Data Visualization is NOT Useful. It’s Indispensable.

(This blog post is the result of a talk a gave a few days ago at Visualizing Europe. You can give a look to the program, you can give a look to my slides directly, or you can see a preview of the pictures taken during the event. Also, this is a quite long post, I am sorry. But I really needed this space to turn my slides into a blog post. If you don’t have time to read it now keep it for later.)

Background

I just came back from Visualizing Europe, a great event organized by Visualizing.org in Brussels to gather several key data visualization actors in the European scene and discuss the state of the art and potential future of visualization.

I had the honor to be invited in the panel session organized by Andrew Vande Moere to discuss the power and potential of visualization together with a bunch or great data visualization designers like Santiago Ortiz, Moritz Stefaner and David McCandless.

To tell you the truth I felt a bit like being the black sheep; but not with a bad feeling! I was the only one in the panel with a core CS visualization research background and the audience too was not to “researchy” as the usual audience I am used to talk to. What a wonderful occasion I thought!

So, when I got the invitation I started thinking about a talk that could send a strong message and stick into people’s mind. And suddenly I recalled I blog post I wanted to write since a long time about the questioned usefulness of information visualization; a complex that afflicts many of us.

I was originally inspired by a message I received from Jorge Camoes (ExcelCharts) in which he pointed me to a couple of interesting blog posts on visualization usefulness.

The first one is from Jorge and is titled: “Is Data Visualization Useful? You’ll Have to Prove It“. In this post he argues that believing that data visualization is useful is an act of faith and that we have no scientific evidence of its usefulness. Along similar lines the second post is from Stephen Few who in his blog post “True Stories about the Benefits of Data Visualization” confesses he has a hard time convincing people of visualization’s usefulness.

After reflecting a little bit about their content and their frustration I recognized the same frustration in me and decided it was time to react. And my reaction, I decided, had to be a strong one. I realized in fact that the whole problem of visualization usefulness can be removed if we change our mindset. I realized that one of the reason why we ask ourselves whether visualization is useful or not is because we don’t have a clear focus on those problems in which visualization is not just useful, but plainly indispensable.

In the rest of the post I’ll try to articulate this idea further, with the hope I will convince you it is important to change our mindset and focus more on problems where the net gain of introducing visualization is extremely high. But let me start by providing a few examples of problems I have personally experienced where I think visualization is clearly indispensable.

Examples of indispensable visualizations

In the following I provide three example of indispensable visualizations. I hope this will convince you that when visualization is indispensable it is evident.

Example #1: Bart and his moving animals
Bart is a biologist and he studies animals. More precisely he wants to understand how animals move and how contextual factors influence their behavior. Think birds migrating from one place to another exploiting the right wind at the right moment or lions searching for a prey and trying to optimize their energy expenditure.

Problem is that animal movement analysis today is a very technological endeavor. Animals have GPS and other sensors attached and can be tracked continuously around the world and generate enormous amounts of data. Plus, this data can be enhanced with contextual factors like weather conditions, making the whole thing more difficult. How do you deal with that? Can you do it without visualization? The problem is that not only they want to test some hypotheses and assumptions they have, but they also want to be able to come up with new ideas.

My colleague and dear friend Florian Mansmann (together with his valuable students) helps Bart and his colleagues understanding how these animals move, and developed a nice tool to visualize traces in time and to correlate them with contextual factors.

Again my question is: do you think it is possible to do that without visualization? I think this is another case of indispensable visualization.

Example #2: Joachim and his hidden molecules
Joachim is a chemist at the University of Konstanz, where I work, and his job is to study small molecules that inhibit a fundamental process in cellular division called “cytokinesis”. I am not going to give any biology lesson here, I am totally inadequate, but it suffices to know that  Joachim studies this stuff because it has implications on understanding cancer. Yes, cancer. Serious stuff.

Problem is that modern biology is done with computational tools and robots and he has to deal with tens of thousands of chemical reactions at once. He uses a technology called High-Throughput Screening by which a biological target can be tested with thousands of molecules in parallel in a matter of few hours. Isn’t it crazy?

But these data must be analyzed then and again there are no clear-cut preconceived hypotheses. Joachim has to delve through long excel spreadsheets and make sense of them. How primitive is this? And how indispensable is visualization here?

So we developed a simple tool to let him explore the association between the activity level of these molecules (how much they react to the given target) and their molecular structure. Nothing too complicated from the visualization point of view: a very interactive and flexible scatter plot with well-designed mappings.

When I showed the capabilities of visualization to him he was blown away, he couldn’t believe this was possible. Won’t you call this indispensable?

