Alumnus Kirk Goldsberry Stresses the Importance of Data Visualization


Alumnus Kirk Goldsberry (PhD 2007) is a visiting scholar at the Harvard Center for Geographic Analysis and an assistant professor of Geography at Michigan State University. He is also the creator of many analytical tools that emphasize data visualization, including a “CourtVision” series of maps that chart scoring data for NBA basketball players (see the June 12, 2012 article, “Alumnus Kirk Goldsberry Featured in New York Times – Again!”). The following is an article he wrote for the Harvard Business Review on September 30, 2013, titled “The Importance of Spatial Thinking Now”:

In its 375 years, Harvard has only ever eliminated one entire academic program. If you had to guess, what program do you think that was and when was it killed off? The answer: Harvard eradicated its Geography Department in the 1940s, and many universities followed suit.

The timing couldn’t have been worse, really. Shortly after the elimination of Geography here at Harvard, the discipline underwent a quantitative and computational revolution that eventually produced innovations like Google Maps and global positioning systems, to name just two. Seventy years later we are paying for a prolonged lack of spatial thinking at American universities. There are too few classes that enable learners to improve their spatial reasoning abilities, with maps and visualizations being of course the most central artifacts to such improvements. The problem is simple: not enough people know how to make maps or handle spatial data sets.

In the meantime, spatial thinking, visualization, contemporary cartography, and the other core competencies of geographic education have never been more relevant or necessary. As this forum has made clear, data visualization is an emerging, important discipline, and spatial thinking—geography—is a fundamental skill for good data visualization.

When talking about data visualization, many begin with the assumption that it’s a new thing, freshly formed in this big data era. Visualization is not new, and it’s much older than the “Napoleon’s March” example cited by Edward Tufte as the best information graphic. For centuries, people have measured and mapped out worldly phenomena. We were collecting and mapping information long before the printing press. Libraries supply us with limitless evidence of visualization masterpieces that predate any automated computation, let alone big data, like Gerardus Mercator’s revolutionary map of the world in 1569.

That’s not to say nothing’s new about this moment in time. What is new is the recent integration of spatial thinking and computing. The current rise of what I prefer to call computational visualization is an obvious and logical extension of human practices that are as old as lines in the sand. But this idea that visualization is new hinders teaching and learning about the act of visualization. Without the proper context, “dataviz” discussions and “data science” curricula neglect the important lessons and huge contributions from the past, contributions that can inform everything from design principles to teaching and learning.

As I look out on the world of data visualization, I see a lot of reinventing of the wheel precisely because so many young, talented visualizers lack geographical training. Those interested in a 21st century career in visualization can definitely learn a lot from 20th century geographers like Jacques Bertin, Terry Slocum, and Cynthia Brewer, and they will identify pre-existing principles, cognate scholarship, and countless masterpieces that are extremely useful guides.

Which brings us back to the sheer lack of geographical training available. Recommitting to a geography curriculum in both our high schools and universities will be crucial to effectively developing a generation of great data visualizers who can tackle our challenges. Quantitative spatial analytics offer vital insights into the world’s most important domains including public health, the environment, the global economy, and warfare.

Without geography—or any teaching that emphasizes spatial thinking—the focus will remain on the data, and that’s a mistake. Yes, data are undeniably important but they are not holy. Data are middlemen. Even the term “data visualization” overemphasizes the role of the middleman, and mischaracterizes the objective of the activity. Nobody wants to see data; nobody learns from that. The best visualizations never celebrate the data; instead they make us learn about worldly phenomena and forget about the data. After all, who looks at the Mona Lisa to think about the paints?

Editor’s note: Many thanks to Zia Salim, a doctoral candidate in the SDSU/UCSB Joint Doctoral Program, for suggesting this article.

Addendum: Don Janell, Researcher and Program Director for the UCSB Center for Spatial Studies, just wrote to say, “I should note that Kirk is no longer at Harvard and has resigned his position at Michigan State. He has signed up with one of the sports channels (not sure which one — there was a bidding war) and has a major presence in the sports blog world.” You can check out some of Kirk’s most recent sports applications of data visualization in his blog on www.grantland.com.

Image 1 for article titled "Alumnus Kirk Goldsberry Stresses the Importance of Data Visualization"
Dr. Kirk Goldsberry: “My research focuses on the visual dimensions of scientific communication. I’m particularly interested in the links between visual form, graphic design, and spatial reasoning. This avenue of research is significantly influenced by the principles of cartography, visualization, cognitive psychology, vision science, spatial analysis, and human computer interaction. The tie that binds all of my research together is the unmatched ability of graphics to simplify and summarize complex spatial narratives. My courses aim to enable students to harness the power of graphic communication by understanding fundamental concepts as well as learning contemporary design techniques” (photo by Mark Fleming, Boston Daily).

Image 2 for article titled "Alumnus Kirk Goldsberry Stresses the Importance of Data Visualization"
The Mercator world map of 1569 is titled Nova et Aucta Orbis Terrae Descriptio ad Usum Navigantium Emendate Accommodata (“New and more complete representation of the terrestrial globe properly adapted for use in navigation”). The title shows that Gerardus Mercator aimed to present contemporary knowledge of the geography of the world and at the same time ‘correct’ the chart to be more useful to sailors. This ‘correction’, whereby constant bearing sailing courses on the sphere (rhumb lines) are mapped to straight lines on the plane map, characterizes the Mercator projection. While the map’s geography has been superseded by modern knowledge, its projection proved to be one of the most significant advances in the history of cartography, inspiring map historian Nordenskiöld to write “The master of Rupelmonde stands unsurpassed in the history of cartography since the time of Ptolemy.” The projection heralded a new era in the evolution of navigation maps and charts and it is still their basis (Wikipedia: Mercator 1569 world map)

Image 3 for article titled "Alumnus Kirk Goldsberry Stresses the Importance of Data Visualization"
Front page of The New York Times, June 12, 2012, featuring Kirk’s analytical visualization of NBA shooting patterns in an article titled “The Hottest Spots for Their Shots”

Image 4 for article titled "Alumnus Kirk Goldsberry Stresses the Importance of Data Visualization"
One of Kirk’s latest sports graphics: “You can see that Tom Brady and Andrew Luck each appear three times. Last season, Brady loved peppering the left side of the field with short passes, but it remains to be seen whether he can be as successful doing that without Wes Welker and Aaron Hernandez. Luck, on the other hand, loves the longer throws, even though his success rate in these areas is below league average. As the NFL enters its own version of the big data era, we should expect to understand the game better. New data and emerging forms of analytics can only stand to improve the characterization of NFL performance.”

Please follow and like us: