IGF 2022 Networking Session #76 Biasing music consumption and possible impacts of recommendation systems

InternetLab
Fernanda K. Martins, InternetLab director, civil society, Latin America, Brazil;
Alice Lana, InternetLab head of research, civil society, Latin America, Brazil;
Clarice Tavares, InternetLab head of research, civil society, Latin America, Brazil;
Ester Borges, InternetLab head of research, civil society, Latin America, Brazil

Speakers

Fernanda K. Martins, InternetLab director, civil society organization, Latin America;
Mariana Valente, University of St. Gallen, Switzerland, Europe;
Alana Rizzo, Public Policy, Brazil and Emerging Markets, YouTube, private sector, Brazil, Latin America;
Ayanda Sisulu, Tidal, Private Sector, South Africa, Africa;
Maria Eriksson, Postdoctoral fellow at Humlab, Umeå University, Academic, Sweden, Europe

Onsite Moderator

Alice Lana

Online Moderator

Ester Borges

Rapporteur

Clarice Tavares

SDGs

5. Gender Equality
5.1
5.5
5.b
10. Reduced Inequalities
10.2
10.3


Targets: The proposal seeks to explore how algorithmic recommendation systems influence the consumption and distribution of music and cause biases in the formation of the identity of consumers-users. In this discussion, the gender and nationality perspectives are fundamental to understanding how algorithms could discriminate against marginalized groups. Thinking about differences between Global North and South and the inequalities between women and men, the discussion may bring perspectives on inequalities that are formed and reinforced in cultural production mediated by algorithms, seeking to bring clarity to the discriminatory potential of algorithms that could impact the economic lives of marginalized groups and interrupt their empowerment.

Format

Roundtable.

Duration (minutes)
60
Language

English

Description

The penetration of streaming platforms globally has changed the way of distributing and accessing music, as well as other cultural goods mediated by them. While the consumption of music in physical media depends on factors such as financial resources and the availability of CD copies, in addition to visits to stores for discoveries, on the platforms consumption is driven by the possibility of free access to music, supposed freedom of choice due to the infinity of options, as well as comfort and ease in discoveries. However, this apparent freedom of choice is confronted with the enormous capacity that streaming platforms have to reinforce or produce preferences and constitute values and subjectivities in consumer users' cultural and social framework through recommendation systems. Fueled by the massive collection that results in a behavioral profile and serves the company's corporate interests, these systems suggest playlists, genres, artists and songs conditioned not only by current tastes, contexts and habits but also by neoliberal and individualist dynamics. Recommended playlists are often based around notions of productivity, well-being, and happiness (such as those for focus or physical activity) and tend to include more established artists, with more commercial music, or are powered by major label sponsorship. Thus, the sociotechnical perspective on algorithmic recommendation systems cannot capture their effects as cultural intermediaries in the construction of identities, the reinforcement of certain worldviews and ideologies, and the exercise of power and control. As certain ideologies or generally hegemonic worldviews guide them, they are crossed by gender, race, and class biases and tend to construct identities and preferences in a uniform and little different way. The lack of transparency about the functioning of these recommendation algorithms is another factor contributing to the limitation of the user's will to consume cultural goods and music. This context raises questions that we will try to answer in this roundtable, such as: (i) what ideals and subjectivities are produced in the organization and presentation of songs by the streaming recommendation systems?; (ii) to what extent do these systems reproduce biases by shaping consumer users' habits and preferences?; (iii) how do they shape emerging music production and the visibility of smaller artists, contributing to the diversification of culture?; (iv) what are the criteria used by the systems to recommend songs, artists, genres and playlists to users?; (v) what determines the entry of a song or artist in a pre-configured playlist of streams?



Considering the presence of an audience on-site and an audience online, as well as speakers on-site and online, we are thinking of ways to integrate these different situations of the public. In the first 5 minutes, we will address questions related to the theme in a pool to online people and questions directed to onsite people. The moderator will raise questions for the next 20 minutes, and the speakers will answer quickly. The objective is to maintain a fluid dynamic. For the next 20 minutes, questions could be submitted by the audience. The online moderator will be responsible for organizing the questions of online participants, while the onsite moderator will be responsible for managing the questions of onsite participants. For the next 8 minutes, the speakers will share their final remarks and for the next 5 minutes, the rapporteur will summarize the main points of the roundtable. In the last 2 minutes, the moderator will finalize the session and thank all of the participants. During the process, onsite and online participants will be able to create notes on the platform mural, while the online moderator will summarize the main points in the same platform.