“Digital History” Can Never Be New

If you claim computational approaches to history (“digital history”) lets historians ask new types of questions, or that they offer new historical approaches to answering or exploring old questions, you are wrong. You’re not actually wrong, but you are institutionally wrong, which is maybe worse.

This is a problem, because rhetoric from practitioners (including me) is that we can bring some “new” to the table, and when we don’t, we’re called out for not doing so. The exchange might (but probably won’t) go like this:

Digital Historian: And this graph explains how velociraptors were of utmost importance to Victorian sensibilities.

Historian in Audience: But how is this telling us anything we haven’t already heard before? Didn’t John Hammond already make the same claim?

DH: That’s true, he did. One thing the graph shows, though, is that velicoraptors in general tend to play much more unimportant roles across hundreds of years, which lends support to the Victorian thesis.

HiA: Yes, but the generalized argument doesn’t account for cultural differences across those times, so doesn’t meaningfully contribute to this (or any other) historical conversation.


New Questions

History (like any discipline) is made of people, and those people have Ideas about what does or doesn’t count as history (well, historiography, but that’s a long word so let’s ignore it). If you ask a new type of question or use a new approach, that new thing probably doesn’t fit historians’ Ideas about proper history.

Take culturomics. They make claims like this:

The age of peak celebrity has been consistent over time: about 75 years after birth. But the other parameters have been changing. Fame comes sooner and rises faster. Between the early 19th century and the mid-20th century, the age of initial celebrity declined from 43 to 29 years, and the doubling time fell from 8.1 to 3.3 years.

Historians saw those claims and asked “so what”? It’s not interesting or relevant according to the things historians usually consider interesting or relevant, and it’s problematic in ways historians find things problematic. For example, it ignores cultural differences, does not speak to actual human experiences, and has nothing of use to say about a particular historical moment.

It’s true. Culturomics-style questions do not fit well within a humanities paradigm (incommensurable, anyone?). By the standard measuring stick of what makes a good history project, culturomics does not measure up. A new type of question requires a new measuring stick; in this case, I think a good one for culturomics-style approaches is the extent to which they bridge individual experiences with large-scale social phenomena, or how well they are able to reconcile statistical social regularities with free or contingent choice.

The point, though, is a culturomics presentation would fit few of the boxes expected at a history conference, and so would be considered a failure. Rightly so, too—it’s a bad history presentation. But what culturomics is successfully doing is asking new types of questions, whether or not historians find them legitimate or interesting. Is it good culturomics?

To put too fine a point on it, since history is often a question-driven discipline, new types of questions that are too different from previous types are no longer legitimately within the discipline of history, even if they are intrinsically about human history and do not fit in any other discipline.

What’s more, new types of questions may appear simplistic by historian’s standards, because they fail at fulfilling even the most basic criteria usually measuring historical worth. It’s worth keeping in mind that, to most of the rest of the world, our historical work often fails at meeting their criteria for worth.

New Approaches

New approaches to old questions share a similar fate, but for different reasons. That is, if they are novel, they are not interesting, and if they are interesting, they are not novel.

Traditional historical questions are, let’s face it, not particularly new. Tautologically. Some old questions in my field are: what role did now-silent voices play in constructing knowledge-making instruments in 17th century astronomy? How did scholarship become institutionalized in the 18th century? Why was Isaac Newton so annoying?

My own research is an attempt to provide a broader view of those topics (at least, the first two) using computational means. Since my topical interest has a rich tradition among historians, it’s unlikely any of my historically-focused claims (for example, that scholarly institutions were built to replace the really complicated and precarious role people played in coordinating social networks) will be without precedent.

After decades, or even centuries, of historical work in this area, there will always be examples of historians already having made my claims. My contribution is the bolstering of a particular viewpoint, the expansion of its applicability, the reframing of a discussion. Ultimately, maybe, I convince the world that certain social network conditions play an important role in allowing scholarly activity to be much more successful at its intended goals. My contribution is not, however, a claim that is wholly without precedent.

But this is a problem, since DH rhetoric, even by practitioners, can understandably lead people to expect such novelty. Historians in particular are very good at fitting old patterns to new evidence. It’s what we’re trained to do.

