The Self-Writing Revolution Happened Twice

The oldest division of labor in recorded music is the one between the person who writes the song and the person who sings it. Tin Pan Alley ran on that division; the committee piece counted the writers' side of the house; this case counts the marriage — the share of recordings, decade by decade, on which at least one performer is also a credited writer of the underlying work. A million and a half sampled recordings, one curve, and the curve turns out to contain two revolutions and a counter-revolution:

The old regime, 1930–1955. For a quarter of a century the line barely moves: 16 to 20 percent. The industry was built factory-style — writers in one room, 1.7 to a song, performers in another — and the fraction of recordings where the two jobs shared a body stayed a steady minority. This is the world where "singer-songwriter" would have sounded like "plumber-notary": a legal possibility, not a career.

The first revolution, 1960–1975. The line climbs nine decades' worth in fifteen years: 22.8 to 34.9 percent. The folk revival, the Beatles' precedent (a band that wrote its own hits, an arrangement contracts had treated as a novelty), the album era's auteur economics — by the mid-seventies more than a third of recordings carried a performer-writer, the highest in recorded history to that point. Every rock memoir describes this revolution. The ledger confirms it, dated and sized.

The counter-revolution, 1983–1995. Here is the chapter the memoirs skip: the marriage rate went back down — 33.8 to a trough of 24.4 — a twelve-year rollback of most of what the sixties won. The MTV decade rebuilt the two-room factory with better carpeting: the professional topline writer, the production team, the interpreter-star, the dance record whose singer was hired after the song was finished. The interregnum has no name in rock history because rock history was embarrassed by it; the credits were not.

The second revolution, 1995–now. Then the longest climb on the chart, thirty unbroken years: 24.4 to 44.9 percent, the all-time high, and still rising. The cause this time is not an ideology but a floor plan: the bedroom studio abolished the two rooms. When the one-song career is a laptop purchase away, writer and performer are the same person because there is only one person. The rapper writes; the producer-artist writes; the SoundCloud population writes by definition. The old-regime split survives at the industrial top of pop — the part journalism can see, the wrong population again — while the population as a whole has quietly become the most self-written in the history of the medium.

The honesty paragraph, per house rules: this curve is computed on recordings whose works carry writer credits at all — the documented fraction, which is biased toward well-cataloged music, and MusicBrainz work-linking is itself richer for famous recordings. The absolute level could shift under fuller data; the shape — rise, rollback, second rise past every precedent — is carried by hundreds of thousands of samples per half-decade, and shapes survive documentation bias far better than levels do. The verdict stands as dated: the singer-songwriter was invented twice, and the second time it was not a movement. It was a default setting.

Method: one-in-seven sample of all recordings with a dated year (1930–2026), a linked work, and at least one writer credit (writing, composition, or lyrics) on that work: ~960k sampled recordings qualified. Self-written = intersection of the recording's performer/credit set with the work's writer set is non-empty. Half-decade shares; bins under 200 samples omitted. Computed August 2026 against the 2026-07-25 MusicBrainz build.