Data Visualization: Keeping the Story Straight

Not long ago, BusinessWeek published a story titled “Data Visualization: Stories for the Information Age” by Maria Popova (self-described as a “digital anthropologist, cultural curator and semi-secret geek aggregating the world’s eclectic interestingness”). The article featured the work of Aaron Koblin of Google’s Creative Labs (self-described as an “artist/designer/programmer”). Popova and Koblin bring fresh perspectives to data visualization, but they are newcomers to the field and have both made statements about it that demonstrate their lack of experience.

Popova describes the field as follows:

Data visualization has nothing to do with pie charts and bar graphs. And it’s only marginally related to “infographics,” information design that tends to be about objectivity and clarification. Such representations simply offer another iteration of the data—restating it visually and making it easier to digest. Data visualization, on the other hand, is an interpretation, a different way to look at and think about data that often exposes complex patterns or correlations.

“Has nothing to do with pie charts and bar graphs”? I would gladly support any effort to dismiss pie charts (with a few exceptions), but the notion that bar graphs and other traditional displays of quantitative data have nothing to do with data visualization is just plain silly. No one who understands data visualization and has done any work in the field would make such a statement, nor would they go on to say that unlike quantitative graphs, “data visualization…is an interpretation, a different way to look at and think about data that often exposes complex pattern or correlations.” In truth, all visual representations of data are interpretations. The very act of selecting information and presenting it in a particular way is an act of interpretation. All forms of data visualization—whether traditional bar graphs or some of the newer animated displays—should be embraced if they bring data alive in clear, simple, and accurate ways to help us understand the stories that live therein. Let’s not be tempted to dismiss the tried and true in our excitement over the novel—or what gives the impression of being novel.

Moving on to Koblin, much of data visualization that is novel technologically works on the same principles as bar graphs, scatterplots, and line graphs. For example, Koblin created an animated display of SMS messages throughout a single day in Amsterdam. The degree to which it provides insight into this information relies on the same basic mechanism that causes bar graphs to work: using objects of varying heights to display differences in quantitative values.

As an artist and technologist, Koblin has done some interesting work. I especially like his animated visualization of worldwide air traffic throughout the course of a day. It tells the high-level story of daily air traffic and works as an effective starting point for further exploration and analysis, which would then require more conventional visualizations. At Google, Koblin has an enviable opportunity to play with data. He has produced several visualizations that are fun and useful for artistic purposes, but fewer that actually present information in meaningful and useful ways. Of the three terms that he uses to describe himself—artist/designer/programmer—self-expression as an artist and programmer appears to come through more frequently and strongly in Koblin’s work than the solutions of a designer. I certainly can’t fault him for the wonderful examples of self-expression that he’s created, but I feel that I must take issue with something he’s said about data visualization: “It’s not about clarifying data…it’s about contextualizing it.” Actually, it’s about both. Without clarity, which is sometimes lacking in Koblin’s visualizations, context can only take us so far.

In spite of its problems, I love the way that Popova concludes her article, with one minor exception:

Ultimately, data visualization is more than complex software or the prettying up of spreadsheets. It’s not innovation for the sake of innovation. It’s about the most ancient of social rituals: storytelling. It’s about telling the story locked in the data differently, more engagingly, in a way that draws us in, makes our eyes open a little wider and our jaw drop ever so slightly. And as we process it, it can sometimes change our perspective altogether.

I balk only at her statement that “it’s about telling the story locked in the data differently.” Differently than other forms of visual display that have worked in the past—like bar graphs, perhaps? To stress that the story must be told differently is to shift our assessment of what works and what’s needed to “innovation for the sake of innovation” to a degree that Popova herself warns against. As artists, programmers, and people from other disciplines and perspectives venture into the realm of data visualization, they will help us expand our horizons. Some will also be too quick to dismiss what they don’t understand as they approach data visualization, in Popova’s words, as “a new frontier of self-expression.” Those of us who have worked in the field for awhile must be open to new ideas; those who are newcomers must respect the work of those who have worked hard to make data visualization useful and effective. I welcome this collaboration.

People sometimes adopt existing terms and give them new meanings to suit their interests, leading to confusion—sometimes usefully, when things need to be shaken up, and sometimes not. Perhaps Koblin and others like him have done this with the term “data visualization” and are using it to describe something related to but in many ways different from data visualization as I know and practice it. I suspect that another term—perhaps “information art”—would better describe what Koblin and others who combine their artistic and technological interests are introducing. Manuel Lima, of the site Visual Complexity, recently made this point in a blog post titled “Information Visualization Manifesto.” Lima proceeded from there to thoughtfully and eloquently describe the characteristics that define data visualization—or more specifically “information visualization”—as distinct from other types of information display. His words are brilliant and timely. I heartily recommend that you read them.

Take care,

8 Comments on “Data Visualization: Keeping the Story Straight”


By Michael. September 14th, 2009 at 3:01 pm

Koblin didn’t say that “about data visualization” he said it about his art - which is totally applicable. I don’t think he’s trying to dismiss the value of traditional visualization he’s just trying to do something more interesting from a cultural/social perspective.

