Hitting the Books: How to build a music recommendation ‘information-space-beast’

ADuring the month of October, singers, songwriters and music producers are uploading 100,000 new songs every day to streaming services like Spotify. That’s a lot of music. There is no reality, alternate or otherwise, where one can hear everything even in thousands of lifetimes. Whether you’re into Japanese noise, Russian hardcore, Senegalese Afro-house, Swedish doom metal, or Bay Area hip hop, the sheer scale of available listening options is paralyzing. This is a big problem that data scientist Glenn McDonald is working to solve. From below in part Computing taste: Algorithms and the makers of music recommendationAuthor and Tuft’s University anthropologist Nick Seaver explores McDonald’s unique landscape-based method for surfacing all the tracks you never knew you could live without.

Also Read :  CeNSE researchers develop energy-efficient computing platform
White background with green wireframe text spread in different colors with plot symbols.

University of Chicago Press

Reprinted with permission from Computing taste: Algorithms and the makers of music recommendation By Nick Seaver, published by the University of Chicago Press. © 2022 by University of Chicago. All rights reserved.

A world of music

“We are now at the dawn of an age of infinitely connected music,” announced the data alchemist from beneath the Space Needle. Glenn Macdonald chose his title himself, preferring “alchemy” with its esoteric associations to the now-common “data science.” His job, as he described it from the stage, was to “use math and typing and computers to help people understand and discover music”.

MacDonald practiced his alchemy for the music streaming service Spotify, where he worked to transmute the basics of big data — logs of audience interactions, bits of digital audio files, and whatever else he could get his hands on — into valuable gold: products that. Can attract and retain paying customers. Macdonald’s mysterious power of alchemy seems to transform ordinary data, if properly processed, from thin interactional traces into dense cultural significance.

It was 2014, and McDonald was presenting at the Pop Conference, an annual gathering of music critics and academics in a building in downtown Seattle designed by Frank Gehry. I was on the other side of the country, and I followed online. That year, the theme of the conference was “Music and Dynamics” and Mc Donald began his speech by describing his personal musical journey, playing samples. “When I was a kid,” he began, “you discovered music by standing still and waiting.” As a child at home, he listened to folk music played by his parents on the stereo. But as he grew up, his listening expanded: the car radio offered heavy metal and new wave; The Internet revealed a world of new and obscure genres to explore. Where once he had been stuck in place, a passive observer of music as it went, he would eventually measure his life’s progress by his ever-widening musical horizons. Macdonald was able to turn this passion into a profession, working to help others explore what he called “the world of music,” which on-demand streaming services made more accessible than ever.

Elsewhere, MacDonald (2013) describes the world of music as a landscape: “Follow any path, no matter how unlikely and difficult it may seem, and you’ll find a hidden valley with hundreds of bands living there. For years, Australian hip hop, Hungarian pop, microhouse Or reconstructing the musical world in methodically- and idiosyncratically-altered miniatures, as in Viking Metal.”

Travelers will find the familiar and the surprising in a world of music — sounds they never imagined and songs they love. Macdonald marveled at this new ability to listen to music from around the world, from Scotland, Australia or Malawi. “The best music for you might come from the other side of the planet,” he said, but that wasn’t a problem: “In music, we have a teleporter.” On-demand streaming provided a kind of musical mobility, allowing listeners to instantly travel through the world of music.

However, he suggests, repeating the usual refrain, the scale of this world can be overwhelming and difficult to navigate. “For this new world to be truly appreciable,” MacDonald said, “we have to find ways to map this space and then build machines to take you down interesting paths.” Recommendation systems offered by companies like Spotify were machines. Macdonald’s recent work focused on maps, or as he described them in another talk: “a kind of thin layer of vaguely perceptible order in the raging, growing, immeasurably expanding information-space-beast of all the world’s music.”

Although his language may be unusually poetic, Macdonald was expressing an understanding of musical diversity that is widely shared among makers of music recommendations: music exists in a kind of place. That place is, in a sense, quite normal — like a landscape you can walk through, encountering new things as you go. But in another sense, this place is deeply strange: behind the valleys and mountains, there is a growling, growing animal, constantly growing and connecting points in space, infinitely connected. Music space can seem as natural as the mountains seen from the top of the Space Needle; But it can also seem like a man-made topological jumble at its core. It is organic and intuitive; It is technical and chaotic.

