With our group projects now underway, I was hopeful that this week’s examples and discussions of the state of digital art history would provide some crucial insight or methods, but it’s left me feeling more ambiguous than ever! On the one hand, I’ve seen examples of art history projects that make use of network analysis, visualization and reconstruction tools, text and corpora analysis to exciting ends (like Visualizing Venice or the Getty’s “Mutual Muses“), and institutions like the Frick are organizing critical discussions around issues of machine learning and computer vision. On the other hand, examples of “distant viewing” as described by Taylor Arnold and Lauren Tilton1 left me a little cold. What happens when we treat visually and thematically complex works as something to be reduced to data? What potential lies in the use of tools that sort materials on their most superficial values, or affirm existing categories and hierarchies? While my skepticism might not go so far as the position stated by Claire Bishop, her concerns of the limits of methods like ‘distant viewing’ nonetheless resonate.
“Does the data set exist in history before being sequenced digitally or is it only actualized once it has been laid out via the digital archive? Are the assembled historical ‘facts’ found or produced?”2
As applied in Arnold and Tilton, with documentary images and formulaic television series, methods like distant viewing (or distant reading) might recognize patterns, which in turn might illuminate new questions, but a “factual” observation goes only so far in explaining the much more interesting questions of why these patterns exist. What is the significance in recognizing visual similarities between artists? (And would these relationships have escaped the notice of an art historian examining the same set of images?) The authors note that their analysis of WPA photos largely illustrates relationships between artists, rather than uncovering something new; we’re given an example where the distant viewing confirms information that was already known or documented. Would this have any use in being applied to a set of lesser-known body of images, or is this a limitation of distant reading?
I’m also curious how (if ever) these kinds of methods can take into account the complexities, code switching, visual subversion, and play of creative works. (I wondered this with some of the text analysis we’ve encountered, but don’t have the lit theory foundation to ask the same questions!) Will it someday be possible to train a neural network to recognize irony? Can it recognize the difference when artists like Julian Schnabel or Sherrie Levine invoke Caravaggio or Walker Evans? Arnold and Tilton rely on images that have close object-subject relationships (a documentary image that depicts a dog, which is meant to represent a dog); can it account for more abstract qualities? (And maybe these are the wrong questions to ask, and looking for a role for computational technologies that they aren’t designed to fill…)
More troublesome, though, are the uncritical applications of certain technologies in these studies, and the assumption that biases in machine learning are simply down to “bad data”? Can we talk about the use of facial-recognition algorithms without interrogating the sources of those technologies? A few of this week’s authors hit on the ethical questions linked to new technologies (most explicitly in Johanna Drucker’s not-so-distant fantasy of the Mus@um3, and I’m looking forward to discussing it more when we get to Module 6!
I’m also still wrangling with the distinctions between a “digitized” art history and digital art history, and where these diverge or overlap as practices. As our group is working out the scope of our project and interests, it’s becoming clear how we’ll need to temper expectations given the time frame, comfort with tools, and (crucially) the availability of materials. What might a minimalist approach to digital art history look like?
![](https://dh.stephaniegrimm.com/wp-content/uploads/2020/09/lisa_lower_expectations.gif)
- Taylor Arnold and Lauren Tilton, “Distant viewing: analyzing large visual corpora,” Digital Scholarship in the Humanities 34 (2019).
- Claire Bishop, “Against Digital Art History,” International Journal for Digital Art History 3 (2018), 127.
- Johanna Drucker, “The Museum Opens,” International Journal for Digital Art History 4 (2019), 2.09)
Leave a Reply