Glass benches: genre and minimalism in data visualization

(Part one of two)

I went to the campus art museum this weekend and found an unexpected puzzle. Which of these objects am I allowed to touch?

IMG_9509

IMG_9502

I go to museums a lot. In my mental model of a museum, anything with a little rope around it or a dark outline on the floor is an Exhibit and Not To Be Touched. By those rules, the bench is forbidden, but the plinth is fair game.

To the security guard’s amusement/alarm, I guessed wrong: I avoided the bench, and tried to put my sticky fingers on the shiny surface of the plinth. I was misled by genre.

In “Graphic Literacies for the Digital Age,” Robert Waller argues that genre indicates where to look for information and how to interpret what we see. The first scene of a crime procedural shows me a crime; the bolded location at the start of a newspaper article tells me where the article was written. I learned to interpret datelines after years of interacting with newspapers.

Not all documents have genres, though, and not all readers are familiar with all genres. Waller writes, “Ideally, document designers can take an existing genre as a model… In the absence of a strong genre, the new design falls back on the core techniques of information design.”

Data visualization has genres of its own. Readers can rely on their mental models to interpret bar charts because they have interacted with them so often. However, not all charts belong to a common genre, and readers might not be familiar with the genres that do exist. In those cases, it’s up to the designer to guide (and sometimes instruct) the reader.

Unusual charts can lean on existing genre conventions. In A People Map of the US, cities are labelled by “their most Wikipedia’d resident.”

Screen Shot 2019-09-23 at 9.58.51 PM

The concept is unusual, but I know how to zoom and pan because it follows the genre conventions of online maps. No further instructions are necessary.

In contrast, What’s killing us now? from The Financial Times uses extensive annotation, rather than relying on reader’s mental models.

whats_killing_us.png

The horizontal axis represents time, from 1841 to 2016. The vertical axis represents age. The colors show mortality rates. It’s a very fine-grained heat map. And I would have struggled to interpret it without the extensive annotations pointing out darker spots with greater mortality, and lighter areas with lower mortality. The context provided by the annotations helps me to understand what I’m seeing, even when a familiar genre (heat map) is applied to an unusual subject (175 years of mortality rates).

A lot of ink is spilled in that heat map and the bubble chart that accompanies it. I wouldn’t call either of these graphics minimalist. They spend ink to contextualize and support data, not just to present it. In doing so, they fly in the face of Edward Tufte’s “data-ink ratio,” which has inspired minimalism in data visualization since The Visual Display of Quantitative Information was published in the 1980s.

According to the data-ink ratio, a graphic should spend as much of its ink as possible in encoding non-redundant data. Removing redundant ink does clear things up one of Tufte’s examples:

IMG_9545 IMG_9546

However, Tufte also argues that redundant data ink includes the bars in a bar chart:

IMG_9547

He suggests the following makeover as well:

IMG_9549  IMG_9548

The remade chart does use less ink. The information is encoded very efficiently. But it’s encoded in a language I can’t read. I have no mental model to support what I’m seeing. In other words: that chart has no genre.

Minimalism relies on genre. Line charts don’t require very much explanation, not because they are perfectly intuitive, but because the reader is used to them and knows how to interpret what they see. This line chart floats in FiveThirtyEight’s margins:

Screen Shot 2019-09-24 at 12.18.15 AM

The designer didn’t spend ink on axes or tick marks. The information is still present, because the reader has seen enough line charts to fill in the blanks. In order to communicate, high data-ink ratios must be subsidized by a reader’s past experience.

Tufte wrote, “Every bit of ink on a graphic requires a reason. And nearly always that reason should be that the ink presents new information.” I would argue in favor of an information-ink ratio: how much of the ink in a graphic is necessary to support a reader’s understanding?

Creative charts can become more intuitive by referring to other genres. Or, they can explain themselves. Part II will take a look at how unusual graphics explain themselves to readers, along with some highly subjective do’s and don’ts. Until then, remember: your reader might not know if they can sit on the bench.

A woman sitting on a glass bench in an art gallery

(The glass bench is Frosted Prism Bench by John Lewis)

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s