New research has revealed music streaming algorithms disproportionately recommend male artists. Scientists have a solution to reverse this trend, but will streaming services change their code to make the industry fairer?
Gender division in the music industry is no secret. Since 2012, women have represented just 21.6% of all artists on the Billboard Hot 100 Year-End Charts and in 2020, this fell to 20.2% of artists.
The division at music festivals is equally damning. Vick Bain, founder of The F-List, an online resource that lists artists, bands, songwriters and composers who identify as female, found that before the pandemic, only 8% of 2020’s festival headliners were due to be female. This included just three acts—Taylor Swift, Little Mix and Haim —topping the bill at the UK’s 16 biggest events. These statistics reflect the sad reality: women’s representation in the music industry has remained embarrassingly low and most alarmingly, stagnant for the last decade.
Could music streaming bridge the divide?
Dr Christine Bauer thinks so. She’s recently been involved in an extensive study examining the ‘Gender Imbalance in Music Recommenders’ and the results could help reshape how we listen to music.
“Our inspiration came from interviews with music artists,” explained Dr Bauer, who is Assistant Professor in the Department of Information and Computing Sciences at Utrecht University. “We asked them how they felt affected by music streaming platforms and what they would like to see improved.
“All interviewees said gender imbalance was a major problem in the music industry—a known reality for decades, but the artists saw potential to use the recommendation algorithms to address this issue and give female artists more visibility.”
These conversations planted the seeds that led to nine years of study. 330,000 participants. And, solid results: only 25% of the music consumed by participants in the study was by female artists.
“Concerning the proportion of female artists in the recommendations, we were not surprised because this unbalance comes from the data; and the representation of women and gender minorities in the music industry is tremendously low,” continued Dr Bauer. “However, we were surprised about the difference in exposure in the recommendations: the first recommended track was by a man, along with the next six.”
Echo chambers in the algorithm
As streaming services use algorithms to recommend new music, artists, and playlists to listeners, these recommendations can greatly impact the new music’s success. Based on previous song choices, these recommendations sometimes play automatically or are pushed to the top of the feed but most importantly, they are given prominence over other songs. The study found that on average, a female artist wasn’t recommended until the seventh or eighth song. That could equate to 20 minutes of listening before a female voice was heard. This lack of exposure in the algorithm was a eureka moment for the researchers.
Streaming is now the most popular way to share and listen to music; over 60,000 songs a day, nearly one per second, is uploaded to Spotify, and streaming now accounts for 81% of UK music consumption, according to industry body the BPI. The study showed how the under-exposure of female artists within streaming services has only exacerbated divisions in the already gate-kept industry.
This revelation led researchers to the root of the problem. “The inspiration to focus on the ranking when amending the algorithms,” notes Dr Bauer. “As users listen to the recommended songs, the algorithm learns from these. This creates a feedback loop. To break this feedback loop, we came up with a simple approach to gradually give more exposure to female artists.”
To do this, the researchers took recommendations suggested by the algorithm and re-ordered them by moving male artists a specified number of positions downwards. Once they had re-ranked the recommendations, the researchers ran a simulation to see what would happen to users listening habits. The results were illuminating.
Breaking the feedback loop
Soon, listeners starting changing their behaviours and listening to more female artists than before.
“Eventually, the recommender started to learn from this change in behaviour,” concluded Dr Bauer. “It began to place females higher up in the recommended list, even before our re-ranking. In other words, we broke the feedback loop.”
Of course, there are limitations with this binary method of looking at under-representation. Due to an unavailability of data beyond binary genders, researchers were unable to incorporate the range of gender identities that exist in their research. Yet, this research can be seen as the first step to making the music industry fairer. With data, people have evidence that something is unfair, so can better hold streaming services to account. And, some are already starting to listen.
“Only one in five artists in the charts are women. Yet, we know how integral women artists’ influence has been on Spotify and the music industry at large,” Spotify said in a recent statement. “To begin to upend this disparity, we must amplify the work of women creators by extending our resources and generating more opportunities for these artists.”
In efforts to course-correct inequality in the industry, the streaming giant has recently expanded its EQUAL hub: a platform, launched on International Women’s Day to help promote women in music, to incorporate the EQUAL Global Music Program. This will highlight female artists from 50 countries to give them more exposure on the platform.
Change may be slow but now streaming services have the key to enable fairer representation. It’s time they broke the loop.
Words by Elouise Hobbs
Support The Indiependent
We’re trying to raise £200 a month to help cover our operational costs. This includes our ‘Writer of the Month’ awards, where we recognise the amazing work produced by our contributor team. If you’ve enjoyed reading our site, we’d really appreciate it if you could donate to The Indiependent. Whether you can give £1 or £10, you’d be making a huge difference to our small team.