I had never before attended the annual Modern Language Association conference — but this year I was invited by the MLA Libraries and Research Forum to join a panel on digitized collections. Here’s the panel abstract:
Digitization processes enable access to library and archival collections – processes requiring decisions that effectively shape how scholars and students will ultimately use these materials online. This panel seeks to open the black box of digitization workflows by starting a discussion about pivotal moments in the process–specifically how collections are selected for digitization; the process of creating OCR (Optical Character Recognition) transcripts, databases, and metadata for searchability; and the development of user interfaces and open access policies.
Each of us was assigned one of those “pivotal moments,” and I was invited to focus on the interface. The overall theme of the conference was “Boundary Conditions” — so I focused, in my seven short minutes, on how the interface itself is a “boundary condition,” and how it can be used to trace or bridge the boundaries between all those other stages in the digital workflow. In short, I wanted to function as a human interface between all the other presenters on the panel.
What follows are my slides and the text of my talk:
We might say that the interface is an exemplar of the “boundary condition.”  Theorists have variously described the interface as a deep surface, a zone of interaction, a gateway, a site of translation, a field of transition, a form of relation, a mediator of agency – a fertile nexus for all kinds of passages and permutations.  It’s simultaneously a uniter and a divider, a “thick” heterotopic space of interaction.
 For the longest time, the ideal interface was supposed to efface itself. The perfect portal was invisible, transparent, immersive, frictionless, seamless. Not any more.  In recent years, designers and theorists have embraced the idea of a productively “seamful” interface: one that doesn’t mediate too effortlessly, that doesn’t too thoroughly conceal all the complexity that lies beneath. There’s pedagogical value in friction, we’ve come to learn. Those seams are the metaphorical water wings that prevent our submergence into that deep surface.  By resisting total immersion, we retain more of our agency, our critical faculties, our ability to see the interface for what it is – an embodiment of epistemology, ontology, and ideology. An interface is an intellectual and political infrastructure.
 Over the past five years or so I’ve written a series of articles about urban data science and cities’ mediated infrastructures. A few of those pieces have focused on interfaces to the so-called “smart city” – the portals or gateways where citizens can peek into, consume, or, if they’re lucky, shape urban operations and civic services. Given the ubiquity of screens – apps, dashboards, public interactives, media facades, transit monitors, etc. – in these data-fied developments, I’ve argued, we need to pay particular attention to how their interfaces are engineering the values of our cities and structuring our subjectivities as urban citizens.
 Drawing inspiration from scholars like Johanna Drucker and designers like Mitchell Whitelaw, I offered a rubric for evaluating urban interfaces. Briefly, it advocated that we attend to the basic composition, materiality, scale, location, and orientation of our interfaces, as well as the modes of presentation and interaction (sight, sound, touch, smell) they make possible. I encouraged us to think about how interfaces impart a sense of orientation, how they help the “user” situate him or herself within the larger landscape; and how they “frame,” or conceptually partition, their content. I drew attention to the data models that undergird all these conceptual and technical architectures – and how those models embody an epistemology and methodology. I asked us to look for the hinges and portals between the technical systems’ myriad layers. And I encouraged us to ask: to whom does the interface speak, who does it exclude – and why, and how?
 These questions could be applied to an urban kiosk – like the Link kiosks now lining New York’s sidewalks –  or to an archival interface. And that’s what I’ll be focusing on in the limited time that remains:  I’ve decided to quickly stitch a few seams between the different presenters on this panel and the stages of workflow that each represents. I want to show how interfaces to digital collections can productively expose the seams of archival process.  While helping people find the collection materials they want, they can also show – for those who want to know – how those collection materials got there, and how they came to show up in their search results.
 First, interfacing selection: I’m sure many of you are familiar with Tim Sherratt’s work. His Invisible Australians project draws attention not only to what records are selected for digitization, but also to what people are recognized as citizens. Using facial detection software, a technology that readily lends itself to many nefarious applications, Sherratt and Kate Bagnall isolated portraits of non-white Australians in the National Archives of Australia. These were the “Invisible Australians,” those whose non-white faces, through their very existence, challenged the country’s White Australia Policy of the early 20th century. He then reinserted those faces back into the Archives’ website. This selection of thumbnails bestows subjectivity and permanent residence to a population selected for surveillance and exclusion.
 As Sherratt explained in his keynote presentation at the 2016 Migrant (Re)Collections conference, the original “process of identification,” which involved portraits and handprints, “helped justify the racist underpinnings of the system.” Today, too, our own identities – and particularly black and brown identities – are selected for capture and digitization: “Our faces are increasingly not our own – they are public signifiers to be captured by systems of identity management and surveillance.” Yet by “exercis[ing] our power to aggregate, and to name,” Sherratt argues, we can “turn these systems on themselves.” Selection-for-surveillance becomes selection-as-subjectification.
 Second, interfacing OCR: Consider the late, great NYPL Labs’ “What’s on the Menu?” project, which highlighted the challenges of OCR’ing typographically or graphically adventurous and handwritten menus.  Its Building Inspector does the same for maps, asking people to confirm the boundaries of individual properties.  Other researchers and developers have proposed, or are developing, annotation tools to supplement OCR. I’d encourage us to think about using our interfaces to go beyond the optical, too – to highlight other modes of automating engagement with digital collections.  Consider Tanya’s work with HiPSTAS, and how and interface can promote fruitful partnerships between automated listening and individual, critical listening.
 Third, interfacing the database:
- Showing the extent of a collection, and helping patrons orient themselves within the grand scheme of a collection
- Other topologies – lists, networks, 10,000-foot views – NYPL Labs interfaces;  Mitchell Whitelaw’s “generous interfaces”
-  Rosten Woo’s Bookstacks – how to take circulation data and make it userful for patrons
 Fourth, interfacing metadata:
- Whitelaw’s work – multiple means of sorting
-  The Cooper Hewitt’s offering of myriad ways to explore their collection
 Fifth, interfacing open access policies:
- Making your politics overt – DPLA, Europeana, Internet Archive
-  Rijksmuseum
 Interfaces are not merely providing access to content; but also opening up the technical, social, and political systems within which that “stuff’ resides, and which gives it context. How can our interfaces provide this larger Latourian actor-network or Foucauldian apparatus surrounding the collection as an option, for patrons who want it?  In closing, I want us to think, too, about how to provide that entry through means other than screens and optics. Our digital collections represent multiple modalities, the analog objects that many of those digital records document come in myriad formats, patrons research and learn through their multisensory bodies and multiple intelligences; lets think about building that richness into our interfaces.