Local metaphors provide an effective language for thinking about the differences between producers of music recommendations, as they do in machine learning and in Euro-American cultures in general. In these contexts, it is easy to imagine certain, similar things being collected over hereWhile others, different things cluster there. In conversations with engineers, it is very common to find a musical space called into existence through gestures, which immerse the speakers in an imaginary environment populated by brief pinches of air and orchestrated by hand waves. One category is on your left, the other on your right. On the whiteboards and windows scattered around the office, you can find the musical space rendered in two dimensions, consisting of an array of points spread across a cluster and a plane.

In the music space, matching music is nearby. If you find yourself in such a place, surround yourself with the music you love. To discover more of it, you only need to look around you and move. In the music space, genres are like regions, playlists are like routes, and tastes are like flowing, archipelagic regions. Your new favorite song may just be over the horizon.

But despite their familiarity, spaces like these are strange: parallels can be found anywhere, and points that seemed far away can suddenly be close. If you ask, you will learn that all these spatial representations are not two or three dimensions, but rather a reduction of much more complex things of space, potentially containing thousands. This is McDonald’s information-space-animal, a mathematical abstraction that stretches human spatial intuitions past their breaking point.

Spaces like these, commonly called “similarity spaces,” are the symbolic terrain on which most machine learning works. To classify data points or recommend objects, machine-learning systems typically find them in spaces, assemble them into clusters, measure distances between them, and draw boundaries between them. Machine learning, as cultural theorist Adrian McKenzie (2017, 63) argues, “represents all difference as distance and direction of movement.” So while musical space is in one sense an informal metaphor (the landscape of musical variation) in another sense it is a highly technical formal object (the mathematical substrate of algorithmic recommendation).

spatial understanding of data travel through technological infrastructures and everyday conversation; They are at once a form of metaphorical expression and a concrete computational exercise. In other words, “space” here is both formality—a restricted, technical concept that conveys precision through abstraction—and what anthropologist Stefan Helmrecht (2016, 468) calls informality—a less disciplined metaphor that travels with formal technology. In practice, it is often difficult or impossible to separate a technical specification from its metaphorical counterpart. When creators of music recommendations speak of place, they speak at once figuratively and technically.

For many critics, this “geometric rationalization” of machine learning (Blank 2018) makes it suicidal toward “culture”: it quantifies qualities, rationalizes passions, and removes cultural objects from their everyday social contexts and relegates them to sterile isolation. Computational Grid. For example, mainstream cultural anthropology has long defined itself in opposition to such formalisms, which seem to lack the thickness, sensitivity, or adequacy of lived experience that we seek through ethnography. As political theorists Lewis Amur and Volha Piotukh (2015, 361) suggest, such analyzes “reduce heterogeneous forms of life and data into uniform spaces of calculation.”

To use geographer Henri Lefebvre’s (1992) terms, equality spaces are clear examples of “abstract space”—a kind of representational space in which everything is measurable and quantifiable, controlled by central authorities in the service of capital. Media theorist Robert Pray (2015, 16), applying Lefebvre’s framework to music streaming, suggests that people like McDonald – “data analysts, programmers and engineers” – are primarily concerned with an abstract, imagined space of computation and measurement. In Lefebvre’s view, the imagined space, the social, is parasitic on the living space, which Prey associates with the audience that resists and reinterprets the technologists’ work. The proliferation of abstract space under capitalism illustrates, in this framework, “the destructive conquest of the living by the pregnant” (Wilson 2013).

But for the people who work with it, musical space doesn’t feel like a sterile grid, even at its most mathematical. The creators of music recommendations do not limit themselves to sophisticated abstractions of imagined space. During their training, they learn to experience the musical space as normal and inhabitable, despite its inherent strangeness. Musical space is as intuitive as a landscape and as exotic as a complex, high-dimensional object of engineering. To use an often problematic distinction from cultural geography, they treat “place” like “place,” as if an abstract, homogeneous grid is a kind of habitable local environment.

Equality vacancies are the result of many decisions; They are by no means “natural,” and people like McDonald know that the choices they make can profoundly rearrange them. However, spatial metaphorizing, moving to speech, gesture, illustration, and calculation, helps make patterns in cultural data feel real. The confusion between maps and territories – malleable representations and objective terrains – is fruitful for those concerned with simultaneously creating objective knowledge and accounting for their own subjective influence in the process. These spatial understandings change the meaning of musical concepts such as genre or social phenomena such as taste, presenting them as a form of clustering.

All products recommended by Engadget are selected by our editorial team, independent of our parent company. Some of our stories contain affiliate links. If you purchase something through one of these links, we may earn an affiliate commission. All prices are correct at the time of publication.


Leave a Reply

Your email address will not be published.

Related Articles

Back to top button