Shaking the “Pretty Picture” Stigma

Coming from the academic and computer science side of visualization, I always assumed that it would be self-evident to anybody that visualization is first and foremost useful, and only happens to also produce nice pictures. Alas, this is not actually the case. To most people, visualization means pretty pictures first, and maybe also a fact or two. We have to fight that or risk the trivialization and marginalization of visualization as an analytic tool.

Insight, not Pictures

Manuel Lima’s Information Visualization Manifesto has caused quite a stir, and it’s an important wake-up call. Lima’s points are excellent, and I can only recommend you read both the article and the comments. I couldn’t possibly agree more with absolutely everything he is saying.

The key problem is that people are much more interested in clicking through interesting pictures than learning about actual analysis work done using visualization. When I complained earlier about the state of visualization on the web, I thought that the problem was simply one of representation. Surely, there is a lot of great analytic work out there, we just have to write about it! But it’s not easy to find them. Stephen Few also recently asked his readers for visualization success stories, with an underwhelming response.

I noticed the same thing at the OECD Seminar a few weeks ago: the main goal seemed to be to make something colorful out of the otherwise dry numbers. Whether the visualizations were any good was secondary, and was not even a question most people there had even thought of asking.

Artistic vs. Pragmatic Visualization

Lima makes the distinction between data visualization and data art. The criteria are not perfectly clear-cut, and it’s easy to discuss a particular example ad nauseam. I did write about possible criteria a while ago though, and think that the distinction is pretty clear. What is the main question: Do you want to communicate the data or a concern? Do you care about perceptual effectiveness or beauty? Is the data a given or is showing the existence of the data part of the point?

There is no ‘right’ answer to these questions, they simply lead you in two very different directions. Some people see a clear distinction between data visualization and data art as a threat, but I don’t see why that would be the case. There is nothing wrong with data art, it’s just not pragmatic data visualization. So let’s stop calling it that.

Not Visualization, but Visual Analysis

We need a new term. Visualization has been around for too long, has too many meanings, and has been used by artists to describe many other things. Visual Analysis is perhaps a better term; it does not have that baggage and contains a key aspect of actual data visualization: analysis. It’s also less politically loaded than the closely related visual analytics.

Visual analysis is not primarily about the pictures, but about finding ways to use our powerful visual systems to analyze data. It’s analysis done in a visual way. It’s visual exploration, visual data analysis, and visual presentation of results.

Visual analysis is difficult. While we certainly want to use visual means of communication to make data accessible, that’s not all there is. Visual analysis can involve incredibly complex and difficult data and visual concepts. There is a common assumption that visualization is easy to understand, but that is only the case for some types of data and some uses.

Visual analysis is not trivial. You have seen the bar and pie charts, but do you actually know what they mean? Do you know how to use them to tease the relevant information out of your data? Can you handle more than two dimensions of data and still find meaningful structures? There is so much more to visual analysis than what Excel offers you.

These things are not just important to make us feel better. If we want to be taken seriously as a field, receive funding for research, and expect businesses to pay for what we are doing, we have to communicate this. Visual analysis is not the clipart to cover up your lack of ideas. Visual analysis is the tool that gives you insight, the microscope that lets you find the hidden structures, and the link between your data and your brain.

Comments

  1. says

    I think this post is right on the spot! It is not enought to draw pretty pictures or to design cool tools with all the bells and whistles the latest computer technologies offer.

    The crucial part is to being able to draw conclusions which are derived from insightful (sets of) graphics.

    Whereas statistics was too conservative in the past to understand the analytic power of graphical displays, computer science did ignore the data analytic aspects of visualizations for the most part.

    IMHO the successful future will be to meet part way and utilize the technologies and tools from computer science in order to get to statistical sound decisions. There is wealth on both sides which just needs to be picked up!

    Will this need a new discipline, a new buzz-word (which we somehow already have) or “just” a fruitful collaboration – being usually the hardest to achieve.

  2. says

    I know you’re biased heavily in favour of analysis, but I think in your haste to defend science and define away “pretty pictures” you understate the role that exploratative visualisation has in generating good forms for analysis. The process is the visualisation… not the answers.

    Sure, the outcome of many explorations ends with the visuals and stops short of what you call “actual” analysis but you won’t succeed in leading people to analysis by passively belittling everyone who doesn’t get that far (and that’s what you’re doing, believe me). I respect that analysis is your goal but it’s too limiting a definition for visualisation in general. It’s a different set of skills and a lot harder, but it doesn’t mean you should dismiss the learning and discovery that happens during a good exploration.

    If I was you I would be worried about succeeding the definition game. Right now you’re frustrated that too many people are calling their work visualisation but just wait until everyone hears your cries and starts calling their pretty pictures “visual analyisis” instead. You can chase definitions forever (and in academia you might be happy to do that) but at some point it’s easier to just set a good example. What good examples of visual analysis are there? (No bad examples please, I don’t care).

Leave a Reply