Algorithms: Next time the Spotify recommendations surprise you or you check your Spotify Wrapped in December, think about how those songs got there and why Spotify thinks you might like that song, writes researcher Sebastian Cole.(Photo: Shutterstock / NTB)
Do you know why you choose the music you listen to?
SHARE YOUR SCIENCE: Algorithmic music recommendations are everywhere. Why do people resist to acknowledge them?
SebastianColeDoctoral Research Fellow at Department of Media and Communication, University of Oslo
People often say that they don’t use algorithms when streaming music. This is often true, but in many cases people don’t notice them.
Algorithmic systems are everywhere, and have been used to try to cure cancer, change grading in schools, and in everyday media. When buying books and clothes online, watching a movie on a streaming service, or reading a post on social media, it’s likely that algorithms are involved.
Hiding in plain sight
Platforms like YouTube, or Netflix, use algorithmic systems to profile users and offer personalised recommendations. They do this by collecting your data, analysing it, comparing to other users, and matching your user behaviours with music information and audio features. In this way, they can learn what a person likes and what the best recommendation should be.
Algorithmic recommendations are also there when listening to music. Spotify learns the music you like and recommends similar music organised in personalised playlists like Discover Weekly and Daily Mix.
However, some Spotify users I talked to in my research explain that they don’t use algorithmic recommendations, and when they do they feel the music recommendations they get are boring, repetitive, and nothing new. Still, they will later talk about how they sometimes use the personalised playlists the platform recommends to them when they are bored or don’t know what to play.
So, why do people struggle to accept or recognise they use algorithmic recommendations?
It depends on your goals
Spotify users I interviewed in my research point towards why it becomes so difficult to recognise algorithms.
Spotify, and other similar recommendation systems in other platforms, have the hard task of recommending content that is the same but different. Users like or dislike the recommendations based on what they need them for. When they want to explore, the recommendations will often look too similar to what they already know, but when they want familiar music, the recommendations might provide too much variation.
The recommendations can be good when people don’t know what to play, or bad if they have a specific craving, so in many cases it’s just easier to use saved playlists and avoid the randomness of recommendations.
Also, music streaming is not the only music source in people’s music lives. People may listen to music through streaming platforms, but they still find new music through family and friends, other platforms like TikTok, or the radio. In this way, platforms like Spotify are databases to save and listen to music, but music discovery often happens elsewhere.
Music is personal and we own our music taste, but music preferences are not the only reasons for not using the recommendations. In many cases, people just do not realise that, or are not interested if, there are algorithms involved.
While most people know there is something taking their data and recommending music, algorithmic systems are complicated and often work in the background. Most people do not know what is going on. As more advanced technology is developed, it becomes more difficult to distinguish if it is a person or a machine making the suggestions.
It’s better to know
Knowing that algorithmic systems are there and thinking about them can be beneficial to resist their influence and be critical about the content we consume. The impact algorithms might have feels less important when we think about music recommendations in comparison to privacy and bias concerns in politics and social media. But music recommendations are personal and important for users and might be the first step towards identifying them in larger systems.
So, next time the Spotify recommendations surprise you or you check your Spotify Wrapped in December, think about how those songs got there and why Spotify thinks you might like that song.