Wednesday, April 16, 2014

Annotation, narrative, and storytelling in infographics and visualization

Screenshot by Robert Kosara
The latest podcast has just been published. As you can see in the photo, Moritz Stefaner, Enrico Bertini, Robert Kosara, and I had a lot of fun for one hour and a half.

We were planning to engage in a heated debate about the glories and shortcomings of storytelling in visualization, but we ended up agreeing with each other a lot. To understand where this all came from, here are some articles and talks that preceded our conversation, and informed it:

• Cole Nussbaumer's series on storytelling techniques 
• Moritz Stefaner's Worlds, not stories
• Periscopic's A Framework for Talking About Data Narration
• Robert Kosara's Stories Are Gateways Into Worlds
• Robert Kosara's Story: A Definition
• Lynn Cherny's Implied Stories (and Data Vis)
• The presentations at Tapestry 2014 and NICAR.

Long story short (no pun intended): Some visualization designers are in favor of storytelling techniques, and others are against them or, at least, against overusing them. And everyone has a personal definition of the term, when applied to information graphics.

In the podcast I explained that I see differences between annotation, narration, and storytelling. After we finished the Skype call, I sent Enrico and Moritz an e-mail to explain myself a bit better. It's copied below (I've made some minor edits.) You may want to read it only after listening to the conversation.
Hi everyone, 
When you publish the podcast, you may want to add an explanation of what I had in mind during the discussion. You can copy and paste this entire e-mail, including this paragraph. I would like to say in advance that all these ideas need a lot of development. I'd love to hear/read some comments on them. 
1. Annotation consists of highlighting certain data points or interesting phenomena in a visualization, and perhaps describing them or putting them in context. For instance, this chart by The New York Times. Notice how the designer provides an explanation for the relevant data points.
Or this interactive visualization on breast cancer rates. See how the information is sequenced in it. 
And why isn't this second graphic a "narration", even if it's organized as a sequence? Keep reading.
2. Narration consists of arranging your charts, maps, and diagrams as a meaningful sequence intended to display cause and effect relationships, no matter how fuzzy they are. This cause-and-effect is the crucial point here. See an example
3. Storytelling: As I mentioned during the podcast, I am growing fond of the definition of story provided by the book The Unpersuadables, by Will Storr: "A story is a description of something happening that contains some form of sensation or drama (...), an explanation of cause and effect that is soaked in emotion."  So, a story in visualization is a narration in which the designer tries to instill an emotional component, rather than relying just on the intrinsic interest of the information presented. 
As an example, I mentioned an infographic about population trends in Brazil that I also described in The Functional Art. Take a look at it
Here, the information is organized in a way that resembles the traditional structure of stories: An opening ("Brazilian population grew between 2000 and 2010,") a surprising fact that becomes a conflict ("but Brazil's fertility rate is way below expected,") the consequences of that conflict ("Brazil's population will start shrinking in 20 years, and it'll become older,") and a conclusion or resolution ("what measures Brazil can adopt to face this future scenario.") All this is based on solid evidence, of course. Storytelling doesn't mean making stuff up. 
We wanted to challenge our readers with this graphic. The emotions we wanted to foster were surprise and, consequently, concern, and curiosity. Most people in Brazil know that its population has been growing healthily in the past, but many don't know that women nowadays have just 1.8 children, which is below the replacement rate, currently at 2.1 children per woman.
Again, I'm aware that this is a very preliminary set of terms, and that the boundaries between them are very blurry. I may change everything completely for my next book.