Information is fascinating at many different levels. Show me a simple graph of the components that make up a whole, that tells me something. I’ve found almost anything is more interesting when looked at across time, since it adds another dimension. This also helps with analysis, because outliers or rapid changes are often related to historical events. Compare a few data sets across time, and you have more perspective, and even more information.
The ebb and flow of the use of trains in border crossings is interesting on its own, but by comparing it to the ebb and flow of personal vehicles used in border crossings, you learn more about both data sets.
Which brings me to the value of parallelism, Tufte’s topic in this week’s reading, “Parallelism: Repetition and Change, Comparison and Surprise” from “Visual Explanations.” He explains how valuable it is to look at data sets at once, in one graph, than have several different graphs across time that each stand on their own. Of course, it is possible to clutter up a visualization with too much information, so one must be careful.
Tufte also focuses on the capabilities of computers in allowing a user to navigate through complex data to the part they want to focus on, permitting them to repeatedly listen to and analyze a certain part of a musical composition, for example. This relates to the power of interactive tools in data viz, and generally journalistic content. Break it down into digestable chunks, let people step in or out of the content as much or little as they want. The details are there, but you don’t have to absorb all of it to get the point. Or go ahead and absorb as much as you want. Have fun exploring one chunk very deeply, because you want to get its nuance. It’s up to the user to make his or her own personalized experience.
Because of these benefits of parallelism, the concept has been used to help people understand information in a literary sense for years — I still remember learning about this in high school, maybe middle school. And a graph showing multiple sets at once is extremely popular, and one of my favorite types of visualizations. Color one line different from the others, and you’re making a point of analysis. Keep it objective, and you’re opening analysis up to the user. Versatile, and ripe for as much depth, or lack of depth, as a user wants.
The first time I was introduced to parallelism was with the Name Voyager, looking at trends in baby name popularity. The concept has showed up in some of the New York Times’ visualizations, such as the 2008 box office receipts graph. And even as I write this, I ran across this tweet from @MacDivaONA (Chrys Wu) pointing me to a graph using parallelism to relate the years of time travel in various movies. I spent 20 minutes looking at it, and I know I still haven’t grasped half the nuances. That’s the point. It is, as Wu writes, a “datavizgasm.”
Now, that last piece may not be telling a news story, per se, but I would argue that journalism means conveying information, and if it captivates someone’s interest, it does tell a story. What’s the earliest year that has been time traveled to in a movie? The latest? Do paths of time traveling characters ever intersect? (When it comes to asking if something’s journalism, I would come down with data guru extraordinaire Adrian Holovaty. When asked the same question re: publishing databases online, such as his Everyblock, Holovaty said, “Who cares?”)
For this last graph, the makers have cataloged their creation process (thanks!) at Information is Beautiful. It’s informative to see the iterations the graph goes through. Parallelism as a graph form does pack a punch. But to do it right, it can’t be too confusing, and there must be a strong Data-Ink ratio, to use a Tufte term. All the lines drawn on the page, or as many as possible, should have to do with data. Creating content concisely is important whether you’re writing, editing, designing or coding. And that sentence, right there, is a parallel structure.