Since writing the short article that appeared in this blog a week ago titled “The Incompatible Marriage of Data Visualization and VR,” I learned indirectly that my skepticism about VR’s role in data visualization was opposed by Erin Pangilinan, the chief editor of a new book about VR titled Creating Augmented & Virtual Realities: Theory & Practice for Next-Generation Computing (O’Reilly Media, 2019). Pangilinan wrote chapter 9 of the book, titled “Data and Machine Learning Visual Design and Development in Spatial Computing,” which promotes VR’s usefulness for data visualization. (Note: In case the term is unfamiliar, “spatial computing” apparently refers to VR, AR, and other related technologies). What I found when I read the chapter, however, utterly failed to make her case.

The only examples that Pangilinan includes in the chapter that are potentially useful in VR are scientific visualizations, not data visualizations. Unfortunately, she fails to make a distinction between these two fields of research and practice. Scientific visualization involves visual representations of the physical world. For example, an MRI scan of someone’s brain is an example of a scientific visualization. It graphically represents the physical structure of the brain in a way that might enable a physician or scientist to examine and understand it. Data visualization (a.k.a., information visualization), on the other hand, graphically represents abstract quantitative data. A graph that shows how sales revenues are changing over time is an example of a data visualization. These two fields overlap in many ways because they both visualize information, but they also significantly differ, and the differences cannot be ignored, certainly not when exploring the potential benefits of VR.

A clear hint regarding Pangilinan’s perspective appears early in the chapter when she writes:

Although whitepapers like the IEEE’s “Cost-benefit Analysis of Visualization in Virtual Environments (VEs)” question the relevance and purpose of visualization in XR [i.e., Extended Reality, which includes VR], asking “do we really need 3D visualization for 3D data?” Quite simply this chapter’s basis assumes from the beginning that the use of VEs enables a better understanding of 3D data, given appropriate context, thoughtful design, and development. (p. 194)

Assuming that VR enhances data visualization is not an appropriate approach—certainly not a scientific approach—to research into VR’s potential.

Here are a few examples of Pangilinan’s claims:

They [i.e., users] can better see layers underlying more complex multidimensional data within spatial computing than other mediums. (p. 196)

These technologies can enhance analytical thinking, given the emergence of computational search and artificial intelligence (AI) displaying multidimensional data that can be more easily explored with new technologies. (p. 197)

New conceptions of data are now encoded into the actual application experiences that improve the user’s interaction with their data. (p. 198)

Affordances in spatial computing allow the user in spatial computing the freedom to do more in a 3D environment, unlike 2D desktop and mobile experiences. (p. 201)

The viewer is also able to appreciate how they can control their data more intuitively and how it deepens their understanding of the substance because of the methodology of design aesthetics (graphic design) and the technology of graphic production (how it is created in spatial computing). These novel interactions, which are possible only in spatial computing, unlock new insights because of being able to view and manipulate data in 3D pace unlike previous design paradigms. (p. 202)

Spatial computing enables more mechanisms to directly manipulate data and offer spatial computing creators the ability to study new design paradigms… (p. 202)

These are bold claims, but at no point in the chapter does she explain or provide evidence for them.

Throughout the chapter, Pangilinan uses the work of Edward Tufte as a foil for her argument, and in the process she misrepresents his work and even misquotes him on several occasions. I was alerted to her misquotations by the fact that some of them are downright inarticulate. You can disagree with aspects of Tufte’s work, but you can’t reasonably accuse him of ever being inarticulate. She belittles Tufte’s position as “backward” and implies that anyone who doubts VR’s usefulness for data visualization is a Luddite who opposes human progress. Here’s a direct quote:

This type of thinking restricts our minds to being limited to backward tooling that keeps humans at a distance from technology and farther away from data that can aid humanity. Instead, we need to utilize technology to become closer to our data and solve humanity’s problems. (p. 201)

We absolutely should use technology to better leverage data for solving human problems, which is precisely why we should never just assume that a specific new technology provides benefits. When considering the potential benefits of VR for data visualization, skepticism is healthy.

Pangilinan’s case that VR is a useful environment for data visualization isn’t serious. A serious approach would begin with a thorough understanding of data visualization, which is not Pangilinan’s area of expertise, and would then proceed scientifically by designing and running experimental studies to test its usefulness. She has not proceeded in this manner, nor has she provided a single example of a data visualization in VR that suggests its usefulness. Her case is hollow. People who develop and promote technologies without first determining if they work are not doing the world a favor. They certainly aren’t using technology to help us “become closer to our data and solve humanity’s problems.”

To her credit, Pangilinan points out that showing 2-D visualizations in VR, as some have done, provides no benefits, and that poorly designed data visualizations do not suddenly become well designed merely by placing them in VR. She errs, however, in assuming that 3-D data visualizations, either on a flat screen or in VR, are inherently useful. The evidence suggests otherwise. She reasonably argues that data visualizations in VR could potentially benefit from new forms of interaction with the data that aren’t available when viewing them on a flat screen, but she never once identifies and describes an interaction that demonstrates this potential. Her entire case is based on assumptions—wishful thinking by someone who is personally invested in VR.

Anyone who genuinely wants to make a case for VR as a useful environment for data visualization must do better than this. We need to get past the cool factor and focus on practical utility. I would not oppose any serious attempt to improve data visualization through VR in ways that seem promising; I’m just not aware of any. Serious explorations of VR’s potential for data visualization would begin with a firm understanding of 1) human perception and cognition, 2) data visualization, and 3) VR, and from there ask, “Given what we know, in what specific ways might VR potentially enhance data visualization?” Given what we know about the ineffectiveness of 3-D data visualizations on a flat screen, what makes us think that viewing them in a virtual environment with the ability to navigate virtually, much like we could in the real world, would enhance the data sensemaking experience? As I explained in the previous article, I can’t think of any reason to believe that VR has anything to offer data visualization other than an expensive distraction. I would embrace any viable evidence that it works, but, so far, no such evidence has emerged. If you believe it has, please share it with me. I’ve been inviting people to provide examples for many years. Perhaps I should offer a reward. Even if you have no evidence but have a specific reason to believe that VR might enhance data visualization, make a reasonable case for that possibility.

As I see it, there’s a long list of ways in which flat-screen data visualization tools could be improved that would offer known benefits. For now, isn’t that where our focus should remain? If we really want to “become closer to our data and solve humanity’s problems,” we must focus on what actually works.