Runs. Can't be meaningfully edited.
Open source music — a working argument
A recording is a binary. Nobody ever shipped the source.
m3c is m + usi + c — music, contracted the
way i18n contracts internationalization. The joke is the point. It is a developer's
abbreviation, worn deliberately, because the whole initiative is an argument that music and software are
closer relatives than either field admits.
Not for lack of talent. Because at the one moment they might have started, there was nothing there to start with. No instrument within reach. Nobody around who could show them the first thing. The window opens, there's nothing in it, and it closes.
Almost every musician can name the accident that let them in — an instrument that happened to be in the house, a friend who knew three chords and was willing to share them. That accident does enormous work, and it is handed out almost entirely by luck.
The point is to manufacture that accident at scale and give it away.
The mission
The tools are the easy half of that problem — they fit in a browser and they cost nothing to run. The hard half is that the music itself is locked, and locked in a way the open source movement solved for code forty years ago and never solved for sound.
Music and code are both enormously complex creative works that compile down to something you can experience but can't easily take apart. You can run a binary without reading its source. You can hear a record without knowing what was played. In both cases the artifact can be imitated — closely, sometimes indistinguishably — and in neither case can it be perfectly recovered.
That's not a loose metaphor. It's the same problem with the same shape. And the Open Source Definition contains a clause with no music equivalent anywhere:
The source code must be the preferred form in which a programmer would modify the program. Deliberately obfuscated source code is not allowed. Intermediate forms such as the output of a preprocessor or translator are not allowed.
The Open Source Definition, clause 2
Legally open. Practically closed.
You may redistribute that MP3, remix it, build on it. You still cannot open it up, because the parts were never published.
Almost everything filed under "open source music" for the last twenty years has been open in the Creative Commons sense and closed in the OSD sense. That mismatch is a real reason the movement never compounded the way open source software did. A shared codebase gets better every time someone reads it. A shared MP3 just gets copied.
Sheet music is already handled. What has never been published is the thing a musician would actually reach for in order to change the work: the session, the stems, the takes.
Runs. Can't be meaningfully edited.
Plays. Can't be taken apart.
Every file, in the form its author edits.
Every stem, take and setting, as recorded.
How it got that way, and every earlier state.
Currently deleted. Nothing preserves them.
Open sourcing is meaningless until there is a source to open. So the order is deliberate: build the thing that keeps every take and every version — abstracted so hard it's invisible — and only then does "publish the source" mean anything at all.
There is a second gift buried in that order. An entire mature ecosystem for building things together — with history, parallel contribution, and permanent ownership — has existed for decades, is free, and is used almost exclusively by software developers. Musicians have precisely the same problem and none of the tooling.
A musician who's spent a month happily restoring old versions of a song already understands version control — they just don't know it has a name. Same first leap. Different instrument.
The mission
Say the goal is that someone learns to play from the records that started American popular music. Under US law, and under the rule this project holds itself to — expired copyright term only, never a fair-use argument — most of that music is not available yet.
Two independent layers have to clear: the recording and the composition. As of 2026, recordings first published through 1925 are in the public domain, and one more year unlocks every January 1st.
Free to transcribe today Term still running
This is not a disappointment. It is the schedule, and it is the argument for building now. Every one of those years arrives whether anyone is ready or not. The question is whether, on the morning the Bristol Sessions enter the public domain, there is an open format, an open corpus and free tooling waiting for them — or whether the first people to the material are the ones who can afford to license it.
There is also plenty already open. A century of recorded music is unlocked, and almost none of it has been transcribed into a form a learner can practice against. The Initiatives page lists where it lives and the rules for using it.