Kyle Devine’s 2019 book Decomposed: A Political Ecology of Musicis making waves. As Devine explains, the book started as an investigation of the nostalgic return of the vinyl record, a seemingly “backwards” trend in current music consumption. However, the more he looked into the issue, the more he was challenged by the story of the music’s material presence as data in the age of mechanical reproduction. His key takeaway, and stark truth: recording technology, from shellac discs, to vinyl LPs, CDs, mp3s, and contemporary streaming services, comes with environmental impacts. What music is made of matters, and its cost to the natural world is a problem of “political ecology.”
In this final episode of our first season Chris takes us through the book, looking for resonances and intersections with the Sounding History project. As a historian of empire he finds parallels, for instance, between the music industry’s environmental costs and empire’s human toll, up to and including mass enslavement: whether in the sugar slaveocracies of the Caribbean, or the server farms of Iceland, empire’s environmental costs have too often been concealed “just over the horizon.”
Yet despite our enthusiasm for the book, we are not entirely convinced by some portions of Devine’s account. We reflect, for example, upon the price (in deforestation and exploitative labor) of the shellac record, as against the liberation and democratization that easily accessible recording technology brought to subaltern and minoritized musical experiences in the early twentieth century. Shellac records made it possible, wherever musicians and technology could come together, for people (“the people,” even) to tell their stories in sound. Without shellac, we believe, there would have been no blues revolution, no Ma Rainey, no Robert Johnson, even, no jazz.
It turns out that the social-environmental-historical-economic impact of datafied music is not an easy nut to crack.
We thus end the podcast (and our first season!) with a quick glimpse of some work Tom is doing at the Alan Turing Institute, the UK’s national center for data science and AI, where he directs the project “Jazz as Social Machine.” Today machine learning agents drive cars, diagnose disease, play chess, and design buildings–among many other human tasks. Such autonomous systems also improvise jazz. It turns out, though, that jazz improvisation is apparently harder than driving a car! Why? The answer has to do with risk, historical “consciousness,” and the attitudes towards “getting it wrong” that underpin the algorithms of the machine learning revolution.
All of the books mentioned in the episode can be found in our Sounding History Goodreads discussion group. Join the conversation!