Most music systems assume you choose what to listen to by name.
Artist names. Album titles. Genre labels that stopped meaning anything once the catalog passed a few thousand entries. When people say their library feels “unusable,” that’s usually the model underneath the complaint.
The false assumption is that recall drives listening – that you wake up wanting a specific thing, and the system’s job is to help you find it.
In practice, that’s rarely true.
Most listening starts with a constraint, not a title.
Energy level. Mood. Context. How much attention you have. Whether you’re driving, working, or trying to reset your nervous system.
Artist-first and genre-first systems collapse those constraints into trivia. Genres get noisy. Artists sprawl across moods. The catalog grows, but the entry points don’t.
The cost shows up slowly.
You rebuild playlists you already built.
You skip tracks that are “good” but wrong for the moment.
You default to the same safe clusters because searching feels like work.
Eventually the library exists, but you’re not really using it.
The leverage point isn’t better recommendations. It’s changing the axis.
When playlists are generated from attributes – mood, energy, context, source – you stop asking “what do I want to hear?” and start answering “what state am I in?” The system does the joining for you.
That shift also tolerates messiness. Tags can be incomplete. Moods can be fuzzy. Over time, play history does more work than perfect metadata ever could.
Half of why I’m building my own player is mundane: I never migrated my old iTunes library, and I wanted sane access to my own music again.
The other half is refusing a model that only works if you already know the answer.
When systems are designed around names, they reward memory.
When they’re designed around attributes, they support intent.
Most people don’t need more music.
They need a way to listen that matches how listening actually starts.
~~~
App 2 of 52
Music & creativity · Writing