Any historical claim (to an acceptable question within the historical paradigm) can easily be countered with “but we already knew that”. Either the question’s been around long enough that every plausible claim has been covered, or the new evidence or theory is similar enough to something pre-existing that it can be taken as precedent.

The most masterful recent discussion of this topic was Matthew Lincoln’s Confabulation in the humanities, where he shows how easy it is to make up evidence and get historians to agree that they already knew it was true.

To put too fine a point on it, new approaches to old historical questions are destined to produce results which conform to old approaches; or if they don’t, it’s easy enough to stretch the old & new theories together until they fit. New approaches to old questions will fail at producing completely surprising results; this is a bad standard for historical projects. If a novel methodology were to create truly unrecognizable results, it is unlikely those results would be recognized as “good history” within the current paradigm. That is, historians would struggle to care.

What Is This Beast?

What is this beast we call digital history? Boundary-drawing is a tried-and-true tradition in the humanities, digital or otherwise. It’s theoretically kind of stupid but practically incredibly important, since funding decisions, tenure cases, and similar career-altering forces are at play. If digital history is a type of history, it’s fundable as such, tenurable as such; if it isn’t, it ain’t. What’s more, if what culturomics researchers are doing are also history, their already-well-funded machine can start taking slices of the sad NEH pie.

Artist's rendition of sad NEH pie. [via]
Artist’s rendition of sad NEH pie. [via]
So “what counts?” is unfortunately important to answer.

This discussion around what is “legitimate history research” is really important, but I’d like to table it for now, because it’s so often conflated with the discussion of what is “legitimate research” sans history. The former question easily overshadows the latter, since academics are mostly just schlubs trying to make a living.

For the last century or so, history and philosophy of science have been smooshed together in departments and conferences. It’s caused a lot of concern. Does history of science need philosophy of science? Does philosophy of science need history of science? What does it mean to combine the two? Is what comes out of the middle even useful?

Weirdly, the question sometimes comes down to “does history and philosophy of science even exist?”. It’s weird because people identify with that combined title, so I published a citation analysis in Erkenntnis a few years back that basically showed that, indeed, there is an area between the two communities, and indeed those people describe themselves as doing HPS, whatever that means to them.

Look! Right in the middle there, it's history and philosophy of science.
Look! Right in the middle there, it’s history and philosophy of science.

I bring this up because digital history, as many of us practice it, leaves us floating somewhere between public engagement, social science, and history. Culturomics occupies a similar interstitial space, though inching closer to social physics and complex systems.

From this vantage point, we have a couple of options. We can say digital history is just history from a slightly different angle, and try to be evaluated by standard historical measuring sticks—which would make our work easily criticized as not particularly novel. Or we can say digital history is something new, occupying that in-between space—which could render the work unrecognizable to our usual communities.

The either/or proposition is, of course, ludicrous. The best work being done now skirts the line, offering something just novel enough to be surprising, but not so out of traditional historical bounds as to be grouped with culturomics. But I think we need to more deliberate and organized in this practice, lest we want to be like History and Philosophy of Science, still dealing with basic questions of legitimacy fifty years down the line.

In the short term, this probably means trying not just to avoid the rhetoric of newness, but to actively curtail it. In the long term, it may mean allying with like-minded historians, social scientists, statistical physicists, and complexity scientists to build a new framework of legitimacy that recognizes the forms of knowledge we produce which don’t always align with historiographic standards. As Cassidy Sugimoto and I recently wrote, this often comes with journals, societies, and disciplinary realignment.

The least we can do is steer away from a novelty rhetoric, since what is novel often isn’t history, and what is history often isn’t novel.


“Branding” – An Addendum

After writing this post, I read Amardeep Singh’s call to, among other things, avoid branding:

Here’s a way of thinking that might get us past this muddle (and I think I agree with the authors that the hype around DH is a mistake): let’s stop branding our scholarship. We don’t need Next Big Things and we don’t need Academic Superstars, whether they are DH Superstars or Theory Superstars. What we do need is to find more democratic and inclusive ways of thinking about the value of scholarship and scholarly communities.

