Culture · 4 min read

Why old songs keep beating new ones

Catalogue music now takes the majority of listening. The reasons are mostly structural, not nostalgic.

Industry data has been telling the same story for several years: recordings more than eighteen months old account for a clear majority of listening, and the share keeps climbing. New releases are losing ground in their own market.

The catalogue got infinite

For most of recorded music history, access was the constraint. You could hear what the radio played, what the shop stocked, and what you owned. A record from thirty years earlier was genuinely difficult to obtain, so new music won by default — it was what was in front of you.

Streaming removed the constraint completely. A song from 1974 and a song from last Friday now sit at the same distance, one search apart, at the same price of nothing. New releases no longer compete against a handful of contemporaries. They compete against everything ever recorded, and most of it has already survived a filter.

Survivorship is a powerful filter

The old music available today is not a fair sample of old music. It is the fraction that people kept choosing for decades, while the forgettable majority disappeared. A new release is compared not against the average of 1975 but against its greatest hits, curated by fifty years of collective attention.

No new song gets that advantage. It arrives unsorted, alongside the roughly one hundred thousand other tracks uploaded that day, and has to earn its place against material that already won.

What algorithms do to discovery

Recommendation systems make this worse in a specific way. They are optimised for the probability that you keep listening, and the safest prediction is something close to what you already play. Familiarity is the strongest available signal, so the system routes attention toward the known.

The result is a loop. Old songs get played, which raises their signal, which gets them recommended more. Editorial playlists and viral moments cut against this occasionally, but the default gradient points backward.

The money followed

This is why catalogue rights started selling for enormous sums. A back catalogue with fifty years of steady streaming is a predictable annuity, which is exactly what a financial buyer wants. An unreleased artist is a venture bet.

New music is not worse and audiences have not stopped caring. The shelf simply stopped clearing. Every year adds to the catalogue and nothing leaves it, so the pile a new release has to climb gets a little taller each time.