Example #3: Security officers keeping us safe
Even if we don’t notice it in our daily life, there are a number of people around us whose job is to monitor our infrastructures to keep us safe and to ensure that we have a steady delivery of fundamental services. We just started a new project here at the University of Konstanz to develop visual analytics tools to help this people monitor big amounts of information in real-time. These are for instance the guys who monitor the power grid infrastructure, a monster with millions of nodes spitting data 24hrs a day, and ensure that you can always recharge your PCs, switch your TV on, or heat up your food with your microwave. These guys do their job in control rooms like those depicted in this picture.

Of course, this data monster is by no means tractable by visualization alone, data analysis and reduction algorithm, as well as complex simulations are needed, but do you think it is possible to do without visualization here? Do you have the feeling it is conceivable to make sense of what is going on there without visualization? I don’t think so.

But that’s a handful of highly skilled people!

A common objection I get when I explain this theory is something along these lines: “ok Enrico this is true, but how many people are in the condition of desperately needing interactive visualization up to the point it is indispensable? The people you describe here are a few very skilled knowledge workers which represent a tiny proportion of the overall population! Visualization has the power to affect a much much larger group”

I agree. It’s actually true that these are a few skilled workers if we reason in terms of proportions. Nonetheless, we have to realize two things:

• The “tiny” proportion is millions of people. It’s not small! Jerome Cukier from OECD approached me after my talk and suggested: “You know what? I think you underestimated these numbers. There are 500 millions Excel users around the world!” Good point.
• The “tiny” proportion are those who work to make our life better. There is a whole army of skilled professionals who study to:
  • find solutions to cure our diseases
  • find solutions to reduce poverty
  • find solutions to make our planet cleaner and safer
  • find solutions to provide innovative and pleasurable products
  • find solutions to keep our cities safe
  • … add another thousand rows

Shouldn’t we spend some time to help these people? I think we certainly have to. If we have the tools with the potential to expand their mind and do great things we have the responsibility to do it!

Visualization use and the cartography cube

Finally, I think it’s necessary to reflect on how we are using visualization today, how we could use it in the future, and how this is related to indispensable visualization.

Fortunately, while thinking about the idea for this talk I stumbled upon a very useful diagram of visualization use. I found it in Alan MacEachren’s “How Maps Work” (by the way, this book is blowing my mind, it’s not at all only about maps and if you are very serious about visualization you should read it).

The model describes visualization use according to three main dimensions:

1) Private vs. Public: Is the visualization targeted to one person or a restricted group of people to solve a specific problem or it is more intended to address public audience?

2) Static vs. Interactive: Is the visualization a static representation or an interactive tool that people can use to explore alternative views?

3) Revealing Knowns vs. Revealing Unknowns: Is the goal to communicate a number of findings or messages to an audience or to provide people ways to explore the data and make sense of it?

My feeling is that visualization today, especially if seen under the lens of what the web and similar media offer to the average reader, is largely used as a communication tool, that is we are at the upper corned of the cube: largely public, mostly static or with little interaction, and mostly to reveal information that has been digested by someone else.

I call this use of visualization “visualization consumerism” (I hope nobody is offended by this name, it’s not my intent), that is, the visualization is prepackaged by someone in a format that can be easily and quickly digested by a large number of people. To be clear, I am not against it, as long as it is done well. Visualization Europe was full of fantastic designers and I love their work. But I think we should do more. Visualization is not only something to consume, we need to look at it in a different way.

What I propose is that we take more care of the knowledge workers I mentioned above and provide them with the indispensable tools they need. I call this: cognitive cyborgs. I think we have the great chance here to help people become cognitive cyborgs. Visualization, as I explained in an older post, has the power to make us more intelligent and we should strive to do that.

And looking at the cube I think we have to shift our attention to the opposite corner of the cube if we want to help people become cognitive cyborgs.

We have to build visualization for private use, highly interactive, to allow easy exploration and focus on revealing unknown, that is, help people generate new knowledge.

Conclusion

I will be totally candid with your here. It’s evidently a bold statement to claim that visualization is not useful but indispensable. To some extent it is provocative on purpose. And it is also dangerous to claim that visualization is indispensable, because some problems might benefit from it but still be solved without it. But I really think it is important to change our mindset and focus more on the impact we might have.

I confess I believe it’s not a totally black and white here, there are several shades of gray in between. But I see a huge opportunity cost here: the more we spend our energies on communication issues, the less we can spend on putting these tools directly in the hands of the people who need them. I see an army of incredibly skilled visualization designers that could do a lot more if only they had the opportunity to work on projects where visualization is really indispensable.

The problem is that this opportunity is not easily visible because many of the people we could empower with our tools don’t know it! That’s the real challenge.

I know, this is a very personal point of view and I am ready to accept some criticism, but that’s it. That’s the way I think and I would love to hear your opinion if you do not agree with me.

I personally don’t want to see visualization trivialized to a mere communication tool of digested data. I want to see cognitive cyborgs all over the place. They are everywhere guys! They are waiting for us but we have to chase them. Let’s make this happen, there are several opportunities at stakes here! Good luck.