This is part of a series of summary posts rounding-up new entries to the Copyright Evidence Wiki (organised thematically). As part of CREATe’s workstream for the AHRC Creative Industries Policy and Evidence Centre, the Wiki catalogues empirical studies on copyright. In this post, we summarise the themes prevalent between 2019 and 2020 using our new Viz tool, and share study highlights.
Year of the Book
In contrast to previous years where music has been, for some time, the subject-matter of most interest, studies from 2019-2020 saw the rise of the book as the most pressing area for research.
Within this cluster, we see more specific topics emerge:
- Libraries. As part of a growing body of work in the eLending project, studies by Giblin (2019a) and (2019b) explore the library e-lending landscape. Through an analysis of licensing terms offered by publishers across multiple jurisdictions, the studies caution that, due to little distinction in the licensing terms offered between lower-demand and higher-demand titles, there is little incentive for libraries to obtain older, culturally relevant books.
- Book availability. Studies by Heald, (2020a) and (2020b), explore copyright’s effects on book availability in two distinct markets. The former utilises a cost-benefit analysis of the proposed reversion right in Canada (more on EU reversion rights in our working paper by Furgal here and the Australian position here), concluding that this may be useful in offsetting the welfare loss caused by long copyright terms (for more on this topic in the music industry see Garcia (2020)). In the latter study, focussing on the proposed term extension in South Africa, the research confirms that the public domain status of a book correlates with its better availability and affordability (similarly confirmed in an earlier study by Reimers (2019) in a study of bestselling fiction novels).
- Fan fiction. A topic of particular interest in 2019 (see: Katz (2019), Fiesler and Bruckman (2019), Khaosaeng (2019)), Flaherty (2020) offers a research agenda for progressing forward. Where previous studies have often approached the study of fanfic qualitatively (often through interviews and thematic coding), this study suggests that quantitative methods may serve to prove the often-suspected positive externalities of user-generated content. In turn, this may provide more evidence in support of a fair dealing exception for fan creativity.
As the honourable runner-up industry, there are plenty of studies worthwhile exploring in relation to music: see Gani (2020) for a study on performers’ creative autonomy and motivations; Pappalardo and Aufderheide (2020) on reuse practices that share a logic with romantic notions of creativity; and, Owen and O’Dair (2020) for the Wiki’s very first study on the viability of blockchain technologies in music copyright.
Year of Enforcement
Enforcement issues (e.g. motivation for infringement, quantifying infringement, court data etc.) continue to be the most popular category of study. In the visualisation below, we see clusters around enforcement issues (category F) in books, music, film, TV and software:
A quick glance at our Word Cloud suggests we still conceptualise these activities primarily as ‘piracy’, despite its often contested connotations.
In particular, we see in recent years a trend towards the study of algorithmic decision making in cases of suspected music infringement (for an earlier example, see Savage (2018)). In their study of “Song-To-Song Similarity Measurements” Newman et al. (2020) test a ‘stemming’ approach to detect similarity in cover songs, by analysing similarities between granular aspects of music such as bass, pitch, drums etc. In cases where plagiarism was previously confirmed in court decisions, the final version of the algorithm confirmed the same in nearly 50% of these cases.
In a slightly different vein, Yuan et al.’s (2020) study pits human against machine, exploring the differences between human perceptions and automated algorithms in detecting music similarity. Using a dataset of court cases where substantial similarity between two songs were in question, the study finds that human perceptions of similarity are perhaps less accurate than automated algorithms (the latter matching court decisions in 71% of cases, compared to approx. 50% for human perceptions).