Unboxing DH – how do we generate knowledge with digital methods and tools?
4th DH-Day of the University of Münster on December 5, 2022
Humanities work using digital methods is generating more and more data-based results of research. However, many DH research methods require tools that leave parts of the knowledge path in the dark. The results of unsupervised machine learning, for example, often come from a black box, a process that cannot be seen. These results become the basis for interpretation work in the humanities, and questioning the genesis processes of the computational results to be interpreted becomes difficult or even impossible. Last but not least, algorithmic operations can render invisible any biases underlying the data itself. In fact, unexplained presuppositions are the subject of criticism even in the application of traditional research methods. Are they―as is not infrequently assumed―removed from any comprehensible criticism as input into a black box? What is clear is that at least some of the DH subdisciplines generate strong epistemic frictions in the digital cognitive process. Already in the selection of what is put into the black box lie presuppositions that need to be exposed. Data are not given, but taken and made (Johanna Drucker speaks in this context, for example, of "capta" instead of "data"); the steps of interpretation should be disclosed at every stage. We need to ask how insights gained through non-reproducible procedures relate to the claims of intersubjective comprehensibility of research and FAIR principles for data. Which role do authorities and which do algorithms play in assessing the value of data and knowledge gained through them? On the Day of DH 2022 at the University of Münster, we want to come together against this background and discuss with each other how we generate knowledge by digital means and which gradations between the black box and the complete traceability and reproducibility of research results are imaginable.