new normal logo

How data intelligence can drive diversity

— Julie Knibbe, Founder Music Tomorrow

How data intelligence can drive diversity — Julie Knibbe, Founder Music Tomorrow

Published:

1.2.2023

Photo:

Nathalie Mohadjer

Julie Knibbe has been working for 10 years with passion on data and music topics. She also worked for Deezer, where she was in charge of strategic management, and she's the founder of Music Tomorrow, to help the music industry on data intelligence and strategy issues, and to shed light on black boxes.

Can you quickly describe yourself, who you are, what you do?

I'm the founder of Music Tomorrow, a data intelligence and strategy consulting company for the music industry. I've been working on data and music issues for 10 years now - first at Deezer where I was in charge of the strategic direction of the streaming platform and the product direction of the music discovery experience, then at Soundcharts where I worked on the evolution of market intelligence solutions for the music industry.

What are your favorite subjects?

My favorite topics are at the intersection of artificial intelligence and music. Recommendation algorithms are now at the heart of how we approach the world. They allow artists to be discovered, to meet their public. However, the functioning of these algorithms is not always transparent or intelligible, and it is difficult for artists and their teams to grasp them. With my team, we want to open these "black boxes" of artificial intelligence.

What is the emerging topic that you think will matter in the near future?

Many issues are of course at the heart of our concerns, such as the climate emergency, food and energy security. We are concerned about this and it impacts us on a daily basis. However, to stay in my field of expertise, the cultural sector still has a role to play, especially artists through the influence they can have on their audience.

How did you became aware of it?

I became aware of imbalances simply by looking at the data. For example, female artists represent only 30% of the new music distributed on the platforms. When you look at the most listened to tracks, this statistic drops to around 20%. But women do not listen to less music than men, do not work less on their art, and are not less talented. It is therefore at the level of the intermediaries and the industry that things are at stake.

What are the stakes?

Bringing more transparency through data allows us to start addressing diversity issues, which are crucial, both from the point of view of the diversity of the music that is offered, the messages it carries, and the carriers of these messages, that is to say the artists.

What are the changes and opportunities that it can bring?

Diversity via algorithms can be promoted simply by introducing measures of diversity, either in the way their performance is measured, or in the selection of results, otherwise it's out of sight, out of mind. In R&D, we develop tools that allow artists to improve their SEO on streaming platforms. Think of it as SEO on Google, where the goal is to rank a site as high as possible, except that in our case, the goal is to be well positioned on Spotify or Deezer.

"Bringing more transparency through data allows us to start addressing diversity issue"

Who are the actors in this field?

Many industry players are well aware of these issues and are taking up these questions of transparency and diversity. To name a few, Deezer has collaborated with the CNRS to measure the possible convergence effects of algorithms (in an article named "Do algorithms push us to always listen to the same style of music?", in French), or Believe and Sony who have put their inclusion and diversity policies at the heart of their recruitment strategies, which can directly impact employee satisfaction.

Where can we find out more?

On our blog, we decrypt these algorithms and try to popularize them as much as possible. To learn more, here is an article we wrote that explains the solutions on the subject, and addresses the issue of measuring fairness, and filter bubbles.