The difference between Lev Manovich’s “Data Science and Digital Art History” and Nathan Yau’s “Representing Data,” provides an interesting point for comparison.
Yau’s chapter gives widespread insights into the best practices for visualizing all types of data. Manovich explores concepts of data science that are specifically relevant to digital art history.
Both authors deal with data science, yet only Manovich’s is specifically tailored to art historical work. Manovich emphasizes that concepts used in data science, such as objects, features, data, feature space, and dimension reduction, are independently important and critical for most people, in most fields to understand: “Anybody who wants to understand how our society ‘thinks with data’ needs to understand these concepts” (Manovich 13). Yet, some of these concepts and ideas are more relevant or useful for digital humanities scholarship.
Yau includes an analysis of many types of visual cues: position, length, angle, direction, shapes, area, volume, color saturation, color hue. He references a study done at AT&T Bell Laboratories in 1985 where William Cleavland and Robert McGill how accurately people perceive visual cues from most to least accurate. The ranking order can be seen here:
I find it interesting that hue is the least accurately perceived visual cue. Yau warns against blindly following this ranking system and brings this ranking up to emphasize the need to understand how a specific visual cue or visualizing technique might be perceived by audiences. I think the perception of hue is an interesting concept to examine in the context of digital art history due to the extreme relevance of color, hue, and saturation in analyzing visual artworks.
The use of hue or saturation can be used as a tool for sorting large datasets of visual artworks. For example, the Software Studies Initiative and Gravity Lab jointly developed the “287 megapixel HIPerSpace super visualization system.” In this software one can explore a set of paintings using cultural analytics techniques by turning paintings into sets of data that can be graphed and turn those graphs into collections of paintings. The video (linked below) demonstrating this software uses the works of painter Mark Rothko. At 1:38 in the video, they take the group of Rothko’s paintings, which are organized from left to right by year created and displayed overlapping one another on this loose chronological timeline, and add a transparency layer to the visual display. This creates a blurring effect that just shows a general impression of hues across the screen. Through this software and these types of sorting we are able to see color trends over the course of Rothko’s career. This example demonstrates the importance and practicality of using hue as a visual cue for understanding sets of visual art historical data.
I am also interested in exploring Manovich’s discussion of the crucial decisions in making representations.. The first decision involves creating a boundary, such as a time period, for the phenomenon being studied (other boundaries could be countries, artists, or groups of artworks). It is within this first decision that we are already feeding software or technology information defined by human terms. The idea that a certain art historical movement is defined on either end by two specific dates (or even decades) is a boundary created by historians. I would argue that we have come to accept some of these “boundaries” as truths within the art historical world despite their human construction. Movements such as the Renaissance, impressionism, or modernism have become widely accepted movements within history with (relatively) defined time periods in which they exist. Although we largely just accept these boundaries as facts, rather than a specific perspective in which to view history, there is still room for interpretation, critique, and alternating modes of understanding movements, time periods, and human created boundaries.
I argue that these human defined boundaries become additionally obscured and hidden once mediated through technologies. Although it can be difficult, it is possible to recognize how an art historian’s writing is clearly specific to their viewpoint, knowledge, and construction.
Once a human created boundary is inputted into a software, it attaches from human construction and becomes a truth on which the computer relies to put out new data, graphs, or information. I have concerns about the ability of technologies to obfuscate and conceal these human created boundaries within digital art historical scholarship.