Visualization and Interpretation by Johanna Drucker (Book Review)

With Visualization and Interpretation, Johanna Drucker expands upon her arguments made in her previous book, Graphesis.1 In both works, she argues that data visualization is a value-driven act of interpretation rather than a “neutral” presentation of data, and that understanding this perspective is essential to a critical study (and deployment) of data visualizations. (Indeed, Graphesis itself could have been alternately titled Visualization and Interpretation.) What Drucker adds in the latter work is a more focused analysis of these theories in and for the context of critical hermeneutics in humanistic studies and critical constructivist perspectives on the interpretive nature of knowledge. As a pragmatic accompaniment to these arguments, Drucker presents two core practical approaches to the practice of data visualization: modeling interpretation and nonrepresentational graphical approaches. Early in the book, Drucker reiterates her arguments from Graphesis: that approaches to data visualization design that position themselves as simply “presenting” data in a value-free process (such as those for which Edward Tufte argues2) obscure both the interpretive and subjective reality of data and the practice of visualization. In contrast, Drucker’s modeling interpretation approach to visualization describes the use of graphic elements and structural frameworks that explicitly require interpretation in their navigation; such uses include “contradiction, ambiguity, parallax, and point of view that are fundamentally interpretative in character” (p. 76). This approach is intended to explicitly situate the viewer not simply as the recipient of knowledge, but as an active coconstructor in the knowledge produced in the work of data visualization. Drucker describes the second of her two approaches to visualization design, nonrepresentational visualization, as a method for creating “interpretative and discursive artifact[s]” (p. 79). She likens the visualization artifact and the underlying data to a map’s relationship to territory and constructs the nonrepresentational perspective as a form of protest against approaches to visualization that set neutrality and transparency as their goal. A nonrepresentational data visualization contrasts with these approaches in that aesthetic decisions are made that do not represent or express a particular data point but instead are intended to create an argument with and through the data. Drucker gives the example of drawing a line between two data points on a chart. This aesthetic act of mark-making creates a relationship between the two data points that draws the reader’s attention to them, but the mark itself is not part of the data set and represents a clear editorial intervention on the part of the designer. The value of these approaches to humanistic work and studies arises through their use of absence, inquiry, interpretation, and the situatedness of audience and author. According to Drucker, these and other elements are core to a humanistic approach to knowledge production. The appendix of the book contains many illustrations that usefully demonstrate these newly proposed graphical approaches in a range of visualization contexts and use cases, but these visual aids are notably absent from the main text itself. This decision to separate the images from the central discussion is a strange one; at times in the text, Drucker seems to be grasping for ways to describe the approaches, when an illustration would have been more instructive. This separation of images is in stark contrast to the approach in Graphesis, which showed a playful and creative approach to enmeshing illustrative and graphic components into theoretical writing about visualization. As Drucker crosses disciplinary boundaries, moving from image analysis to textual analysis, she seeks to show the wider applications of her subject-oriented approach to knowledge production. However, there is a notable absence of non-visual media to further expand this discussion. For example, interpretation is inherent in the act of listening, in that “[l]istening discovers and generates the heard.”3 The relationship between seeing, hearing, and interpretation would seem to make the field of data sonification a prime space of investigation in this context. Much like the relationship between a map and its territory, sound is always already distinct from its source.