Tool Trouble: Thoughts on differences

As I noted a few months ago, I’ve been spending this semester preparing for my Ph.D comprehensive exams. (T-minus ten days until they start, by the way.) I tried mightily to focus myself on art historical reading alone for the past few months. In the meantime, however, I’ve lived vicariously through some of the many blogs, talks, and tweets of our stars of DH. Thus, inevitably, I’d resort to binging on PDF whitepapers and RSS feeds during lunch and coffee breaks before coming back to the (seriously dusty, you would not believe) tomes of seventeenth-century Dutch art history. I reassured myself that I’d be able to return to the DH realm in high gear come summer, finally ready to devote some necessary time and attention to the question of art history’s place in the digital humanities — an increasingly urgent question as I begin to outline the computational models I would like to employ for my dissertation.

This is a roundabout way of apologizing for the nature of this blog post, which I write in response to the latest issue of differences.¹ The editors devoted this issue to the theme “In the Shadows of the Digital Humanities”, echoing the name of a contentious MLA13 panel and incorporating papers from that panel as well as responses to it. This post has vanishingly little of the theoretical apparatuses that the differences authors have all constructed for their contributions. Coming from a period specialty that tends not to swim in the waters of critical theory quite as deeply as some other concentrations in the history of art, I admit to feeling like a complete interloper commenting on this kind of scholarship. I follow what threads I can, falling down the rabbit hole of tweets, blogs, and comments by folks like Natalia Cecire, Bethany Nowviskie, Ted Underwood, and Stephen Ramsay among many others. And as often happens, it seems so much has already been said so well by people who are so out-of-your-league smart you just want to curl up in a ball.

So, with that…

As seems to happen in almost any extended conversation about “the state” of the digital humanities, terms like “technological determinism” (Rhody, 8), “solutionism” (Kirschenbaum, 53), and “positivism” (Galloway, 109) slip into the discussion. I was deeply impressed by the subtlety with which many of these authors approached these terms, by in large deconstructing straw-man representations (or “constructs”, as Kirschenbaum aptly puts it) of a great deal of DH scholarship. By comparison, Alexander Galloway’s contribution to this issue struck an odd tone. Galloway posits a “Cybernetic Hypothesis,” suggesting that the tools of the digital humanities transcend a particular set of methodologies, instead participating in a knowledge regime (125) connected to/complicit with:

1. The larger capitalist/neo-liberal hegemony (which Galloway invokes via everyone’s favorite G-word)
2. The “deskilling” of the scholar from a person with the authority to “make claims” to a person dependent upon the outputs of devices.

As to the former, I find specific critiques of this relationship (e.g. the MOOC craze/debacle, or complicity in academic funding rhetoric) far more satisfying than the bland assertion that defines the computer and computer user as disempowered synecdoches of global capital and surveillance. As to the latter, Scott Weingart quickly noted his displeasure with this implicit equation of the quantitative with positivism. I agreed, particularly given Lisa Marie Rhody’s strident demonstration in the very same issue that it is more often the self-styled opponents of DH that claim the positivism/determinism that its practitioners take care to avoid. She rightly points out that the “Digital Humanities” so-called (not to mention its several genealogical strands dating back decades) “grew out of a thoughtful and reflective awareness of technology’s potential, as well as its dangers, and not a ‘vapid embrace of the digital.’” (Rhody, 9)

I wonder, though, at the longevity of the technological positivism canard. Why does Galloway succumb to it in an otherwise well-considered exploration of the history of technological knowledge structures? One passage in particular stood out to me (the boldface emphasis is mine):

The digital humanities assume certain things about human subjects. While seeming to embed scholars and students more firmly in data, digital tools tend to do the opposite. Highly codified interfaces reduce the spectrum of possible input to a few keywords or algorithmic parameters. Those who were formerly scholars or experts in a certain area are now recast as mere tool users beholden to the affordances of the tool — while students spend ever more time mastering menus and buttons, becoming literate in a digital device rather than a literary corpus. (127)

I wonder if the determinism/positivism trouble will persist as long we conceptualize computational approaches to humanistic subjects by using the term “tool”. In fact, I noted in my reading of this issue that the word “tool” pops up once or more in virtually every article, sometimes in such close order it feels like, well, the sound of some machine (e.g. Golumbia, 172). Yet — remarkably for a venue like differences — the term passes by all but uninterrogated.