This is relevant here, and good, but tough to reconcile with the earlier post. In an ideal world, without disciplinary brandings, we can all try to be welcoming of works on their own merits, without relying our preconceived disciplinary criteria. In the present condition, though, it’s tough to see such an environment forming. In that context, maybe a unified digital history “brand” is the best way to stay afloat. This would build barriers against whatever new thing comes along next, though, so it’s a tough question.

10 thoughts on ““Digital History” Can Never Be New”

  1. The best work being done now skirts the line, offering something just novel enough to be surprising, but not so out of traditional historical bounds as to be grouped with culturomics.

    Would you mind listing, say, the top 5 books/articles you would currently categorize under this description? I’m interested in the features of this work, as you see it—both the research itself but also features of the people and venues publishing it.

    1. “Beware, demon!” he intoned hollowly. “I am not without defenses.”
      “Oh yeah? Name three.”
      -Robert Asprin, Another Fine Myth

      More people should ask the question “Oh yeah? Name three.” – thanks for doing so. Obviously, since I do networks, those are the first that come to mind, but here are some examples of projects that skirt the line pretty well.

      Using networks to study history from below, allowing the frequently unnamed to, in aggregate, point towards unexpected larger truths about individuals or groups (especially, eg, returning women to the narrative):

      • Ahnert, Ruth, and Sebastian E. Ahnert. “Protestant Letter Networks in the Reign of Mary I: A Quantitative Approach.” ELH 82, no. 1 (2015): 1–1. doi:10.1353/elh.2015.0000.
      • Horowitz, Sarah. Friendship and Politics in Post-Revolutionary France. Penn State University Press (2014), Edition: 1, 240 pages, 2014.
      • Parsons, Elaine Frantz. Ku-Klux: The Birth of the Klan during Reconstruction. The University of North Carolina Press, 2016.

      Network as frame of exploration (similar to above):

      • Düring, Marten. “The Dynamics of Helping Behavior for Jewish Refugees during the Second World War: The Role of Brokerage Chains.” In Knoten Und Kanten III. Soziale Netzwerkanalyse in Den Geschichts- Und Politikwissenschaften. Bielefeld, 2013.

      Spatial history and results of analysis as worthy of explanation:

      • Blevins, Cameron. “Space, Nation, and the Triumph of Region: A View of the World from Houston.” Journal of American History 101, no. 1 (June 1, 2014): 122–47. doi:10.1093/jahist/jau184.

      Revision of previous historical claims, and reframing discussion:

      • Hitchcock, Tim, Adam Crymble, and Louise Falcini. “Loose, Idle and Disorderly: Vagrant Removal in Late Eighteenth-Century Middlesex.” Social History 39, no. 4 (October 2, 2014): 509–27. doi:10.1080/03071022.2014.975943.

      Two fabulous examples of using the process of reconstruction as a form of interrogation are, as far as I am aware, unfortunately unpublished, are:

      • Ben Schmidt’s work on revisualizing Turner’s maps of the frontier
      • Devon Elliott’s dissertation on fabrication for studying the history of stage magic

      Also, of course, the project by Matthew Lincoln already linked here.

  2. I think this relates to what Moretti said in LARB (remember when we used to like that publication? :P) who remarked that meaningful computational results often have a very different character from the pleasingly-polemic types of literary criticism that declare “Romanticism, in fact, didn’t start in this decade, but in this one!” Such a declaration stakes out a new position, but does so comfortably within a familiar discursive ground (Romanticism still exists as a concept for you, no matter in which decade you locate its origin.)

    A few observations from my dissertation experience. The four art historians on the committee found it easiest to understand my identification of structurally-important, but previously-underrecognized artists. The approach may have been novel, and the particular formulation of importance (one based on the logistics of production) rather different from their usual visual evidence. But the idiom of my argument — “I’m going to show you an artist you haven’t heard of but should pay attention to!” — was utterly familiar.

    The _longue durée_ arguments, on the other hand, were more difficult for them to grok (less difficult for the info scientist on the committee, though.) There, I was less declaring “this formation of a national school happened faster and earlier than you thought”, but rather “your concept of national school formation _exists in parallel_ with a totally different way of thinking about printmaking and nationalism that looks like this.” That continuity existed alongside, and in fact caused a great deal of, discontinuity – that was a trickier concept. It’s just a more foreign idiom for art history.

