The Matter of Metadata: Archival Constructs between Index Cards and AI

If you were to browse through the complete catalog of the saai–one of the largest architecture collections in Germany with the estates of Frei Otto, Egon Eiermann and Günter Behnisch, to name just the best known–one would find works by over 200 offices and architects. They would be sorted by name, referenced by material and medium, size and date, cataloged according to predefined categories and saved for future visitors. One would not find many women, their practices or spaces for or by marginalized groups or diverse practices. Like many institutionalized archives in the Western world, this speaks to the long-standing conventions of a discipline constructed by so-called canonical figures—and their work. However, the metadata of the archival catalog does not consist of objective, stable categories. If metadata are understood as objects, as are their processes, their infrastructures and their consequences, then techniques for their virtual indexing can be learned from the human-made information from collected objects. Not only which parameters have to be entered into which databases and how, but above all which metadata promote and prevent which research focuses; which techniques of knowledge transfer into catalogs and knowledge systems create blind spots or lead to erosions and omissions. This is because metadata is still the interface of the archive, what can be seen (and what is visible at all) for users. Every bias, every value formation that is already established in the archives' systems is necessarily not only mapped in data-generating AI solutions (be it for digitization or indexing), but overlaid with their already existing bias (from data pools and programming). Only if material techniques and the specific archival-valuing biography of the archive are critically scrutinized can blind spots become visible - and be approached differently when transposed into AI recordings.

The project is therefore based on two pillars from the outset: an archival-material and a medial-institutional one. A humanities dissertation that will examine selected archival materials in the saai archive in terms of their materiality and their role in metadata production in the archive, for example annotations, index cards or sorting logics in folders and binders. These case studies will lay the foundations for a material understanding of metadata as a material-medial practice and demonstrate the relevance of the processes of archiving for the objects and their reception. The second, as an interdisciplinary postdoctoral project in the humanities with a “wide-angle” perspective, will examine the processes of institutionalization of archival materials and their medial impact from the archive into society. This will reveal archival practices and policies in architectural collections at German-speaking technical universities such as the saai over longer periods of time and inter-institutionally. Both will cooperate and contrast findings with a newly established data technology project at saai: a postdoctoral research project with a background in computer sciences and digital humanities that examines the question of metadata genesis and born digital archives as a praxeological question and explores alternative tools and approaches to data technology.