The data visualization – data art continuum

Data visualization is becoming a catch-all concept with little analytic usefulness. The infographic plague we have to endure is not helping.

It doesn’t have to be that way. Stephen Few wrote recently about the distinction between data visualization (“the goal that data be visualized in a way that leads to understanding”) and data art (“visualizations of data that seek primarily to entertain or produce an aesthetic experience”). Robert Kosara gives data art a different meaning.

While I strongly agree with this, I am unable to see where to draw the line: there is no data visualization without graphic design and no data art without data. For the sake of the argument, let’s think of data visualization as 10% of graphic design and 90% of data and data art as 90% of graphic design and 10% of data. It’s a continuum, where the nature and the purpose change when the percentages change.

It’s fun to play with these meaningless percentages, so let’s carry on. Suppose that a typical Excel user with no graphic design skills or training becomes aware of what data visualization is all about. She understands that some basic rules can improve effectiveness and make her charts aesthetically pleasing. She’s still using traditional charts, but she’s moving the split to the 75% data / 25%  design mark. At the 80%/20% mark we can start talking about “functional aesthetics”. We haven’t left data visualization yet.

Around 60% data / 40% design, things start to change (I would place a typical New York Times visualization around here). Design can no longer improve data visualization and it slowly becomes data illustration (at 49% data / 51% design). The designer is making a unique piece and he uses a lot more textures and non-traditional charts, but he still believes (or wants us to believe) that the main goal is to inform and  entertain. I would place David McCandless around the 30% data / 70% design mark.

Finally, at the 25%/75% mark data illustration becomes data art and the graphic designer becomes an artist. He’s now free of all constraints.  The data can be the starting point, but he uses it to create a unique work of art.

Again, don’t take these percentages too seriously. I do believe that aesthetics play a major role in data visualization and it’s always there, no matter what. But if, at some point, we find ourselves trying to make out charts more beautiful and memorable than more effective, we are changing their nature and their usefulness in a corporate environment. As Stephen Few puts it, it’s “frivolous, costly, and harmful”.

Leave data art to art galleries and please, please, don’t try to be an artist with Excel and canned effects.

So, where are you in the data visualization – data art continuum?


7 thoughts on “The data visualization – data art continuum”

  1. Excellent post. There certainly is a continuum. My concern is terminology. As I commented on Stephen Few’s blog, the term data visualization means different things to different people and many include data art within data visualization. I suggested statistical graphics for the left end. The term data graphics is synonymous with statistical graphics which would allow all terms to start with data but does not make the distinction from data art as clearly.

    I would put Tableau to the left of Excel since Tableau pays much more attention to perceptual accuracy and doesn’t include pseudo three dimensional effects, ribbon graphs and other options intended to be artistic. On the other hand, a figure from Tableau is much more aesthetically pleasing than one from Excel with defaults used.

  2. Naomi: I would split data visualization (the first 40%) into statistical graphics (first 10%) where I would include Bertin, Cleveland and your own book, simple aesthetics (10%-20%) for good but typical Excel charts (with no default formatting) and functional aesthetics (20%-40%) with Tableau, because it’s easier in Tableau to find the right balance between accuracy/effectiveness and aesthetically pleasing charts. I’m excluding from this continuum really bad charts (many of the charts in the Excel chart library) and bad infographics (the link-bait variety).

    Maybe we should arrange an international summit to finally define the terms we are using in this field.

  3. You had me at “continuum” — infosthetics and I wrote a paper 5 years ago on the information aesthetic visualisation continuum between mapping focus / data focus, and the relationship between data, art, and interaction.

    Perhaps your struggle with a continuum is that is could maybe be a triangle instead? =)


  4. “I would place a typical New Your Times visualization around here”


    Should be New York Times?

Comments are closed.