The J. Paul Getty Trust 2014 Report
"Digital" Art History
Breslauer Professor of Bibliographical Studies in GSEIS at the University of California, Los Angeles
The phrase "digital art history" is as misleading as it is awkward, suggesting that the encounter with computational technology has produced a distinct subfield in art-historical studies. Though digitization has had profound effects on the way we do our business, it remains to be seen if it has changed the field. For all intents and purposes, most of us (at least in the privileged, wired world) do our daily work digitally-accessing images, documents, research materials, and references online. The intellectual dimensions of this transformation get subsumed in our familiar habits. Digital techniques have been with us for several decades, with electronic records management and text analysis dating back years before the appearance of the World Wide Web's graphical user interfaces in the 1990s. As computational capacity and networked bandwidth increased, research and teaching have been affected by practices such as digitization, data creation, analysis, augmentation, and reception. We can point to similar impacts across the humanities, sciences, and social sciences, but it bears keeping in mind that artworks pose specific challenges as well as opportunities. Since we are already several decades into the encounter with the digital and computational, an assessment might be of use in directing future work, projects, and institutional investments.
Digital technologies extend traditional approaches to art history by increasing the scale and scope of available materials, the capacity to aggregate collections and their records, and the ability to cross-reference information about works of art. Making vast amounts of the cultural record accessible exposes materials never integrated into canonical narratives or given the spotlight in collections or exhibitions. Combined with creating online presences for once-marginalized communities and geographies of practice, this change in scope alters the corpus that constitutes the field. As our sense of masterworks and mainstreams changes, the assessment of the social production of value might alter considerably, as will structures of authority, gate-keeping, legitimacy, and other features of our discipline. While all of this might be described as a change in infrastructure—the management of resources and means for their use—the result supports a shift from dominant grand narratives into a multiplicity of histories of art.
Such changes combine technical innovation with intellectual implications. Every decision—even basic digitization of visual materials—has consequences that cause us to revisit basic assumptions. What is an image? Is it defined by its medium? By the symbolic or economic worth of its materials, by formal values such as color and composition, reception history, through the history of production, or by its model of space, volume, point of view, and/or its role in power relations? What features can be digitized or called to attention in the digitization process? What can be encoded? What can be extracted? Digitization is data creation, and the files become the objects on which processing, analysis, and other computational techniques act. Seeing art-historical objects anew through the ways digital protocols encode their information—a problem much more difficult than remediating texts through strings of keyboard strokes—creates an opportunity to reflect on the very essence of pictorial, sculptural, and spatial forms, and the historicity of vision as a social and cultural act.
Once they are digital, files of objects can be analyzed with computational processes. These include techniques of data mining, image analysis, digital forensics (combining material sciences and computation), and the use of visualization tools to examine quantitative information extracted from images and documents. The possibilities are staggering. But should we be skeptical about the ways these techniques are being applied to art history? After all, art-historical objects are not fundamentally statistical, and the interpretative approaches of humanists are, by definition and design, situated, embodied, culturally and historically specific, eschewing the very claims to universal rules and repeatable results that are central to empirical ones. One challenge is to create analytic tools that embody the interpretive principles of art history. Alternatively we can focus our critical skepticism on the character and quality of digital files, data, and other surrogates on which computational techniques depend.
What does it mean to talk about art-historical data? Works that can be mapped, measured, or modeled in the history of art, such as gothic cathedrals, archaeological remains, or networks of communication and trade can have these properties recorded as metrics. These apparently empirical acts are in fact acts of interpretation in which some information about an object has been included, other aspects excluded. A cathedral is not equal to the list of its measurements, nor a historical location simply reducible to a point on a map, particularly one supplied by Google. Likewise, the transfer of information in the documentary record (auction catalogues and dealer account books) creates a quantitative translation for analysis, but historical information is rarely clearly structured or consistent, so much modification and translation takes place in the process. Such projects extract information selectively, abstract it into quantifiable units, and, as in the case of digitized images, turn the surrogates into the object of study and analysis.
Increasingly, the analysis of images as data has become possible, but the computation processes only the digital files and should not be confused with the analysis of the original objects. In the analog world, the tonal values or color range of any analog artifact will shift depending on light conditions, circumstances of viewing, and other aspects of its situation. Works of art are phenomenological objects, subject to change. Though a digital file will be displayed differently on individual browsers, a work of art and its digital surrogate have very different properties, introducing caveats and qualifications that should not be ignored in any serious scholarship.
Increased computational capacities have added other possibilities to the inventory of digital techniques in the last decade or so. These include virtual rendering and conservation done only in simulated images and rendering engines onscreen, material sciences work and its connection to forensics, modeling complex social systems, dealing with networks and the study of social relations, and creating augmented reality applications that deliver situated simulations on mobile platforms and devices. Each of these requires considerable technical sophistication—and hopefully, equally sophisticated critical skepticism to evaluate the nature of the translations and interpretative dimensions that are part of the radical remediation of images and artworks into digital files.
Another substantive change, and new source of potential research, is the ability to collect viewers' experiences in an unprecedented way. The changes in reception studies will no doubt only be fully realized ahead, as indeed, changes in studies of production will be altered by the ways analysis of economic markets, materials and their use, circulation, and value.
