The Marginalia Machine is a drawing robot – a Cartesian plotter – that reproduces editorial notes from manuscripts in the archive of Bloodaxe Books. Taking a scanned poetry manuscript, custom-made computer vision software differentiates the text of the poem itself and the handwritten notes around it. The main text is discarded and the notes are sent to the machine where they are drawn, with a pen, on a continuous scroll of paper. Over the course of an exhibition a number of scrolls are produced combining the notes of many manuscripts in a series of enigmatic snatches of text and looping abstract marks.
The Marginalia Machine was produced as part of a Research Through Design process within a larger, funded research project, ‘The Poetics of the Archive, Creative and Community Engagement with the Bloodaxe Books Archive’ based in the department of English Language and Linguistics at Newcastle University, UK. Bloodaxe Books is a small but internationally significant publisher of contemporary poetry, whose archive, consisting mostly of edited manuscripts, was purchased by Newcastle University in 2013. Our role within this research project – as artists and interaction designers based in Culture Lab an interdisciplinary research centre focused on digital culture – was to create exploratory and provocative interactions with the archive both online and in physical space and the Marginalia Machine was part of a larger ecosystem of works of creative art and interaction design.
With the machine itself as with other things we made, we aimed to be responsive not only to the aspirations of the various project stakeholders (Bloodaxe Books themselves, our colleagues in English and in the library Special Collections department as well as the community of poets, scholars and other parties interested in the archive) but also to the materiality of the archive itself. We conceived of this materiality as consisting not only of the manuscripts and records themselves but also in a broader sense as being embodied in the technical infrastructure, working practices and physical spaces of the university. The Marginalia Machine explores the archive materials themselves but in the context of their transition from an uncatalogued archive to one that is formally described, searchable and publically available. Not only was the archive being formally catalogued throughout the design process but some of the materials were being digitized, a process which profoundly alters the way we see such material and the way that it is used. As such we saw this period as a particular kind of opportunity within the life of the archive to experiment with its presentation to the public in the course of learning about it ourselves. Archives are often held to be seats of a particular kind of knowledge and with that knowledge comes a kind of power to produce a partial description of the past.
By varying and pluralizing the description and presentation of the archive with the Marginalia Machine and with other creative work we hoped to open this description to interpretation and to at least feather the edges of some of that power to present the past.
Research Through Design uses the unexpected consequences of building and using physical things to find things out. The Marginalia Machine – although presented as an artwork – was part of a design process that allowed us to think about the archive and its new home in the university and helped us to design later interfaces to it. One particular thing we came to consider was the way that we had thought of the archive as a static entity that we were simply describing in different ways. The demands of the production process of the Marginalia Machine caused us to engage with our colleagues in the library and learn about what they did. Being unused to the actual practices of cataloguing and digitization our prejudice was to conceive of them as something which stood effectively apart from the materials themselves, as something that lay ‘on top’ so-to-speak. On the one hand there would be the materials themselves and on the other a set of meta-data, image files of digitized material and interfaces on the web written in code. These things we assumed would be formally separate. Of course in practice this is radically untrue. Cataloguing physically disturbs the material in a number of significant ways. Some items are literally discarded. Duplicates and enclosing materials such as envelopes are trashed, materials are re-packaged and re-arranged sitting in new groups suggesting new correspondences. Items are marked in pencil with their newly assigned identity numbers. They are clipped with brass paper clips and bound with unbleached cloth. These actions are performed in order to preserve the archive, to arrange it according to a vision of use and to render it tractable for study but there is also a kind of violence at work, a tough love as heterogeneous materials are made homeogenous, filtered, sorted, marked and boxed. Most of all there is a web of professional activity and technological/material transformation around the manuscripts that caused us to conceive of the archive as ‘live’. Liveness as we conceived it brought about a view of the archive as a dynamic, processual entity open to active reconfiguration.
With the Marginalia Machine we were particularly interested in aspects of the archive itself that were not represented in the formal catalogue. The margin notes may be referred to in the catalogue description but typically would not be except in cases where they are particularly significant in the definition of an item. Even then the description is terse and formal. We consequently saw an opportunity to foreground this rich facet of the materials at exactly the point (i.e. during cataloguing) where it eludes description. In this sense we were attempting to refer to aspects of the process of archiving at a professional, material and technical level.
The software behind the marginalia machine takes newly created digitized images and decomposes them into background, typewritten and handwritten/ hand drawn parts. At the point where an item was formally ingested into the system our software intervened and effectively ‘forked’ its development into a new, unexpected strand. This sense of timeliness also led to a related design intervention within our project. @BloodaxeArchive is a Twitter bot, an automated agent that tweets marginalia images from the archive as soon as they are digitized alongside a link to the catalogue entry describing the original item. Taking advantage of our own position within the university’s internal network we were able – with the agreement of our colleagues in the library – to programmatically watch the technical systems with which the manuscripts were digitized and mark the appearance of each new item, in an indexical fashion, with a public manifestation. The timeline of @BloodaxeArchive actually reads as an alternative history of the archive’s archiving, one uttered by a hacking of the technical infrastructure of the archiving process. Scrolling through this history it is possible to observe batches of works from a particular author, sets of cover pages even to get a sense of how much digitization was being done and when. In the Marginalia Machine too some images stand alone with legible notes describing no-longer-present text in a kind of poetic elision. Others form new correspondences as formal similarities emerge between the (now) abstract shapes of crossings-out, parentheses and typographical annotations. These sequences tell a story of archiving in enigmatic visual language quite apart from their catalogue descriptions.
The Marginalia Machine and @BloodaxeArchive were also attempts to engage publics with the emergence of the archive and the interfaces to it. With them we hoped to engage both expected audiences (@BloodaxeArchive has around one hundred followers who are almost exclusively poets, literary scholars, archivists and librarians) and – with the public exhibition of the Marginalia Machine in venues like the V&A – to attract new ones to the fascinating material painstakingly catalogued by the Special Collections department at Newcastle University. The exploratory and experimental nature of the Marginalia Machine also inflected the interface that we built to the archive. In a direct reference back to these designs, marginalia-only versions of scanned images are available for many items but more generally our various off-kilter ‘lenses’ on the collection reflect our interest in pluralizing and loosening the ways that the items are described. ‘Shapes’ for instance allows users to sketch the form of a poem with the mouse and explore poems with a similar outline. ‘Words’ uses text-mining techniques to explore network-like correspondences between manuscripts.
Our aim with these designs was to take a step outside of conventional ways of seeing, describing and searching archival materials and by doing so to think about the potential of digital culture heritage, not only as a source of new knowledge but as a site of creative potential. We hope that you will explore the archive and make something new.