Each month, the open access team of the University of Groningen Library (UB) puts a recent open access article by UG authors in the spotlight. This publication is highlighted via social media and the library’s newsletter and website.

The article in the spotlight for the month of February 2024 is titled The algorithmic network imaginary: How music artists understand and experience their algorithmically constructed networks, written by Robert Prey and Marc Esteve Del Valle (Faculty of Arts).

Abstract

In this article we develop the concept of “algorithmic network imaginary” to understand how musicians imagine and relate to the networks of “related artists” they are algorithmically sorted into on Spotify. To address this question, we collected data on the related artist networks of 22 musicians constructed by Spotify’s Fans Also Like feature and conducted semi-structured, in-depth interviews with each musician. We used the Qualitative Structural Analysis method for data analysis. Our findings provide insight into what musicians think Fans Also Like is and is for, and reveals how cultural creators understand and experience their algorithmic networks. More broadly, they provide insights into how social actors perceive, understand, and experience their algorithmically constructed peer networks.

We asked authors Robert Prey and Marc Esteve Del Valle a few questions about the article:

Spotify’s Fans Also Like (FAL) feature uses an algorithm to sort artists in a network of related artists. How do artists experience this list of related artists?

For our research we generated visualizations of Spotify’s related artist network for each music artist that we interviewed. These visualizations elicited discussion about how our interviewees thought FAL worked, and how they thought it should work.

Most of the artists we talked to took genre as the standard to which FAL’s recommendations should be compared. This expectation led to disappointment or confusion when musical acts from different genres appeared in their FAL. Indeed, almost everyone we interviewed had complaints or disagreements about some of the artists they were linked to in FAL. Some complained that they were mainly connected with artists from their same country who shared a similar level of popularity, irrespective of genre or musical style. Several of our interviewees mentioned that FAL connected them with artists from genres and scenes that they feel they have long moved on from. This was important to them because - as artists - they highly value the ability to evolve and change styles.

Most of our interviewees felt that FAL should not only help listeners discover new artists and music but that it should also connect artists with the scene they feel an affinity with.

In your findings you describe the network of related artists created by Spotify’s FAL as an “abstraction of a music scene”. Can you explain what this means and how this was perceived by your interviewees?

To create its network of related artists, FAL algorithmically combines the listening habits of fans on Spotify with music discussions and trends happening around the internet. It thus draws from online practices that partially constitute music scenes. In doing so, it provides a data-driven abstraction of a given act’s music scene. It is for this reason that we argue that the network of related artists produced by FAL should be understood as an abstraction of a scene: FAL “abstracts” from music scene practices across the web. 

Musicians have always derived visibility, status, and distinction from the scene they are associated with. Scenes are assemblages of shared practices, aspirations, and spaces. In our interviews we heard a lot about how FAL represents or misrepresents the “scene” these artists felt they belonged to. 

What can artists do if they don’t agree with their FAL network? Are some artists ‘gaming the system’, manipulating the algorithm to change their Spotify’s FAL network?

In its description of the feature, Spotify clearly states “Since FAL is not manually curated, it’s not possible to edit the artists that appear in this tab.” Music artists are very aware of the importance of FAL for how they are perceived and for attracting new fans. As a result, many of our interviewees complained about how they weren’t able to directly dispute their FAL associations. 

Interestingly, complaints to Spotify used to be higher when the feature was called “Related Artists”. The feature was rebranded in order to change the perception amongst musicians that Spotify was manually creating an artist network for each music artist. The new name - “Fans Also Like” - was chosen to communicate that the feature simply measured what “fans” were listening to. 

Spotify advises musicians who are displeased with the acts they are related to on FAL to generate more data. On its website, Spotify tells musicians that they can fix their FAL by “staying active on social media”; “encouraging (their) fans to tune in to your music on Spotify”; engaging in “conversations about (their) music across the internet”; and “creating and sharing playlists”. While some of our interviewees admitted to trying to ‘game the system’ in order to improve their FAL, most of them recognized the practical difficulties of actually accomplishing this.  

You worked with qualitative data (semi-structured, in-depth interviews). Is the data you collected FAIR? How is it stored and is it accessible to the public? 

The semi-structured interview data we collected includes personally identifiable information of the interviewees, which is relatively easy to discern due to the small size of the music acts. Thus, for this project, we decided not to adhere to the FAIR principles to safeguard the requested anonymity of the interviewees more effectively.

Could you reflect on your experiences with open access and open science in general?

We believe that making scholarly work and scientific findings freely available to everyone is key to helping science, and society, advance. Putting this belief into action, in past projects involving social media data, such as an investigation into the impact of Reddit users on the collective memory of Michael Jackson (Moonwalking together: Tracing Redditors’ digital memory work on Michael Jackson), we rigorously followed the FAIR principles and subsequently made the complete dataset available via the Open Science Framework (OSF).

Useful links:

The UG’s Digital Competence Centre supports UG researchers throughout the entire research (data) life cycle, from grant proposal to FAIR data archiving.

An overview of the support available at the UG for staff members who want to engage in open science practices. These practices (and the support) include open access, FAIR data, open education, public engagement and more.

Citation:

Prey, R., & Esteve-Del-Valle, M. (2023). The algorithmic network imaginary: How music artists understand and experience their algorithmically constructed networks. The Information Society, 40(1), 18–31. https://doi.org/10.1080/01972243.2023.2274070

If you would like us to highlight your open access publication here, please get in touch with us.