But the single most substantial shift in research methods across the humanities, art history included, might be the intellectual and critical engagement with structured data. The notion that one's research is organized by structured categories, classification schemes, and that these are modeled according to a research question but also provide a way to assess the problems and limits of research, is a powerful one. The task of making explicit the many facets of a complex argument in order to marshal one's research into a data structure that can be searched and queried, while an artifice of discipline to some degree, is an exercise that forces research to be clearly organized, its armature exposed.
Do new research questions emerge? Yes and no. To some extent the digitization of the cultural record, art historical or other, extends critical inquiry through familiar approaches such as social art history, where quantitative information can be brought to bear on the study of art markets. Stylistic analysis (or stylometrics) and techniques of digital forensics combine material sciences, formal analysis, and computational capacity. Work in cultural analytics hints at possibilities ahead for the computational assessment of features in images (tone, scale, value, composition, etc.) that might prove useful for study across millions of images. Techniques of distant reading, as they are termed in the literary humanities, tend to signal patterns of readership, markets, publication history, or citation that would be difficult, if not impossible, to detect using traditional methods. The shift in the scale of what can be processed shifts the questions that can be asked. Concerns that such techniques would displace or replace the closer reading of aesthetic works seem alarmist, since these distant readings often point to areas of the cultural record that have been ignored or whose quality and character reward close engagement once their existence is made more clear.
We imagine, mistakenly, that much of the world's art is available online. Not only is this not true, but it is unlikely to be true in the near future for technical as well as cultural reasons (scale, scope, intellectual property, and the disadvantaged condition of many communities). Even if that (imperialist) goal could be realized, what would the boundary of that project be? Would the art-historical corpus include clown paintings? Travel sketches? Engineering diagrams? Graphic arts? Textile designs? Mass circulation magazines for teens? Every shard of every pot ever unearthed from an archaeological dig? The range will always exceed the grasp unless the digital map extends to cover the entire analog earth and its many layers and conditions of viewing history. The totalizing impulse is doomed to failure. Images will always proliferate faster than the inventory of art history can manage. The vision of unlimited access, of fully integrated networked infrastructure for cultural heritage, is fraught with questions of power, authority, hierarchy, and control. And questions of influence and style, as well as those of identity and community, require more nimble and flexible interpretative frameworks of analysis, not just a vast catalog of searchable objects presented as if they exist in isolation.
Many challenges remain. The apparatus of digital scholarship needs to be capable of more subtle arguments. Authoring platforms need to be able to integrate analytic tools with writing spaces. The federation of large-scale and less-populated repositories is a challenge, as is finding repositories and being able to search their contents in a unified query. Establishing stable publishing platforms and finding ways to sustain the use of intellectual content in a long term way will also take time, since the migration of files forward will depend on many interrelated institutional, political, technical, and financial elements. Interest in publication formats and in creating authoring environments specific to the compare, contrast, and formal analysis of objects has grown, and with it, debates and discussions about how to address intellectual property and copyright issues.
These are all matters that will require resources, but also, committed communities of practice and consensus. Conventions have to emerge around all aspects of networked and onscreen argument, augmented and virtual spaces, and collaborative, even communal modes of authorship and scholarship. Another challenge, intellectual as much as technical, is to design an information architecture that exposes cultural differences. Historical systems of knowledge organization have shaped art history as much as they have shaped other disciplines. Differently ordered and structured cosmologies, classification, and nomenclature systems are embedded in our institutional records and approaches to the field. The colonization of knowledge in digital environments risks repeating the worst kind of imperial moves if we do not attend to the ways data structures, metadata, and the very systems on which we depend for integration and interoperability impose rules of law in the name of convenience and efficiency. The challenge is to incorporate difference—not as a value within established order—but as the value that creates that order. A single concept, like artistic authorship, is not a universal given, but a cultural one, loaded with baggage and assumptions. Information structures designed to incorporate multiple perspectives could change the ways such assumptions are deployed.
The final challenge is pedagogical as much as technical. To date, the ability to think fluently in terms of data structures, narrative, the rhetorical force of images and simulated spaces is not something that has been part of the art historian's training. Understanding digitization at the level of file formats and production, data structures and their design and use, standards for surrogate creation, and the classification and ordering schemes through which access is mediated are all critical skills for twenty-first century art historians. So are the techniques for mapping, visualization, and analysis, for modeling, data mining, and simulation. Not all historians of art will want to engage with the specific kinds of analysis fostered by digital means any more than they embraced statistical, formal, critical, or theoretical approaches uniformly. But what is certain is that computational techniques and digital platforms will be fully integral to art history ahead. Knowing how such techniques work and what they do, how they operate, and what the intellectual implications of the shaping rhetorics of technology means might be, will be an essential tool of critical understanding and intellectual inquiry as well as of scholarly production.
An endless search for the new and novel easily blinds us to the extent to which the subtler and more complete transformation of our daily work has been effected by the coming of digital techniques and networked platforms. Leadership is crucial. Taking seriously the advocacy role that major institutions, like the Getty, can play in pushing for open content, fair use, broadly accessible ways of publishing and vetting the best scholarship in the field, and creating pedagogical opportunities that create cross-cultural dialogues from diverse communities are all essential. Intellectual vision has to lead the technological implementation towards solid, long-term platforms, rather than frivolous projects with trick effects. Innovation for its own sake is pointless, but so is resistance to transformation and change. The prospect of a fully networked cultural infrastructure will require skilled diplomacy and intellectual imagination as well as considerable underwriting if it is to work effectively as an enduring resource for the common good.