I have been posting an occasional series to the office email about unhelpful visualisations.
Having been fascinated by good qualities of The Cambridge Phenomenon influence map (1) (2), I then wrote about the Star Wars influence map:
This influence map tracks the same kind of connections as the Cambridge Phenomenon one I sent out a few weeks ago. Except in this case, people, organisations, and projects are given the same status.
Overlapping dotted lines and many unabeled link-types produce a confusing diagram. We can guess the nature of the relationship between two connected items (it's a guess because there are not labels), but it's not possible to know the relationship among three connected items. Lines seem to have directionality, but it's hard to know how that works, especially when lines directly intersect around nothing in particular.
It's a pretty diagram, and it effectively communicates its point. Although the function of such graphs is to support and illustrate that point, the unclear presentation doesn't satisfy far beyond the initial impression.
Today, I came across another unhelpful diagram: a history of information, a stacked graph which apparently displays data on media users since 1800. This beautiful and impressive diagram shows the transitions between epochs of news history: newspapers slowly supplant word-of-mouth, television and radio start edging in, and by 2020, thin bands of online options -- websites, blogs, social networks, social news, and targeted news-- completely supplant the old world.
The design elements of authority are all there to support belief in a coming future of targeted news: evenly spaced labels, vertical bars for each label, and specific datapoints for each medium for each labeled year. The graph tells the whig history of media as a story of inevitable progress from one new medium to the next. Next to the progression of newspapers, television, and websites, social news and targeted media look like sure winners in the next decade.
The graph, however, is a fabrication. Baekdal cites no actual sources for his graph(although he says the last 10 years are based loosely on what he has seen from "probably" 1000 surveys he has conducted. For data about life before 1990, he interviewed people and googled some stuff). But even if it were properly researched, it would be deceptive at best:
- Time is stretched inconsistently: the intervals between labels begin at 50 years, then 20, to 10, 5, 2, 1, and then 5. With time made so flexible, we cannot trust the implicit argument of progress, which relies on the visual similarity of subsequent regions of color. When questioned about this, Baekdal said, "I fail to see how the scale of the graph can be seen as a lie. You can clearly see each year."
- The total sum remains constant: The last 200 years have seen vast increases in media audiences: literacy, communications technology, and the rise of the gobal middle class have all expanded the pool of media consumers. A study that recorded socioeconomic data alongside media use would be able to display or (possibly) adjust for population growth within that group.
To be fair, there is no way to design the graph correctly. It is simply not honest to present non-quantitative opinions using means reserved for quantitative information. Although he argues in the comments that his graph is based on interviews, surveys, and "careful analysis", he finally admits that the "graph does not illustrate the size of the different forms of media. It illustrate their importance."
P.S. Notice how the color selection subtly supports Baekdal's argument. Newspapers, radio, and television have colors which contrast each other, while the electronic media "websites, blogs, social networks, and social news" all have similar colors. The eye naturally groups them, contrasting them strongly with television.