Introduction
The last few decades have seen the emergence of interdisciplinary and transdisciplinary research areas such as Computational Social Science, Digital Humanities, and STEAM. From my own professional experience and network highly similar activities, perhaps under different labels such as Data Scientist or Creative Technologist, have also arisen in industry and government sectors.
These collaborations have seen a variety of computational data tools and methods traditionally associated with STEM disciplines harnessed, for example to explore rich cultural material that has been digitised or mine constantly evolving sources of user-generated content.
It would be expected, therefore, to find Jupyter notebooks, the go-to notebook for computational data tools and methods \cite{choice} fairly popularly used at the cutting-edge of Humanities, Social Science, and Arts research. Although difficult to guage accurate saturation, there are indications of this, for example the Research section of this maintained collection of Digital Humanities Jupyter notebooks \cite{quinnanyadh-jupyter} or topic searches of github code repositories.
Whilst there is clearly room for greater researcher adoption, the focus of this article is however on what the evidence, or lack of as will be discussed, that there exists a monumental opportunity for Jupyter notebooks to be utilized for teaching Humanities, Social Science, and Arts, and unlock hugely impactful pedagogical benefits for both instructors and students, replicating benefits frequently reported from Jupyter notebook-based STEM teaching.
My own search to find and document Humanities, Social Science, and Arts courses at undergraduate and secondary education level in the US or UK using Jupyter notebooks threw up a mere 15 programs. Worse, several of these items are highly related hence not unique, and also include my own courses. Even accounting for the inaccessibility to access/uncover what could be a majority of the courses of interest, the true signal of adoption rates is still likely to be disappointing low.
The aim is therefore to present the findings from these few use-cases. Whilst data points are scarce, some educators have written up or been quoted in-depth about their courses.
Noting the relevance of selection bias both in the sample of existing courses able to collect as well as any write-up, the next sections outline the data collection process (...), present the dataset (...), examine the available evaluations and provide common themes and concrete/granular examples as to any benefits or negatives (...). The article ends with a conclusion outling actions recommended on the basis of the evidence.