts following Tufte’s rules:
Left: Holmes’s graphics, right: minimalistic versions of the charts. Source: Useful Junk? The Effects of Visual Embellishment on Comprehension and Memorability of Charts, Bateman et al
The researchers found that the embellished charts did a better job in getting the message across, and that people had a better memory of the charts after a three week period. So adding illustrations and non-data visual elements does not seem to affect communicating a message in (simple) data sets. However, what the researchers did not study, was how well people were able to read data values from both kind of charts.
A sane way of looking at the Tufte versus Holmes is to see them as two extremes of a continuum. How much ink you can and should spend on non-data elements in your designs depends on the medium, the audience, the topic and also of your own taste and style. If you are after readable values, effective comparisons, and a scientifical look: go towards minimalistic. If you want engagement, some humour and memorability, go more for the Nigel Holmes style.
