Data Found

After collecting and analyzing the data, we reached results that corroborate the research findings of Ferraro, Serra and Bauer1. By using two public databases containing information about the behavior of users of music streaming services, the authors sought to understand how gender was approached by the platforms. The results suggested a significant difference in the quantity of recommended female and male artists; in general, users had to wait for six tracks to be played before reaching a female artist. Below, we demonstrate the main differences we observed when looking specifically at the consumption of music in Brazil and from Brazilian artists. 

Gender differences

The recommendations of artists identified with the male gender are the majority, even when observed from different angles:

i. When looking at the overall total of recommendations: 

Gender percentage of artists played by the platform 

Male Female Mixed genders

On both platforms (Deezer and Spotify), the percentage of recommended male artists was above 50%. In the overall total of Spotify recommendations, male gender occupies 55.76%, female, 24.01% and mixed genders (groups or feat), 19.25%. In the overall total of Deezer recommendations, 58.39% were male, 19.76% female and 21.38% mixed.

ii. The total, considering the list of artists present on the Top 5 of both platforms, as well as those randomly chosen by the researchers, named here as Not Top 5, was:

Gender percentage of played artists

When we look only at the Top 5, the percentage of recommended male artists increases. On Spotify Top 5, 65.41% of the recommendations were of male artists, 15.70% of female and 17.47% of mixed gender artists. On Deezer, the percentage of recommendations of male artists rises to 73.23%, female drops to 12.90% and mixed, to 13.87%.

The recommendations of Not Top 5 keep the numbers closer to the general. In this category, on Spotify, the male gender occupies 49.74%, female 29.21% and mixed gender 20.37%. On Deezer, the male gender appears with 49.11% of the recommendations, female with 24.96% and mixed with 26.08%.  

iii) Individual playlists

Gender percentage of played artists (TOP5)

When looking at playlists by artists, we could observe that, on the Top 5, Marília Mendonça’s2 playlist on Spotify was the only one that did not have a majority of recommendations of male artists.

When we observed playlists of artists from the group Not Top 5, we noticed a majority of recommendations of female artists on the playlists of Aline Barros, Adriana Calcanhotto and Drik Barbosa on Spotify, and Drik Barbosa and Rico Dalasam on Deezer.

iv. Analysis through bots

Gender percentage of played artists by bot gender

The broad recommendation of male artists remains, independently of the gender of the user, or rather, of the gender assigned to the Bot.

On Spotify, the recommendations for the female Bot were: 55.65% male artists, 24.1% female and 19.27% mixed; for the male Bot, the recommendations were: 54.85% male artists, 23.88% female and 20.31% mixed; for the non-binary Bot, the recommendations were: 56.79% male, 24.1% female and 18.18% mixed.

On Deezer, recommendations of male artists were even higher. For the female Bot, they were: 58.15% male artists, 20.8% female and 21.33% mixed; for the male Bot, the recommendations were: 58.53% male artists, 19.42% female and 21.53% mixed; for the non-binary Bot, the recommendations were: 58.49% male, 19.79% female and 21.29% mixed.

v. Similarities between shifts

Gender percentage of played artists by shift

We observed that, regardless of the shift in which the radio is played, the high percentage of recommendation of male artists is maintained. On Spotify, in the morning shift, 55.75% of male artists, 24.37% of female and 18.97% of mixed gender were recommended. In the same platform, in the evening, 55.78% of male artists, 23.66% female and 19.53% mixed were recommended. On Deezer, in the morning shift, 57.96% of male artists, 20.61% female and 21.03% mixed were recommended. On the same platform, in the evening, 58.82% of male artists, 18.92% female and 21.74% mixed were recommended.

vi. Presence of female artists on feats 

Female solo participation X feats, groups, etc.

The presence of female solo artists on the recommendations occurred approximately the same number of times as on feats, bands, groups, etc. On some playlists, the participation of female artists in feats was higher than of solo female artists:

• Deezer: Barões da Pisadinha, Eli Soares, Jorge e Mateus, Mc Dricka, Mc Poze
• Spotify: Barões da Pisadinha, Eli Soares, Gilberto Gil, Gustavo Lima, Henrique e Juliano, Jorge e Mateus, Mc Dricka, Mc Poze

Male solo participation X feats, groups, etc.

The presence of male solo artists on the recommendations occurred almost three times as often as on feats, bands, groups, etc. This imbalance may indicate that men in solo careers seem to have more prominence or may be more easily recommended by algorithms than women in solo careers, which dialogues with other research in which it is argued that male musicians tend to be more successful in their careers than women3.

vii. Sequences of artists of the same gender

Top 5 and Not-Top 5: On both platforms, there were long sequences of songs by male artists. Spotify had the longest sequences continuously featuring male and female artists.

On Top 5, the longest sequence occurred on Spotify, on Jorge and Mateus’s playlist, with 31 songs by male artists consecutively played.

Top 10 longest sequences of artists of the same gender (Top 5)

On Not-Top 5, the longest sequence also occurred on Spotify, on Mc Poze’s playlist, with 45 songs by male artists consecutively played, that is, there was no occurrence of any woman for 45 songs.

Top 10 longest sequences of artists of the same gender (Not Top 5)

On Top5, the longest sequence of songs by female artists occurred on Spotify, on Marília Mendonça’s playlist. It was a sequence of eight female artists.

Top 10 longest sequences of artists of non-male gender (Top 5)

Top 10 longest sequences of artists of non-male gender (Not Top 5)

On Not-Top 5, the longest sequence of songs by female artists occurred on Spotify, on Adriana Calcanhotto’s playlist. There was a sequence of 17 recommendations of female artists. Therefore, the longest female sequence is less numerous than half the size of the longest male sequence.

viii. Probability of transition of genders

Within the data collected, the average probability of recommendation by artist gender on Top 5 was:

The average probability of transition on Top 5 was:

On data collected, the average probability of recommendation by artist gender on Not-Top 5:

The average probability of transition on Not-Top 5 was:

The idea here is not to stick to the values, but to show that, on both platforms, the probability of a platform recommending a male artist is higher both in isolation and in sequence. That is, the probability of a platform recommending a male artist is higher, and the probability of the next recommended artist being a male is higher. This interpretation reflects the long sequences of recommended songs by male artists.

An important consideration should be made here. As mentioned above, not all should be seen as a platforms’ flaw or responsibility. There is a marked gender inequality in the music industry, which varies among musical styles and which has not been created by streaming platforms and, thus, cannot not be overcome solely by adjusting the algorithm4. However, we see that this inequality is reinforced and perpetuated by the way algorithms have been making their recommendations on both the platforms studied here – which could, in turn, act in the opposite direction, precisely seeking to increase the visibility of women and non-binary artists. It is also worth mentioning the very low presence of non-binary artists in the highlighted data.

Differences among gender identities

Presence of non-binary and transvestite artists 

In the collection carried out, transvestite and non-binary artists were recommended only on the playlists of Adriana Calcanhotto, Drik Barbosa and Rico Dalasam. Transvestite artists were recommended only on Rico Dalasam’s playlist and on Spotify.