By Stephen Few. September 14th, 2009 at 3:11 pm

Michael,

The article suggests that Koblin made the statement about “data visualization.” Here’s the complete passage from the article:

“Data visualization is a way to make sense of the ever-increasing stream of information with which we’re bombarded and provides a creative antidote to the “analysis paralysis” that can result from the burden of processing such a large volume of information. ‘It’s not about clarifying data,’ says Koblin. ‘It’s about contextualizing it.’”

Did Popova take Koblin’s statement about contextualizing out of context?

By Dan Murray. September 15th, 2009 at 4:47 am

To me, data visualization is about:

1. Relevance
2. Simplicity
3. Context

These qualities when combined in appropriate ways yield greater understanding more quickly. Tools that enhance the novice’s ability to analyze information are valuable now and will be more so in the future.

By jerome cukier. September 15th, 2009 at 4:49 am

there are many people who could benefit from data visualization in ways they don’t even begin to suspect. but the lack of awareness of infovis is really a problem. If there were more demand for good visualizations, corporate america would be so much better off. (among others).
so while I agree with the points you raise, I welcome all articles that bring attention to the notion of data visualization, even if what they call “data visualization” is somewhat different from the definition its practitioners would give.

By Benjamin Wiederkehr. September 17th, 2009 at 6:43 am

I struggled with the definition of data visualization and information visualization for some time now. After recently reading “Now you see it” I follow the definition that information visualization together with scientific visualization are both parts of data visualization. Thus in my point of view, the discussion in the article are more focused on information visualization as on the field of data visualization.

Maybe using this differentiation it is easier to adequately place the excellent work of Aaron and the likes.

By chris. September 18th, 2009 at 6:15 pm

I’m new to data visualisation and I know its easy to agree, but… pie charts, bar charts are as you say stephen, interpretations of data. they made selections on data to tell/show, are selections not interpretations? otherwise your telling/showing everything.

i can see what popova is getting at… (graphic designer here), she was wrong to dismiss the bar charts, pie charts as not interpretations… and as you’re right stephen collaboration with newcomers is essential, but when she says differently tell the story in data, like the work of koblin (with his house of cards and others mentioned), we all want to make it more engaging, jaw dropping, further innovative in style and interpretations.

maybe yes i agree many of koblin’s examples dont necessarily communicate explicit interpretations and are probably information art, giving more awareness insights into data such as the air traffic, rather than definitive decisions/choices analytical insights.

maybe its like in graphic design we search for the latest trend to make us (audience) engage and drop jaws, such as is done in typography the revival of flourishes/swashes. We want to give data a visual representation that balances form (swashes, bar graphs, pie charts, aesthetic) and function (insights, comparisons). Sometimes i think visualisers need to explicitly define its category, which is strange.. to me as a graphic designer, as (maybe its just me) our work fields can mix quite well i.e. branding into illustrative swashes, data visualistion into illustration (again trying to balance aesthetic/form and function). I would probably care to add, easily, having a solid basis of understanding what has gone before will warrant an appreciation of not to abandon this, or tweak to fit interpretations/defintions too much.

i think the innovation can arise from this crossing of boundaries but people should try not dismiss the knowledge of those other boundaries, Info Art into Data Vis (and vice versa) and explicate these crossings. If we dont have innovation for innovations sake, how can we progress, as an inherently visual person, i love this field for its volume of innovative styles/forms and its challenge to visualise data sets, so definetly pro adaption & development.

well hope i make sense and i am useful. as you can see your post got my juices going if they might not be all be clear and must stop… thanks stephen and all featured/commented. forgive i have proof checked a few times (its late, let me off).

By Maria Popova. April 16th, 2010 at 10:23 am

Stephen,

Insightful article. I see your point about pie charts and bar graphs *being* data visualization which, taken out of the context of the original article, is absolutely right. But that’s precisely the point – it’s out of context. Please bear in mind that the main purpose of the BusinessWeek article was to explore data visualization as an emerging creative discipline and art form, not to pontificate on the scientific merits of an age-old discipline. Hence the slideshow of 21 artists working in it.

And semantic disagreements aside (I don’t think I used the word “differently” in the closing the way your read into it), I wholeheartedly agree with you about embracing collaboration. But isn’t collaboration – both artistically and just socially, in interacting with others – the most valuable stage for expressing ourselves and relating to the world and others?

Cheers.

By Stephen Few. April 16th, 2010 at 11:07 am

Maria,

I did not take your description of “data visualization” out of context. In any context, your description was inaccurate, uninformed, and misleading in a way that creates confusion. Data visualization is an established and evolving medium of expression. Much of the “emerging creative discipline and art form” that you call data visualization is something entirely different. What you described might be “data art”, but that is quite different from data visualization and shouldn’t be confused with it. Data visualization helps people explore, make sense of, and communicate data. It gives the data a clear voice to inform us. Much of what you featured in your article fails to inform. At best, it entertains. That might be useful, but it isn’t data visualization.

Semantics matter. As someone who works with people every day to help them make sense of data and communicate its message effectively to others in an effort to make better decisions, I care deeply about the distinction between data visualization that effectively informs and that which calls itself data visualization but fails to inform. This distinction might not matter to you as a “digital anthropologist, cultural curator and semi-secret geek aggregating the world’s eclectic interestingness,” but it matters to people like me who work hard to use information effectively.