Data viz

It seems to me that the main issue we’re grappling with in this week’s readings is that of clarity in visual communication. In this respect, Tufte and Drucker lie at opposite poles: Tufte’s work is focused on how to achieve clarity, while Drucker’s is concerned with problematizing it.

At the heart of Tufte’s book is a concern with presenting complex information on a two-dimensional page—“flatland”—in a manner that we can comprehend it: “Escaping this flatland is the essential task of envisioning information—for all the interesting worlds (pysical, biological, imaginary, human) that we seek to understand are inevitably and happily multivariate in nature” (12; Tufte’s emphasis). This is a concern which should seem familiar to a historian: the information we work with is always “happily multivariate,” so much so that we don’t often even think of it as reducible to variables at all.

While many of the examples Tufte uses may seem distant from humanistic topics, several are quite relevant. One example is the maps of Tang, Song, Ming, and Qing dynasty poets, marked with circles scaled to the number of poets hailing from a particular locale (74–75). Two things are noteworthy: first, they show quite clearly a southern drift in the “center of gravity” of Chinese cultural production; second, with a brief caption, these Chinese-language visualizations are entirely comprehensible to a viewer with no ability to decipher Chinese writing. The data visualization transcends linguistic barriers, which is no small thing.

On the other hand, Drucker’s piece argues that standard techniques of data visualization are fundamentally flawed, because they purport to describe an objective reality that she denies—and even if we accept “an observer-independent reality available to description” (¶2), she argues that clear visual presentation falsely ascribes certainty to ambigous data (or “capta,” in her preferred terminology). There is some merit to interrogating more closely what data visualizations are showing us: how are we to interpret, for example, the southward shift we see during the Song dynasty in the maps Tufte includes? The dates given (969–1279) span both the Northern Song and the Southern Song; the obvious question is whether the distribution is different before and after the Jurchen’s conquest of northern China in 1127. Was the shift already in progress in 1127? Or is what we see the result of an extinction of poetic production in the Jurchen-held areas after 1127?

This is not an issue that brings into question the utility of the map form, though; it simply calls for disaggregating the Northern and Southern Song dynasties if we want insight into this question. Drucker might ask other questions, though: How are we defining “poet,” for example? Or, how reliable is our information on these poets’ birthplaces? These are entirely legitimate questions to ask—and ones that cannot easily be answered in clean and tidy data visualization. Unfortunately, Drucker’s ideas for how we might produce “humanistic” visualizations are not especially successful; or, rather, they succeed only in illustrating uncertainty and ambiguity, at the expense of communicating any useful information.

In the end, I suppose the lesson should be that we must not ask too much of our visualizations. As producers, we should accompany them with textual explanations of their underlying data and assumptions; as consumers, we should demand the same.

About

I’m a first-year Ph.D. student in Northeastern’s world history program. Coming from a background is nineteenth-century French history, I’m moving toward comparative nineteenth-century European imperialism with a particular interest in the effects of modern transport and communication technologies.

Posted in Uncategorized