    So wrapped up in this problem about posturing and novelty, I think, is the old discipline-specific discourse problem, where the types and scale of evidence, and the questions considered worthwhile, are wrapped up in crushingly lengthy disciplinary histories. That probably explains why I got the most ringingly-positive feedback on my historiographical chapter that distinguishes between how “graphs” do their work, and what art history’s fuzzy conception of networks has looked like over our ~150yr modern disciplinary history. For better or for worse, it looks like that research, more so than any of the more-standalone analytical chapters, may have the most potential energy for a future book project. Maybe because historiography fits a familiar discourse, but also (I hope?) because these more senior scholars are looking for more rich signposts to help them understand how to build their own intellectual bridges (if not praxis ones) to the computational scholarship being produced by their graduate students.

    1. Thanks Matt, I think that’s spot-on, though worrisome. In DH venues, we spend too much time talking about method and the bridge between method and theory; I was hoping in traditional venues we’d be able to focus more on the analytic results, rather than more meta-discussions.

      At any rate, I think your experience speaks to my argument; that what was a “new type of question” wasn’t so interesting to the art historians, because it was difficult to recognize, and what was interesting to them was that which looked most similar to standard contributions to the discipline. Which isn’t a bad thing, I suppose, since disciplinary knowledge crawls forward carefully, but it does mean we must work to fit new wine into old wineskins if we want our work to be read/appreciated. It also brings us back to the fact that we need to convince historians that bringing something totally new to the table isn’t a good metric for the success of digital history.

      1. But I think the meta-discussion _is_ one of the most valuable products. TBH, I don’t really care about Jonas Suyderhoef or about Dutch regionalism, but I do care about how art historians in general think about identifying the kinds of people or monuments that we choose to raise up in our scholarship… or about how we conceive of things like genre, attribution, regional schools, etc. It’s one of the reasons I often say that I won’t mind if someone comes along in a few years and runs better, richer data against my methods and ends up with significantly different results. They will have used the method, and for doing so, they will think differently about how they practice art history.

        “Just the results” can often be quite boring if you sit outside the very small group of sub-specialists in your particular discipline. I would also argue the most impactful essays in art history (and, perhaps, every other field?) are valued for their methodological provocations, not their putative subject matter. You might think that we focus way too much on German limewood sculpture given how much Michael Baxandall gets cited in art history syllabi – but it was his concept of the period eye as a way to reject a Hegeleian teleology that made his scholarship so important. Don’t get me wrong, you certainly _need_ results to advocate for a methodology – we can’t just armchair scholar the shit out of this. But the results in and of themselves may have a far shorter half-life than the intellectual frameworks used to produce them.

        1. Hmm, this is tough. I disagree that the method is the meat of it, but do agree that it’s important and under-discussed in history. The most impactful/shared sources are certainly method and meta-discussions, but I’m not convinced it’s because they are any more intrinsically meaningful; it’s just that they’re nearly alone in their ability to be relevant to other historians. But even if a wide swath engages with some new method (topic modeling, anyone?) because it’s approachable, it doesn’t mean that approach is any more important or less parochial than what analysis it may produce. Basically, I imagine method and meta-discussions are more widely-read not because they ought to be, but because they simply can be.

          If digital history winds up reinvigorating the discussion of method and approach in history (as it already is – http://ahr.oxfordjournals.org/content/121/2/377.full ), then that’s great. It’s overdue. Without the specific contributions, though, why study history?

          I suppose another way to put this is that framing discussions offer nomothetic contributions, and “specific results” offer idiographic ones, but I can’t imagine either being useful without the other, regardless of the half-lives of specific findings.

      2. [Reposted comment. Not sure if other went through. Feel free to delete either one.]

        “I was hoping in traditional venues we’d be able to focus more in the analytic results, rather than more meta-discussions.”

        Ha. I’m currently having to defend my use of a simple bar graph in an historical journal. (To be fair, though, it’s in the context of an otherwise content-focused R&R, and wasn’t an outright rejection.)

        Regarding your “method is the meat of it” comment below, even though I share Matthew’s inclination, it’s an inclination I know I need to push against, so I have this quote from Lucio Colleti (via Moretti) on a post-it next to my desktop: Methodology is the science of those who have nothing.

Leave a Reply