Your music app knows that you enjoy a certain kind of song. This is helpful until it decides that you would like to enjoy that kind of song for the remainder of your earthly existence.
Spotify’s recommendations can introduce unfamiliar artists and make a vast catalogue manageable. They can also feel repetitive. The interesting question is how to use the convenience without handing every listening decision to the same set of predictions.
A taste profile is an interpretation, not your soul
Spotify describes its taste profile as an interpretation of what and how you listen. That is a useful distinction. A system sees listening behaviour; it does not automatically know whether an album became a favourite or merely kept playing while you looked for your keys.
Recommendation systems can use patterns in listening and relationships between tracks. But reducing Spotify’s changing service to three fixed algorithms is too confident. Its complete current machinery is not established by an old diagram circulating online.
Does discovery become a bubble?
A 2020 Spotify research paper examined listening diversity using earlier platform data. Most users in its comparison had more diverse self-directed listening than algorithmically programmed listening. The researchers explicitly cautioned that this did not establish the recommendations caused narrower taste: people with narrower preferences might choose those features more often.
The work also linked diversity with favourable retention and conversion measures. So the simple claim that Spotify must always prefer to narrow your taste does not follow from this research. Neither does it prove that every recommendation experience is adventurous.
The practical tension is familiar: a satisfying suggestion should feel relevant, while a surprising one must move beyond the obvious. Too familiar becomes wallpaper; too surprising becomes “why is my breakfast accompanied by experimental bagpipes?”
Tell the system when a listen was not a vote
Spotify provides a useful control for this. Its taste-profile help page explains how to exclude a playlist or track through its menu. Past and future listens then have less influence on recommendations and taste summaries. The change is reversible and may take time to affect weekly recommendations.
That can help when sleep music or a situational playlist dominates your listening time. Excluding a playlist is not a universal block on its songs: listening to them outside that playlist can still count. It is a steering control, not a magical personality reset.
Leave room for a human detour
Ask a friend for an album rather than another individual track. Follow a musician’s collaborators. Try a radio programme or a record-shop recommendation. These routes can introduce a different person’s organizing principle into your listening.
You do not have to abandon streaming to do this. Discovery and deliberate choice can coexist on the same device. The point is to occasionally decide where the journey begins instead of accepting every next turn.
If you want a particularly unusual detour, our tour of six unconventional instruments includes sounds that make an ordinary guitar look almost suspiciously reasonable. Or consider the slower ritual behind vinyl’s revival.
A recommendation engine can be a good guide. It does not need to be the only person choosing the soundtrack.
🐾 Curiosity looks good on you. Explore the Goodies, or find our illustrated books on Amazon.