One can hardly blame the authors, though. The damn term pops up everywhere in the DH sphere. From the Bamboo DiRT project that seeks to catalog digital “tools” for humanities scholarship, to introductory sessions on DH “tools” in almost any THATCamp, to the One Week | One Tool project, to the Chronicle’s ProfHacker blog absolutely lousy with “tools” to help your academic labor, the “tool” has become absolutely ubiquitous.

It isn’t hard to understand why. A “tool” sounds reassuringly tangible. A tool is approachable. A tool is something you can pick up, utilize, and then put down. A tool can be trusty. If a tool breaks, you can just go get a new one! It evokes the idea of expert engineering coalesced (or perhaps crafted is a better word) into an discrete object, the great benefit being not only that you don’t have to build the tool yourself, but you don’t need to understand the engineering behind its production.

I wonder, though, if the “tool” enthusiasm endemic to DH (as a side note, don’t forget Adeline Koh’s article on DH “niceness”, also in this issue) is inscribing not only a impression/construct of DH, but in fact actual modes of practice, that embrace all too closely the understanding of a tool as “a device or implement used to carry out a particular function.”

To wit: a few weeks ago I headed a very informal discussion about various DH methodologies at one of the DC area’s many great art historical research centers. I was the “DH specialist” in the room (I know, laugh if you must…), and I had been asked to come and help introduce some digital tools to a few of their scholars who were interested in applying, among other approaches, network and geographic visualization and analysis to their research questions. I hope I can say the discussion was interesting and thought-provoking, if not perhaps immediately useful to all involved. One barrier we kept running into, however, was their expectation that I would be talking about “plug-and-play” tools, when in fact I obstreperously kept asking about their research methodologies, and how exactly they were hoping to make those methodologies interact with the methodologies offered by geographic and network paradigms.

Like Galloway, these scholars may have been entirely reasonable in their expectation that the digital humanities comprised a set of tools — tools that might take time to learn, no doubt, but tools that could essentially be “used” with an extant methodology, few or no questions asked. After all, we just can’t seem to shut up about how great all these “tools” are.

As in many things, I tried to channel Elijah Meeks when he talks and writes about the ORBIS and Kindred Britain projects. He has argued that these digital constructions are “tools” only in the way that a monograph or a scholarly article are “tools”. If I understand Elijah correctly, he means that a mature work of computational scholarship in the humanities not only acknowledges the methodological assumptions underpinning its models and logic (down to level of code architecture), but actually presents a theoretical model for understanding its subject. The models underlying network structures, GIS, topic modeling through R or MALLET, etc. simply don’t provide evidence for an argument per se. Properly deployed, these models become an argument. Or, as Franco Moretti argues in “Operationalizing”: to integrate the methodologies of these analytical models with your own is to claim that your objects of research can be usefully conceptualized as such, and also requires you to acknowledge that, yes, you are in fact transforming those objects.

It is for this reason that I really would like to retire “tool” from my own DH vocabulary. It seems to me to encourage a “black box” understanding of the digital humanities methods, not to mention obsession with tying oneself to transient platforms and implementations, rather than engaging with the underlying methodologies and theories.² We end up with potentially powerful interpretive projects that release merely the first “interesting” results to pop out of their “tool” of choice. (For example, I think here of the Shakeosphere network visualization project that nods to the existence of consequences of using a network model but has not (yet!) dug into examples in their website of what these implications might actually be.)

Certainly, I fear that the predilection to “niceness” in DH may oppose dropping the “tool” concept, as it frankly raises the barrier to entry by demanding closer methodological engagement before the fun can begin. I also wonder to what extent funding sources such as the NEH, who seem to prefer multi-use platforms — a.k.a. “tools” — may oppose a retrenchment into the methodologically-tailored approach I am advocating. All the same, I think the pitfalls of the simplistic, tool-centric outlook on DH are too great to ignore.